gnu/packages/bioinformatics.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2014, 2015, 2016, 2017, 2018 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2015, 2016, 2017, 2018 Ben Woodcroft <donttrustben@gmail.com>
   4 ;;; Copyright © 2015, 2016 Pjotr Prins <pjotr.guix@thebird.nl>
   5 ;;; Copyright © 2015 Andreas Enge <andreas@enge.fr>
   6 ;;; Copyright © 2016 Roel Janssen <roel@gnu.org>
   7 ;;; Copyright © 2016, 2017, 2018 Efraim Flashner <efraim@flashner.co.il>
   8 ;;; Copyright © 2016 Marius Bakke <mbakke@fastmail.com>
   9 ;;; Copyright © 2016, 2018 Raoul Bonnal <ilpuccio.febo@gmail.com>
  10 ;;; Copyright © 2017, 2018 Tobias Geerinckx-Rice <me@tobias.gr>
  11 ;;; Copyright © 2017 Arun Isaac <arunisaac@systemreboot.net>
  12 ;;;
  13 ;;; This file is part of GNU Guix.
  14 ;;;
  15 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  16 ;;; under the terms of the GNU General Public License as published by
  17 ;;; the Free Software Foundation; either version 3 of the License, or (at
  18 ;;; your option) any later version.
  19 ;;;
  20 ;;; GNU Guix is distributed in the hope that it will be useful, but
  21 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  22 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  23 ;;; GNU General Public License for more details.
  24 ;;;
  25 ;;; You should have received a copy of the GNU General Public License
  26 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  27
  28 (define-module (gnu packages bioinformatics)
  29   #:use-module ((guix licenses) #:prefix license:)
  30   #:use-module (guix packages)
  31   #:use-module (guix utils)
  32   #:use-module (guix download)
  33   #:use-module (guix git-download)
  34   #:use-module (guix hg-download)
  35   #:use-module (guix build-system ant)
  36   #:use-module (guix build-system gnu)
  37   #:use-module (guix build-system cmake)
  38   #:use-module (guix build-system ocaml)
  39   #:use-module (guix build-system perl)
  40   #:use-module (guix build-system python)
  41   #:use-module (guix build-system r)
  42   #:use-module (guix build-system ruby)
  43   #:use-module (guix build-system scons)
  44   #:use-module (guix build-system trivial)
  45   #:use-module (gnu packages)
  46   #:use-module (gnu packages autotools)
  47   #:use-module (gnu packages algebra)
  48   #:use-module (gnu packages base)
  49   #:use-module (gnu packages bash)
  50   #:use-module (gnu packages bison)
  51   #:use-module (gnu packages boost)
  52   #:use-module (gnu packages check)
  53   #:use-module (gnu packages compression)
  54   #:use-module (gnu packages cpio)
  55   #:use-module (gnu packages cran)
  56   #:use-module (gnu packages curl)
  57   #:use-module (gnu packages documentation)
  58   #:use-module (gnu packages databases)
  59   #:use-module (gnu packages datastructures)
  60   #:use-module (gnu packages file)
  61   #:use-module (gnu packages flex)
  62   #:use-module (gnu packages gawk)
  63   #:use-module (gnu packages gcc)
  64   #:use-module (gnu packages gd)
  65   #:use-module (gnu packages gtk)
  66   #:use-module (gnu packages glib)
  67   #:use-module (gnu packages graph)
  68   #:use-module (gnu packages groff)
  69   #:use-module (gnu packages guile)
  70   #:use-module (gnu packages haskell)
  71   #:use-module (gnu packages image)
  72   #:use-module (gnu packages imagemagick)
  73   #:use-module (gnu packages java)
  74   #:use-module (gnu packages jemalloc)
  75   #:use-module (gnu packages ldc)
  76   #:use-module (gnu packages linux)
  77   #:use-module (gnu packages logging)
  78   #:use-module (gnu packages machine-learning)
  79   #:use-module (gnu packages man)
  80   #:use-module (gnu packages maths)
  81   #:use-module (gnu packages mpi)
  82   #:use-module (gnu packages ncurses)
  83   #:use-module (gnu packages ocaml)
  84   #:use-module (gnu packages pcre)
  85   #:use-module (gnu packages parallel)
  86   #:use-module (gnu packages pdf)
  87   #:use-module (gnu packages perl)
  88   #:use-module (gnu packages perl-check)
  89   #:use-module (gnu packages pkg-config)
  90   #:use-module (gnu packages popt)
  91   #:use-module (gnu packages protobuf)
  92   #:use-module (gnu packages python)
  93   #:use-module (gnu packages python-web)
  94   #:use-module (gnu packages readline)
  95   #:use-module (gnu packages ruby)
  96   #:use-module (gnu packages serialization)
  97   #:use-module (gnu packages shells)
  98   #:use-module (gnu packages statistics)
  99   #:use-module (gnu packages swig)
 100   #:use-module (gnu packages tbb)
 101   #:use-module (gnu packages tex)
 102   #:use-module (gnu packages texinfo)
 103   #:use-module (gnu packages textutils)
 104   #:use-module (gnu packages time)
 105   #:use-module (gnu packages tls)
 106   #:use-module (gnu packages vim)
 107   #:use-module (gnu packages web)
 108   #:use-module (gnu packages xml)
 109   #:use-module (gnu packages xorg)
 110   #:use-module (srfi srfi-1)
 111   #:use-module (ice-9 match))
 112
 113 (define-public r-ape
 114   (package
 115     (name "r-ape")
 116     (version "5.0")
 117     (source
 118      (origin
 119        (method url-fetch)
 120        (uri (cran-uri "ape" version))
 121        (sha256
 122         (base32
 123          "0q59pmxawz498cb9mv5m49lhiwxib8ak94yyydz7qg8b6lpd4bn3"))))
 124     (build-system r-build-system)
 125     (propagated-inputs
 126      `(("r-lattice" ,r-lattice)
 127        ("r-nlme" ,r-nlme)
 128        ("r-rcpp" ,r-rcpp)))
 129     (home-page "http://ape-package.ird.fr/")
 130     (synopsis "Analyses of phylogenetics and evolution")
 131     (description
 132      "This package provides functions for reading, writing, plotting, and
 133 manipulating phylogenetic trees, analyses of comparative data in a
 134 phylogenetic framework, ancestral character analyses, analyses of
 135 diversification and macroevolution, computing distances from DNA sequences,
 136 and several other tools.")
 137     (license license:gpl2+)))
 138
 139 (define-public aragorn
 140   (package
 141     (name "aragorn")
 142     (version "1.2.38")
 143     (source (origin
 144               (method url-fetch)
 145               (uri (string-append
 146                     "http://mbio-serv2.mbioekol.lu.se/ARAGORN/Downloads/aragorn"
 147                     version ".tgz"))
 148               (sha256
 149                (base32
 150                 "09i1rg716smlbnixfm7q1ml2mfpaa2fpn3hwjg625ysmfwwy712b"))))
 151     (build-system gnu-build-system)
 152     (arguments
 153      `(#:tests? #f ; there are no tests
 154        #:phases
 155        (modify-phases %standard-phases
 156          (delete 'configure)
 157          (replace 'build
 158                   (lambda _
 159                     (zero? (system* "gcc"
 160                                     "-O3"
 161                                     "-ffast-math"
 162                                     "-finline-functions"
 163                                     "-o"
 164                                     "aragorn"
 165                                     (string-append "aragorn" ,version ".c")))))
 166          (replace 'install
 167                   (lambda* (#:key outputs #:allow-other-keys)
 168                     (let* ((out (assoc-ref outputs "out"))
 169                            (bin (string-append out "/bin"))
 170                            (man (string-append out "/share/man/man1")))
 171                       (mkdir-p bin)
 172                       (install-file "aragorn" bin)
 173                       (mkdir-p man)
 174                       (install-file "aragorn.1" man))
 175                     #t)))))
 176     (home-page "http://mbio-serv2.mbioekol.lu.se/ARAGORN")
 177     (synopsis "Detect tRNA, mtRNA and tmRNA genes in nucleotide sequences")
 178     (description
 179      "Aragorn identifies transfer RNA, mitochondrial RNA and
 180 transfer-messenger RNA from nucleotide sequences, based on homology to known
 181 tRNA consensus sequences and RNA structure.  It also outputs the secondary
 182 structure of the predicted RNA.")
 183     (license license:gpl2)))
 184
 185 (define-public bamm
 186   (package
 187     (name "bamm")
 188     (version "1.7.3")
 189     (source (origin
 190               (method url-fetch)
 191               ;; BamM is not available on pypi.
 192               (uri (string-append
 193                     "https://github.com/Ecogenomics/BamM/archive/"
 194                     version ".tar.gz"))
 195               (file-name (string-append name "-" version ".tar.gz"))
 196               (sha256
 197                (base32
 198                 "1f35yxp4pc8aadsvbpg6r4kg2jh4fkjci0iby4iyljm6980sac0s"))
 199               (modules '((guix build utils)))
 200               (snippet
 201                `(begin
 202                   ;; Delete bundled htslib.
 203                   (delete-file-recursively "c/htslib-1.3.1")
 204                   #t))))
 205     (build-system python-build-system)
 206     (arguments
 207      `(#:python ,python-2 ; BamM is Python 2 only.
 208        ;; Do not use bundled libhts.  Do use the bundled libcfu because it has
 209        ;; been modified from its original form.
 210        #:configure-flags
 211        (let ((htslib (assoc-ref %build-inputs "htslib")))
 212          (list "--with-libhts-lib" (string-append htslib "/lib")
 213                "--with-libhts-inc" (string-append htslib "/include/htslib")))
 214        #:phases
 215        (modify-phases %standard-phases
 216          (add-after 'unpack 'autogen
 217            (lambda _
 218              (with-directory-excursion "c"
 219                (let ((sh (which "sh")))
 220                  ;; Use autogen so that 'configure' works.
 221                  (substitute* "autogen.sh" (("/bin/sh") sh))
 222                  (setenv "CONFIG_SHELL" sh)
 223                  (substitute* "configure" (("/bin/sh") sh))
 224                  (zero? (system* "./autogen.sh"))))))
 225          (delete 'build)
 226          ;; Run tests after installation so compilation only happens once.
 227          (delete 'check)
 228          (add-after 'install 'wrap-executable
 229            (lambda* (#:key outputs #:allow-other-keys)
 230             (let* ((out  (assoc-ref outputs "out"))
 231                    (path (getenv "PATH")))
 232               (wrap-program (string-append out "/bin/bamm")
 233                 `("PATH" ":" prefix (,path))))
 234             #t))
 235          (add-after 'wrap-executable 'post-install-check
 236            (lambda* (#:key inputs outputs #:allow-other-keys)
 237              (setenv "PATH"
 238                      (string-append (assoc-ref outputs "out")
 239                                     "/bin:"
 240                                     (getenv "PATH")))
 241              (setenv "PYTHONPATH"
 242                      (string-append
 243                       (assoc-ref outputs "out")
 244                       "/lib/python"
 245                       (string-take (string-take-right
 246                                     (assoc-ref inputs "python") 5) 3)
 247                       "/site-packages:"
 248                       (getenv "PYTHONPATH")))
 249              ;; There are 2 errors printed, but they are safe to ignore:
 250              ;; 1) [E::hts_open_format] fail to open file ...
 251              ;; 2) samtools view: failed to open ...
 252              (zero? (system* "nosetests")))))))
 253     (native-inputs
 254      `(("autoconf" ,autoconf)
 255        ("automake" ,automake)
 256        ("libtool" ,libtool)
 257        ("zlib" ,zlib)
 258        ("python-nose" ,python2-nose)
 259        ("python-pysam" ,python2-pysam)))
 260     (inputs
 261      `(("htslib" ,htslib-1.3) ; At least one test fails on htslib-1.4+.
 262        ("samtools" ,samtools)
 263        ("bwa" ,bwa)
 264        ("grep" ,grep)
 265        ("sed" ,sed)
 266        ("coreutils" ,coreutils)))
 267     (propagated-inputs
 268      `(("python-numpy" ,python2-numpy)))
 269     (home-page "http://ecogenomics.github.io/BamM/")
 270     (synopsis "Metagenomics-focused BAM file manipulator")
 271     (description
 272      "BamM is a C library, wrapped in python, to efficiently generate and
 273 parse BAM files, specifically for the analysis of metagenomic data.  For
 274 instance, it implements several methods to assess contig-wise read coverage.")
 275     (license license:lgpl3+)))
 276
 277 (define-public bamtools
 278   (package
 279     (name "bamtools")
 280     (version "2.4.1")
 281     (source (origin
 282               (method url-fetch)
 283               (uri (string-append
 284                     "https://github.com/pezmaster31/bamtools/archive/v"
 285                     version ".tar.gz"))
 286               (file-name (string-append name "-" version ".tar.gz"))
 287               (sha256
 288                (base32
 289                 "0jr024kcrhjb82cm69i7p5fcg5375zlc1h3qh2n1v368hcd0qflk"))))
 290     (build-system cmake-build-system)
 291     (arguments
 292      `(#:tests? #f ;no "check" target
 293        #:phases
 294        (modify-phases %standard-phases
 295          (add-before
 296           'configure 'set-ldflags
 297           (lambda* (#:key outputs #:allow-other-keys)
 298             (setenv "LDFLAGS"
 299                     (string-append
 300                      "-Wl,-rpath="
 301                      (assoc-ref outputs "out") "/lib/bamtools")))))))
 302     (inputs `(("zlib" ,zlib)))
 303     (home-page "https://github.com/pezmaster31/bamtools")
 304     (synopsis "C++ API and command-line toolkit for working with BAM data")
 305     (description
 306      "BamTools provides both a C++ API and a command-line toolkit for handling
 307 BAM files.")
 308     (license license:expat)))
 309
 310 (define-public bcftools
 311   (package
 312     (name "bcftools")
 313     (version "1.5")
 314     (source (origin
 315               (method url-fetch)
 316               (uri (string-append
 317                     "https://github.com/samtools/bcftools/releases/download/"
 318                     version "/bcftools-" version ".tar.bz2"))
 319               (sha256
 320                (base32
 321                 "0093hkkvxmbwfaa7905s6185jymynvg42kq6sxv7fili11l5mxwz"))
 322               (patches (search-patches "bcftools-regidx-unsigned-char.patch"))
 323               (modules '((guix build utils)))
 324               (snippet
 325                ;; Delete bundled htslib.
 326                '(delete-file-recursively "htslib-1.5"))))
 327     (build-system gnu-build-system)
 328     (arguments
 329      `(#:test-target "test"
 330        #:configure-flags (list "--with-htslib=system")
 331        #:make-flags
 332        (list
 333         "USE_GPL=1"
 334         "LIBS=-lgsl -lgslcblas"
 335         (string-append "prefix=" (assoc-ref %outputs "out"))
 336         (string-append "HTSDIR=" (assoc-ref %build-inputs "htslib") "/include")
 337         (string-append "HTSLIB=" (assoc-ref %build-inputs "htslib") "/lib/libhts.so")
 338         (string-append "BGZIP=" (assoc-ref %build-inputs "htslib") "/bin/bgzip")
 339         (string-append "TABIX=" (assoc-ref %build-inputs "htslib") "/bin/tabix")
 340         (string-append "PACKAGE_VERSION=" ,version))
 341        #:phases
 342        (modify-phases %standard-phases
 343          (add-before 'check 'patch-tests
 344            (lambda _
 345              (substitute* "test/test.pl"
 346                (("/bin/bash") (which "bash")))
 347              #t)))))
 348     (native-inputs
 349      `(("htslib" ,htslib)
 350        ("perl" ,perl)))
 351     (inputs
 352      `(("gsl" ,gsl)
 353        ("zlib" ,zlib)))
 354     (home-page "https://samtools.github.io/bcftools/")
 355     (synopsis "Utilities for variant calling and manipulating VCFs and BCFs")
 356     (description
 357      "BCFtools is a set of utilities that manipulate variant calls in the
 358 Variant Call Format (VCF) and its binary counterpart BCF.  All commands work
 359 transparently with both VCFs and BCFs, both uncompressed and BGZF-compressed.")
 360     ;; The sources are dual MIT/GPL, but becomes GPL-only when USE_GPL=1.
 361     (license (list license:gpl3+ license:expat))))
 362
 363 (define-public bedops
 364   (package
 365     (name "bedops")
 366     (version "2.4.14")
 367     (source (origin
 368               (method url-fetch)
 369               (uri (string-append "https://github.com/bedops/bedops/archive/v"
 370                                   version ".tar.gz"))
 371               (file-name (string-append name "-" version ".tar.gz"))
 372               (sha256
 373                (base32
 374                 "1kqbac547wyqma81cyky9n7mkgikjpsfd3nnmcm6hpqwanqgh10v"))))
 375     (build-system gnu-build-system)
 376     (arguments
 377      '(#:tests? #f
 378        #:make-flags (list (string-append "BINDIR=" %output "/bin"))
 379        #:phases
 380        (modify-phases %standard-phases
 381          (add-after 'unpack 'unpack-tarballs
 382            (lambda _
 383              ;; FIXME: Bedops includes tarballs of minimally patched upstream
 384              ;; libraries jansson, zlib, and bzip2.  We cannot just use stock
 385              ;; libraries because at least one of the libraries (zlib) is
 386              ;; patched to add a C++ function definition (deflateInit2cpp).
 387              ;; Until the Bedops developers offer a way to link against system
 388              ;; libraries we have to build the in-tree copies of these three
 389              ;; libraries.
 390
 391              ;; See upstream discussion:
 392              ;; https://github.com/bedops/bedops/issues/124
 393
 394              ;; Unpack the tarballs to benefit from shebang patching.
 395              (with-directory-excursion "third-party"
 396                (and (zero? (system* "tar" "xvf" "jansson-2.6.tar.bz2"))
 397                     (zero? (system* "tar" "xvf" "zlib-1.2.7.tar.bz2"))
 398                     (zero? (system* "tar" "xvf" "bzip2-1.0.6.tar.bz2"))))
 399              ;; Disable unpacking of tarballs in Makefile.
 400              (substitute* "system.mk/Makefile.linux"
 401                (("^\tbzcat .*") "\t@echo \"not unpacking\"\n")
 402                (("\\./configure") "CONFIG_SHELL=bash ./configure"))
 403              (substitute* "third-party/zlib-1.2.7/Makefile.in"
 404                (("^SHELL=.*$") "SHELL=bash\n"))
 405              #t))
 406          (delete 'configure))))
 407     (home-page "https://github.com/bedops/bedops")
 408     (synopsis "Tools for high-performance genomic feature operations")
 409     (description
 410      "BEDOPS is a suite of tools to address common questions raised in genomic
 411 studies---mostly with regard to overlap and proximity relationships between
 412 data sets.  It aims to be scalable and flexible, facilitating the efficient
 413 and accurate analysis and management of large-scale genomic data.
 414
 415 BEDOPS provides tools that perform highly efficient and scalable Boolean and
 416 other set operations, statistical calculations, archiving, conversion and
 417 other management of genomic data of arbitrary scale.  Tasks can be easily
 418 split by chromosome for distributing whole-genome analyses across a
 419 computational cluster.")
 420     (license license:gpl2+)))
 421
 422 (define-public bedtools
 423   (package
 424     (name "bedtools")
 425     (version "2.27.1")
 426     (source (origin
 427               (method url-fetch)
 428               (uri (string-append "https://github.com/arq5x/bedtools2/releases/"
 429                                   "download/v" version "/"
 430                                   "bedtools-" version ".tar.gz"))
 431               (sha256
 432                (base32
 433                 "1ndg5yknrxl4djx8ddzgk12rrbiidfpmkkg5z3f95jzryfxarhn8"))))
 434     (build-system gnu-build-system)
 435     (arguments
 436      '(#:test-target "test"
 437        #:make-flags
 438        (list (string-append "prefix=" (assoc-ref %outputs "out")))
 439        #:phases
 440        (modify-phases %standard-phases
 441          (delete 'configure))))
 442     (native-inputs `(("python" ,python-2)))
 443     (inputs
 444      `(("samtools" ,samtools)
 445        ("zlib" ,zlib)))
 446     (home-page "https://github.com/arq5x/bedtools2")
 447     (synopsis "Tools for genome analysis and arithmetic")
 448     (description
 449      "Collectively, the bedtools utilities are a swiss-army knife of tools for
 450 a wide-range of genomics analysis tasks.  The most widely-used tools enable
 451 genome arithmetic: that is, set theory on the genome.  For example, bedtools
 452 allows one to intersect, merge, count, complement, and shuffle genomic
 453 intervals from multiple files in widely-used genomic file formats such as BAM,
 454 BED, GFF/GTF, VCF.")
 455     (license license:gpl2)))
 456
 457 ;; Later releases of bedtools produce files with more columns than
 458 ;; what Ribotaper expects.
 459 (define-public bedtools-2.18
 460   (package (inherit bedtools)
 461     (name "bedtools")
 462     (version "2.18.0")
 463     (source (origin
 464               (method url-fetch)
 465               (uri (string-append "https://github.com/arq5x/bedtools2/"
 466                                   "archive/v" version ".tar.gz"))
 467               (file-name (string-append name "-" version ".tar.gz"))
 468               (sha256
 469                (base32
 470                 "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))
 471     (arguments
 472      '(#:test-target "test"
 473        #:phases
 474        (modify-phases %standard-phases
 475          (delete 'configure)
 476          (replace 'install
 477            (lambda* (#:key outputs #:allow-other-keys)
 478              (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
 479                (for-each (lambda (file)
 480                            (install-file file bin))
 481                          (find-files "bin" ".*")))
 482              #t)))))))
 483
 484 (define-public ribotaper
 485   (package
 486     (name "ribotaper")
 487     (version "1.3.1")
 488     (source (origin
 489               (method url-fetch)
 490               (uri (string-append "https://ohlerlab.mdc-berlin.de/"
 491                                   "files/RiboTaper/RiboTaper_Version_"
 492                                   version ".tar.gz"))
 493               (sha256
 494                (base32
 495                 "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
 496     (build-system gnu-build-system)
 497     (arguments
 498      `(#:phases
 499        (modify-phases %standard-phases
 500          (add-after 'install 'wrap-executables
 501            (lambda* (#:key inputs outputs #:allow-other-keys)
 502              (let* ((out (assoc-ref outputs "out")))
 503                (for-each
 504                 (lambda (script)
 505                   (wrap-program (string-append out "/bin/" script)
 506                     `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
 507                 '("create_annotations_files.bash"
 508                   "create_metaplots.bash"
 509                   "Ribotaper_ORF_find.sh"
 510                   "Ribotaper.sh"))))))))
 511     (inputs
 512      `(("bedtools" ,bedtools-2.18)
 513        ("samtools" ,samtools-0.1)
 514        ("r-minimal" ,r-minimal)
 515        ("r-foreach" ,r-foreach)
 516        ("r-xnomial" ,r-xnomial)
 517        ("r-domc" ,r-domc)
 518        ("r-multitaper" ,r-multitaper)
 519        ("r-seqinr" ,r-seqinr)))
 520     (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
 521     (synopsis "Define translated ORFs using ribosome profiling data")
 522     (description
 523      "Ribotaper is a method for defining translated @dfn{open reading
 524 frames} (ORFs) using ribosome profiling (ribo-seq) data.  This package
 525 provides the Ribotaper pipeline.")
 526     (license license:gpl3+)))
 527
 528 (define-public ribodiff
 529   (package
 530     (name "ribodiff")
 531     (version "0.2.2")
 532     (source
 533      (origin
 534        (method url-fetch)
 535        (uri (string-append "https://github.com/ratschlab/RiboDiff/"
 536                            "archive/v" version ".tar.gz"))
 537        (file-name (string-append name "-" version ".tar.gz"))
 538        (sha256
 539         (base32
 540          "0wpbwmfv05wdjxv7ikm664f7s7p7cqr8jnw99zrda0q67rl50aaj"))))
 541     (build-system python-build-system)
 542     (arguments
 543      `(#:python ,python-2
 544        #:phases
 545        (modify-phases %standard-phases
 546          ;; Generate an installable executable script wrapper.
 547          (add-after 'unpack 'patch-setup.py
 548            (lambda _
 549              (substitute* "setup.py"
 550                (("^(.*)packages=.*" line prefix)
 551                 (string-append line "\n"
 552                                prefix "scripts=['scripts/TE.py'],\n")))
 553              #t)))))
 554     (inputs
 555      `(("python-numpy" ,python2-numpy)
 556        ("python-matplotlib" ,python2-matplotlib)
 557        ("python-scipy" ,python2-scipy)
 558        ("python-statsmodels" ,python2-statsmodels)))
 559     (native-inputs
 560      `(("python-mock" ,python2-mock)
 561        ("python-nose" ,python2-nose)))
 562     (home-page "http://public.bmi.inf.ethz.ch/user/zhongy/RiboDiff/")
 563     (synopsis "Detect translation efficiency changes from ribosome footprints")
 564     (description "RiboDiff is a statistical tool that detects the protein
 565 translational efficiency change from Ribo-Seq (ribosome footprinting) and
 566 RNA-Seq data.  It uses a generalized linear model to detect genes showing
 567 difference in translational profile taking mRNA abundance into account.  It
 568 facilitates us to decipher the translational regulation that behave
 569 independently with transcriptional regulation.")
 570     (license license:gpl3+)))
 571
 572 (define-public bioawk
 573   (package
 574     (name "bioawk")
 575     (version "1.0")
 576     (source (origin
 577       (method url-fetch)
 578       (uri (string-append "https://github.com/lh3/bioawk/archive/v"
 579                           version ".tar.gz"))
 580       (file-name (string-append name "-" version ".tar.gz"))
 581       (sha256
 582        (base32 "1daizxsk17ahi9n58fj8vpgwyhzrzh54bzqhanjanp88kgrz7gjw"))))
 583     (build-system gnu-build-system)
 584     (inputs
 585      `(("zlib" ,zlib)))
 586     (native-inputs
 587      `(("bison" ,bison)))
 588     (arguments
 589      `(#:tests? #f ; There are no tests to run.
 590        ;; Bison must generate files, before other targets can build.
 591        #:parallel-build? #f
 592        #:phases
 593        (modify-phases %standard-phases
 594          (delete 'configure) ; There is no configure phase.
 595          (replace 'install
 596           (lambda* (#:key outputs #:allow-other-keys)
 597             (let* ((out (assoc-ref outputs "out"))
 598                    (bin  (string-append out "/bin"))
 599                    (man (string-append out "/share/man/man1")))
 600               (mkdir-p man)
 601               (copy-file "awk.1" (string-append man "/bioawk.1"))
 602               (install-file "bioawk" bin)))))))
 603     (home-page "https://github.com/lh3/bioawk")
 604     (synopsis "AWK with bioinformatics extensions")
 605     (description "Bioawk is an extension to Brian Kernighan's awk, adding the
 606 support of several common biological data formats, including optionally gzip'ed
 607 BED, GFF, SAM, VCF, FASTA/Q and TAB-delimited formats with column names.  It
 608 also adds a few built-in functions and a command line option to use TAB as the
 609 input/output delimiter.  When the new functionality is not used, bioawk is
 610 intended to behave exactly the same as the original BWK awk.")
 611     (license license:x11)))
 612
 613 (define-public python2-pybedtools
 614   (package
 615     (name "python2-pybedtools")
 616     (version "0.6.9")
 617     (source (origin
 618               (method url-fetch)
 619               (uri (string-append
 620                     "https://pypi.python.org/packages/source/p/pybedtools/pybedtools-"
 621                     version ".tar.gz"))
 622               (sha256
 623                (base32
 624                 "1ldzdxw1p4y3g2ignmggsdypvqkcwqwzhdha4rbgpih048z5p4an"))))
 625     (build-system python-build-system)
 626     (arguments `(#:python ,python-2)) ; no Python 3 support
 627     (inputs
 628      `(("python-matplotlib" ,python2-matplotlib)))
 629     (propagated-inputs
 630      `(("bedtools" ,bedtools)
 631        ("samtools" ,samtools)))
 632     (native-inputs
 633      `(("python-cython" ,python2-cython)
 634        ("python-pyyaml" ,python2-pyyaml)
 635        ("python-nose" ,python2-nose)))
 636     (home-page "https://pythonhosted.org/pybedtools/")
 637     (synopsis "Python wrapper for BEDtools programs")
 638     (description
 639      "pybedtools is a Python wrapper for Aaron Quinlan's BEDtools programs,
 640 which are widely used for genomic interval manipulation or \"genome algebra\".
 641 pybedtools extends BEDTools by offering feature-level manipulations from with
 642 Python.")
 643     (license license:gpl2+)))
 644
 645 (define-public python-biom-format
 646   (package
 647    (name "python-biom-format")
 648    (version "2.1.6")
 649    (source
 650     (origin
 651      (method url-fetch)
 652      ;; Use GitHub as source because PyPI distribution does not contain
 653      ;; test data: https://github.com/biocore/biom-format/issues/693
 654      (uri (string-append "https://github.com/biocore/biom-format/archive/"
 655                          version ".tar.gz"))
 656      (file-name (string-append name "-" version ".tar.gz"))
 657      (sha256
 658       (base32
 659        "08cr7wpahk6zb31h4bs7jmzpvxcqv9s13xz40h6y2h656jvdvnpj"))))
 660    (build-system python-build-system)
 661    (propagated-inputs
 662     `(("python-numpy" ,python-numpy)
 663       ("python-scipy" ,python-scipy)
 664       ("python-future" ,python-future)
 665       ("python-click" ,python-click)
 666       ("python-h5py" ,python-h5py)
 667       ("python-pandas" ,python-pandas)))
 668    (native-inputs
 669     `(("python-nose" ,python-nose)))
 670    (home-page "http://www.biom-format.org")
 671    (synopsis "Biological Observation Matrix (BIOM) format utilities")
 672    (description
 673     "The BIOM file format is designed to be a general-use format for
 674 representing counts of observations e.g. operational taxonomic units, KEGG
 675 orthology groups or lipid types, in one or more biological samples
 676 e.g. microbiome samples, genomes, metagenomes.")
 677    (license license:bsd-3)
 678    (properties `((python2-variant . ,(delay python2-biom-format))))))
 679
 680 (define-public python2-biom-format
 681   (let ((base (package-with-python2 (strip-python2-variant python-biom-format))))
 682     (package
 683       (inherit base)
 684       (arguments
 685        `(#:phases
 686          (modify-phases %standard-phases
 687            ;; Do not require the unmaintained pyqi library.
 688            (add-after 'unpack 'remove-pyqi
 689              (lambda _
 690                (substitute* "setup.py"
 691                  (("install_requires.append\\(\"pyqi\"\\)") "pass"))
 692                #t)))
 693          ,@(package-arguments base))))))
 694
 695 (define-public bioperl-minimal
 696   (let* ((inputs `(("perl-module-build" ,perl-module-build)
 697                    ("perl-data-stag" ,perl-data-stag)
 698                    ("perl-libwww" ,perl-libwww)
 699                    ("perl-uri" ,perl-uri)))
 700          (transitive-inputs
 701           (map (compose package-name cadr)
 702                (delete-duplicates
 703                 (concatenate
 704                  (map (compose package-transitive-target-inputs cadr) inputs))))))
 705     (package
 706       (name "bioperl-minimal")
 707       (version "1.7.0")
 708       (source
 709        (origin
 710          (method url-fetch)
 711          (uri (string-append "https://github.com/bioperl/bioperl-live/"
 712                              "archive/release-"
 713                              (string-map (lambda (c)
 714                                            (if (char=? c #\.)
 715                                                #\- c)) version)
 716                              ".tar.gz"))
 717          (sha256
 718           (base32
 719            "12phgpxwgkqflkwfb9dcqg7a31dpjlfhar8wcgv0aj5ln4akfz06"))))
 720       (build-system perl-build-system)
 721       (arguments
 722        `(#:phases
 723          (modify-phases %standard-phases
 724            (add-after
 725             'install 'wrap-programs
 726             (lambda* (#:key outputs #:allow-other-keys)
 727               ;; Make sure all executables in "bin" find the required Perl
 728               ;; modules at runtime.  As the PERL5LIB variable contains also
 729               ;; the paths of native inputs, we pick the transitive target
 730               ;; inputs from %build-inputs.
 731               (let* ((out  (assoc-ref outputs "out"))
 732                      (bin  (string-append out "/bin/"))
 733                      (path (string-join
 734                             (cons (string-append out "/lib/perl5/site_perl")
 735                                   (map (lambda (name)
 736                                          (assoc-ref %build-inputs name))
 737                                        ',transitive-inputs))
 738                             ":")))
 739                 (for-each (lambda (file)
 740                             (wrap-program file
 741                               `("PERL5LIB" ":" prefix (,path))))
 742                           (find-files bin "\\.pl$"))
 743                 #t))))))
 744       (inputs inputs)
 745       (native-inputs
 746        `(("perl-test-most" ,perl-test-most)))
 747       (home-page "http://search.cpan.org/dist/BioPerl")
 748       (synopsis "Bioinformatics toolkit")
 749       (description
 750        "BioPerl is the product of a community effort to produce Perl code which
 751 is useful in biology.  Examples include Sequence objects, Alignment objects
 752 and database searching objects.  These objects not only do what they are
 753 advertised to do in the documentation, but they also interact - Alignment
 754 objects are made from the Sequence objects, Sequence objects have access to
 755 Annotation and SeqFeature objects and databases, Blast objects can be
 756 converted to Alignment objects, and so on.  This means that the objects
 757 provide a coordinated and extensible framework to do computational biology.")
 758       (license license:perl-license))))
 759
 760 (define-public python-biopython
 761   (package
 762     (name "python-biopython")
 763     (version "1.70")
 764     (source (origin
 765               (method url-fetch)
 766               ;; use PyPi rather than biopython.org to ease updating
 767               (uri (pypi-uri "biopython" version))
 768               (sha256
 769                (base32
 770                 "0nz4n9d2y2dg849gn1z0vjlkwcpzzkzy3fij7x94a6ixy2c54z2a"))))
 771     (build-system python-build-system)
 772     (arguments
 773      `(#:phases
 774        (modify-phases %standard-phases
 775          (add-before 'check 'set-home
 776            ;; Some tests require a home directory to be set.
 777            (lambda _ (setenv "HOME" "/tmp") #t)))))
 778     (propagated-inputs
 779      `(("python-numpy" ,python-numpy)))
 780     (home-page "http://biopython.org/")
 781     (synopsis "Tools for biological computation in Python")
 782     (description
 783      "Biopython is a set of tools for biological computation including parsers
 784 for bioinformatics files into Python data structures; interfaces to common
 785 bioinformatics programs; a standard sequence class and tools for performing
 786 common operations on them; code to perform data classification; code for
 787 dealing with alignments; code making it easy to split up parallelizable tasks
 788 into separate processes; and more.")
 789     (license (license:non-copyleft "http://www.biopython.org/DIST/LICENSE"))))
 790
 791 (define-public python2-biopython
 792   (package-with-python2 python-biopython))
 793
 794 (define-public bpp-core
 795   ;; The last release was in 2014 and the recommended way to install from source
 796   ;; is to clone the git repository, so we do this.
 797   ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
 798   (let ((commit "7d8bced0d1a87291ea8dd7046b7fb5ff9c35c582"))
 799     (package
 800       (name "bpp-core")
 801       (version (string-append "2.2.0-1." (string-take commit 7)))
 802       (source (origin
 803                 (method git-fetch)
 804                 (uri (git-reference
 805                       (url "http://biopp.univ-montp2.fr/git/bpp-core")
 806                       (commit commit)))
 807                 (file-name (string-append name "-" version "-checkout"))
 808                 (sha256
 809                  (base32
 810                   "10djsq5vlnkilv436gnmh4irpk49v29pa69r6xiryg32xmvn909j"))))
 811       (build-system cmake-build-system)
 812       (arguments
 813        `(#:parallel-build? #f))
 814       (inputs
 815        `(("gcc" ,gcc-5))) ; Compilation of bpp-phyl fails with GCC 4.9 so we
 816                           ; compile all of the bpp packages with GCC 5.
 817       (home-page "http://biopp.univ-montp2.fr")
 818       (synopsis "C++ libraries for Bioinformatics")
 819       (description
 820        "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
 821 analysis, phylogenetics, molecular evolution and population genetics.  It is
 822 Object Oriented and is designed to be both easy to use and computer efficient.
 823 Bio++ intends to help programmers to write computer expensive programs, by
 824 providing them a set of re-usable tools.")
 825       (license license:cecill-c))))
 826
 827 (define-public bpp-phyl
 828   ;; The last release was in 2014 and the recommended way to install from source
 829   ;; is to clone the git repository, so we do this.
 830   ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
 831   (let ((commit "0c07167b629f68b569bf274d1ad0c4af83276ae2"))
 832     (package
 833       (name "bpp-phyl")
 834       (version (string-append "2.2.0-1." (string-take commit 7)))
 835       (source (origin
 836                 (method git-fetch)
 837                 (uri (git-reference
 838                       (url "http://biopp.univ-montp2.fr/git/bpp-phyl")
 839                       (commit commit)))
 840                 (file-name (string-append name "-" version "-checkout"))
 841                 (sha256
 842                  (base32
 843                   "1ssjgchzwj3iai26kyly7gwkdv8sk59nqhkb1wpap3sf5m6kyllh"))))
 844       (build-system cmake-build-system)
 845       (arguments
 846        `(#:parallel-build? #f
 847          ;; If out-of-source, test data is not copied into the build directory
 848          ;; so the tests fail.
 849          #:out-of-source? #f))
 850       (inputs
 851        `(("bpp-core" ,bpp-core)
 852          ("bpp-seq" ,bpp-seq)
 853          ;; GCC 4.8 fails due to an 'internal compiler error', so we use a more
 854          ;; modern GCC.
 855          ("gcc" ,gcc-5)))
 856       (home-page "http://biopp.univ-montp2.fr")
 857       (synopsis "Bio++ phylogenetic Library")
 858       (description
 859        "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
 860 analysis, phylogenetics, molecular evolution and population genetics.  This
 861 library provides phylogenetics-related modules.")
 862       (license license:cecill-c))))
 863
 864 (define-public bpp-popgen
 865   ;; The last release was in 2014 and the recommended way to install from source
 866   ;; is to clone the git repository, so we do this.
 867   ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
 868   (let ((commit "e472bac9b1a148803895d747cd6d0c5904f85d9f"))
 869     (package
 870       (name "bpp-popgen")
 871       (version (string-append "2.2.0-1." (string-take commit 7)))
 872       (source (origin
 873                 (method git-fetch)
 874                 (uri (git-reference
 875                       (url "http://biopp.univ-montp2.fr/git/bpp-popgen")
 876                       (commit commit)))
 877                 (file-name (string-append name "-" version "-checkout"))
 878                 (sha256
 879                  (base32
 880                   "0yn82dzn1n5629nzja68xfrhi655709rjanyryb36vzkmymy6dw5"))))
 881       (build-system cmake-build-system)
 882       (arguments
 883        `(#:parallel-build? #f
 884          #:tests? #f)) ; There are no tests.
 885       (inputs
 886        `(("bpp-core" ,bpp-core)
 887          ("bpp-seq" ,bpp-seq)
 888          ("gcc" ,gcc-5)))
 889       (home-page "http://biopp.univ-montp2.fr")
 890       (synopsis "Bio++ population genetics library")
 891       (description
 892        "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
 893 analysis, phylogenetics, molecular evolution and population genetics.  This
 894 library provides population genetics-related modules.")
 895       (license license:cecill-c))))
 896
 897 (define-public bpp-seq
 898   ;; The last release was in 2014 and the recommended way to install from source
 899   ;; is to clone the git repository, so we do this.
 900   ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
 901   (let ((commit "6cfa07965ce152e5598a89df2fa80a75973bfa33"))
 902     (package
 903       (name "bpp-seq")
 904       (version (string-append "2.2.0-1." (string-take commit 7)))
 905       (source (origin
 906                 (method git-fetch)
 907                 (uri (git-reference
 908                       (url "http://biopp.univ-montp2.fr/git/bpp-seq")
 909                       (commit commit)))
 910                 (file-name (string-append name "-" version "-checkout"))
 911                 (sha256
 912                  (base32
 913                   "1nys5jq7jqvdg40d91wsmj3q2yzy4276cp7sp44n67p468f27zf2"))))
 914       (build-system cmake-build-system)
 915       (arguments
 916        `(#:parallel-build? #f
 917          ;; If out-of-source, test data is not copied into the build directory
 918          ;; so the tests fail.
 919          #:out-of-source? #f))
 920       (inputs
 921        `(("bpp-core" ,bpp-core)
 922          ("gcc" ,gcc-5))) ; Use GCC 5 as per 'bpp-core'.
 923       (home-page "http://biopp.univ-montp2.fr")
 924       (synopsis "Bio++ sequence library")
 925       (description
 926        "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
 927 analysis, phylogenetics, molecular evolution and population genetics.  This
 928 library provides sequence-related modules.")
 929       (license license:cecill-c))))
 930
 931 (define-public bppsuite
 932   ;; The last release was in 2014 and the recommended way to install from source
 933   ;; is to clone the git repository, so we do this.
 934   ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
 935   (let ((commit "c516147f57aa50961121cd505bed52cd7603698b"))
 936     (package
 937       (name "bppsuite")
 938       (version (string-append "2.2.0-1." (string-take commit 7)))
 939       (source (origin
 940                 (method git-fetch)
 941                 (uri (git-reference
 942                       (url "http://biopp.univ-montp2.fr/git/bppsuite")
 943                       (commit commit)))
 944                 (file-name (string-append name "-" version "-checkout"))
 945                 (sha256
 946                  (base32
 947                   "1y87pxvw0jxjizhq2dr9g2r91md45k1p9ih2sl1yy1y3p934l2kb"))))
 948       (build-system cmake-build-system)
 949       (arguments
 950        `(#:parallel-build? #f
 951          #:tests? #f)) ; There are no tests.
 952       (native-inputs
 953        `(("groff" ,groff)
 954          ("man-db" ,man-db)
 955          ("texinfo" ,texinfo)))
 956       (inputs
 957        `(("bpp-core" ,bpp-core)
 958          ("bpp-seq" ,bpp-seq)
 959          ("bpp-phyl" ,bpp-phyl)
 960          ("bpp-phyl" ,bpp-popgen)
 961          ("gcc" ,gcc-5)))
 962       (home-page "http://biopp.univ-montp2.fr")
 963       (synopsis "Bioinformatics tools written with the Bio++ libraries")
 964       (description
 965        "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
 966 analysis, phylogenetics, molecular evolution and population genetics.  This
 967 package provides command line tools using the Bio++ library.")
 968       (license license:cecill-c))))
 969
 970 (define-public blast+
 971   (package
 972     (name "blast+")
 973     (version "2.6.0")
 974     (source (origin
 975               (method url-fetch)
 976               (uri (string-append
 977                     "ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/"
 978                     version "/ncbi-blast-" version "+-src.tar.gz"))
 979               (sha256
 980                (base32
 981                 "15n937pw5aqmyfjb6l387d18grqbb96l63d5xj4l7yyh0zbf2405"))
 982               (patches (search-patches "blast+-fix-makefile.patch"))
 983               (modules '((guix build utils)))
 984               (snippet
 985                '(begin
 986                   ;; Remove bundled bzip2, zlib and pcre.
 987                   (delete-file-recursively "c++/src/util/compress/bzip2")
 988                   (delete-file-recursively "c++/src/util/compress/zlib")
 989                   (delete-file-recursively "c++/src/util/regexp")
 990                   (substitute* "c++/src/util/compress/Makefile.in"
 991                     (("bzip2 zlib api") "api"))
 992                   ;; Remove useless msbuild directory
 993                   (delete-file-recursively
 994                    "c++/src/build-system/project_tree_builder/msbuild")
 995                   #t))))
 996     (build-system gnu-build-system)
 997     (arguments
 998      `(;; There are two(!) tests for this massive library, and both fail with
 999        ;; "unparsable timing stats".
1000        ;; ERR [127] --  [serial/datatool] datatool.sh     (unparsable timing stats)
1001        ;; ERR [127] --  [serial/datatool] datatool_xml.sh     (unparsable timing stats)
1002        #:tests? #f
1003        #:out-of-source? #t
1004        #:parallel-build? #f ; not supported
1005        #:phases
1006        (modify-phases %standard-phases
1007          (add-before
1008           'configure 'set-HOME
1009           ;; $HOME needs to be set at some point during the configure phase
1010           (lambda _ (setenv "HOME" "/tmp") #t))
1011          (add-after
1012           'unpack 'enter-dir
1013           (lambda _ (chdir "c++") #t))
1014          (add-after
1015           'enter-dir 'fix-build-system
1016           (lambda _
1017             (define (which* cmd)
1018               (cond ((string=? cmd "date")
1019                      ;; make call to "date" deterministic
1020                      "date -d @0")
1021                     ((which cmd)
1022                      => identity)
1023                     (else
1024                      (format (current-error-port)
1025                              "WARNING: Unable to find absolute path for ~s~%"
1026                              cmd)
1027                      #f)))
1028
1029             ;; Rewrite hardcoded paths to various tools
1030             (substitute* (append '("src/build-system/configure.ac"
1031                                    "src/build-system/configure"
1032                                    "src/build-system/helpers/run_with_lock.c"
1033                                    "scripts/common/impl/if_diff.sh"
1034                                    "scripts/common/impl/run_with_lock.sh"
1035                                    "src/build-system/Makefile.configurables.real"
1036                                    "src/build-system/Makefile.in.top"
1037                                    "src/build-system/Makefile.meta.gmake=no"
1038                                    "src/build-system/Makefile.meta.in"
1039                                    "src/build-system/Makefile.meta_l"
1040                                    "src/build-system/Makefile.meta_p"
1041                                    "src/build-system/Makefile.meta_r"
1042                                    "src/build-system/Makefile.mk.in"
1043                                    "src/build-system/Makefile.requirements"
1044                                    "src/build-system/Makefile.rules_with_autodep.in")
1045                                  (find-files "scripts/common/check" "\\.sh$"))
1046               (("(/usr/bin/|/bin/)([a-z][-_.a-z]*)" all dir cmd)
1047                (or (which* cmd) all)))
1048
1049             (substitute* (find-files "src/build-system" "^config.*")
1050               (("LN_S=/bin/\\$LN_S") (string-append "LN_S=" (which "ln")))
1051               (("^PATH=.*") ""))
1052
1053             ;; rewrite "/var/tmp" in check script
1054             (substitute* "scripts/common/check/check_make_unix.sh"
1055               (("/var/tmp") "/tmp"))
1056
1057             ;; do not reset PATH
1058             (substitute* (find-files "scripts/common/impl/" "\\.sh$")
1059               (("^ *PATH=.*") "")
1060               (("action=/bin/") "action=")
1061               (("export PATH") ":"))
1062             #t))
1063          (replace
1064           'configure
1065           (lambda* (#:key inputs outputs #:allow-other-keys)
1066             (let ((out     (assoc-ref outputs "out"))
1067                   (lib     (string-append (assoc-ref outputs "lib") "/lib"))
1068                   (include (string-append (assoc-ref outputs "include")
1069                                           "/include/ncbi-tools++")))
1070               ;; The 'configure' script doesn't recognize things like
1071               ;; '--enable-fast-install'.
1072               (zero? (system* "./configure.orig"
1073                               (string-append "--with-build-root=" (getcwd) "/build")
1074                               (string-append "--prefix=" out)
1075                               (string-append "--libdir=" lib)
1076                               (string-append "--includedir=" include)
1077                               (string-append "--with-bz2="
1078                                              (assoc-ref inputs "bzip2"))
1079                               (string-append "--with-z="
1080                                              (assoc-ref inputs "zlib"))
1081                               (string-append "--with-pcre="
1082                                              (assoc-ref inputs "pcre"))
1083                               ;; Each library is built twice by default, once
1084                               ;; with "-static" in its name, and again
1085                               ;; without.
1086                               "--without-static"
1087                               "--with-dll"))))))))
1088     (outputs '("out"       ;  21 MB
1089                "lib"       ; 226 MB
1090                "include")) ;  33 MB
1091     (inputs
1092      `(("bzip2" ,bzip2)
1093        ("zlib" ,zlib)
1094        ("pcre" ,pcre)
1095        ("perl" ,perl)
1096        ("python" ,python-wrapper)))
1097     (native-inputs
1098      `(("cpio" ,cpio)))
1099     (home-page "http://blast.ncbi.nlm.nih.gov")
1100     (synopsis "Basic local alignment search tool")
1101     (description
1102      "BLAST is a popular method of performing a DNA or protein sequence
1103 similarity search, using heuristics to produce results quickly.  It also
1104 calculates an “expect value” that estimates how many matches would have
1105 occurred at a given score by chance, which can aid a user in judging how much
1106 confidence to have in an alignment.")
1107     ;; Most of the sources are in the public domain, with the following
1108     ;; exceptions:
1109     ;;   * Expat:
1110     ;;     * ./c++/include/util/bitset/
1111     ;;     * ./c++/src/html/ncbi_menu*.js
1112     ;;   * Boost license:
1113     ;;     * ./c++/include/util/impl/floating_point_comparison.hpp
1114     ;;   * LGPL 2+:
1115     ;;     * ./c++/include/dbapi/driver/odbc/unix_odbc/
1116     ;;   * ASL 2.0:
1117     ;;     * ./c++/src/corelib/teamcity_*
1118     (license (list license:public-domain
1119                    license:expat
1120                    license:boost1.0
1121                    license:lgpl2.0+
1122                    license:asl2.0))))
1123
1124 (define-public bless
1125   (package
1126     (name "bless")
1127     (version "1p02")
1128     (source (origin
1129               (method url-fetch)
1130               (uri (string-append "mirror://sourceforge/bless-ec/bless.v"
1131                                   version ".tgz"))
1132               (sha256
1133                (base32
1134                 "0rm0gw2s18dqwzzpl3c2x1z05ni2v0xz5dmfk3d33j6g4cgrlrdd"))
1135               (modules '((guix build utils)))
1136               (snippet
1137                `(begin
1138                   ;; Remove bundled boost, pigz, zlib, and .git directory
1139                   ;; FIXME: also remove bundled sources for murmurhash3 and
1140                   ;; kmc once packaged.
1141                   (delete-file-recursively "boost")
1142                   (delete-file-recursively "pigz")
1143                   (delete-file-recursively "google-sparsehash")
1144                   (delete-file-recursively "zlib")
1145                   (delete-file-recursively ".git")
1146                   #t))))
1147     (build-system gnu-build-system)
1148     (arguments
1149      '(#:tests? #f ;no "check" target
1150        #:make-flags
1151        (list (string-append "ZLIB="
1152                             (assoc-ref %build-inputs "zlib")
1153                             "/lib/libz.a")
1154              (string-append "LDFLAGS="
1155                             (string-join '("-lboost_filesystem"
1156                                            "-lboost_system"
1157                                            "-lboost_iostreams"
1158                                            "-lz"
1159                                            "-fopenmp"
1160                                            "-std=c++11"))))
1161        #:phases
1162        (modify-phases %standard-phases
1163          (add-after 'unpack 'do-not-build-bundled-pigz
1164           (lambda* (#:key inputs outputs #:allow-other-keys)
1165             (substitute* "Makefile"
1166               (("cd pigz/pigz-2.3.3; make") ""))
1167             #t))
1168          (add-after 'unpack 'patch-paths-to-executables
1169           (lambda* (#:key inputs outputs #:allow-other-keys)
1170             (substitute* "parse_args.cpp"
1171               (("kmc_binary = .*")
1172                (string-append "kmc_binary = \""
1173                               (assoc-ref outputs "out")
1174                               "/bin/kmc\";"))
1175               (("pigz_binary = .*")
1176                (string-append "pigz_binary = \""
1177                               (assoc-ref inputs "pigz")
1178                               "/bin/pigz\";")))
1179             #t))
1180          (replace 'install
1181           (lambda* (#:key outputs #:allow-other-keys)
1182             (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
1183               (for-each (lambda (file)
1184                           (install-file file bin))
1185                         '("bless" "kmc/bin/kmc"))
1186               #t)))
1187          (delete 'configure))))
1188     (native-inputs
1189      `(("perl" ,perl)))
1190     (inputs
1191      `(("openmpi" ,openmpi)
1192        ("boost" ,boost)
1193        ("sparsehash" ,sparsehash)
1194        ("pigz" ,pigz)
1195        ("zlib" ,zlib)))
1196     (supported-systems '("x86_64-linux"))
1197     (home-page "https://sourceforge.net/p/bless-ec/wiki/Home/")
1198     (synopsis "Bloom-filter-based error correction tool for NGS reads")
1199     (description
1200      "@dfn{Bloom-filter-based error correction solution for high-throughput
1201 sequencing reads} (BLESS) uses a single minimum-sized bloom filter is a
1202 correction tool for genomic reads produced by @dfn{Next-generation
1203 sequencing} (NGS).  BLESS produces accurate correction results with much less
1204 memory compared with previous solutions and is also able to tolerate a higher
1205 false-positive rate.  BLESS can extend reads like DNA assemblers to correct
1206 errors at the end of reads.")
1207     (license license:gpl3+)))
1208
1209 (define-public bowtie
1210   (package
1211     (name "bowtie")
1212     (version "2.3.2")
1213     (source (origin
1214               (method url-fetch)
1215               (uri (string-append "https://github.com/BenLangmead/bowtie2/archive/v"
1216                                   version ".tar.gz"))
1217               (file-name (string-append name "-" version ".tar.gz"))
1218               (sha256
1219                (base32
1220                 "0hwa5r9qbglppb7sz5z79rlmmddr3n51n468jb3wh8rwjgn3yr90"))
1221               (modules '((guix build utils)))
1222               (snippet
1223                '(substitute* "Makefile"
1224                   ;; replace BUILD_HOST and BUILD_TIME for deterministic build
1225                   (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
1226                   (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\"")))))
1227     (build-system gnu-build-system)
1228     (inputs
1229      `(("perl" ,perl)
1230        ("perl-clone" ,perl-clone)
1231        ("perl-test-deep" ,perl-test-deep)
1232        ("perl-test-simple" ,perl-test-simple)
1233        ("python" ,python-2)
1234        ("tbb" ,tbb)
1235        ("zlib" ,zlib)))
1236     (arguments
1237      '(#:make-flags
1238        (list "allall"
1239              "WITH_TBB=1"
1240              (string-append "prefix=" (assoc-ref %outputs "out")))
1241        #:phases
1242        (modify-phases %standard-phases
1243          (delete 'configure)
1244          (replace 'check
1245            (lambda* (#:key outputs #:allow-other-keys)
1246              (zero? (system* "perl"
1247                              "scripts/test/simple_tests.pl"
1248                              "--bowtie2=./bowtie2"
1249                              "--bowtie2-build=./bowtie2-build")))))))
1250     (home-page "http://bowtie-bio.sourceforge.net/bowtie2/index.shtml")
1251     (synopsis "Fast and sensitive nucleotide sequence read aligner")
1252     (description
1253      "Bowtie 2 is a fast and memory-efficient tool for aligning sequencing
1254 reads to long reference sequences.  It is particularly good at aligning reads
1255 of about 50 up to 100s or 1,000s of characters, and particularly good at
1256 aligning to relatively long (e.g. mammalian) genomes.  Bowtie 2 indexes the
1257 genome with an FM Index to keep its memory footprint small: for the human
1258 genome, its memory footprint is typically around 3.2 GB.  Bowtie 2 supports
1259 gapped, local, and paired-end alignment modes.")
1260     (supported-systems '("x86_64-linux"))
1261     (license license:gpl3+)))
1262
1263 (define-public tophat
1264   (package
1265     (name "tophat")
1266     (version "2.1.0")
1267     (source (origin
1268               (method url-fetch)
1269               (uri (string-append
1270                     "http://ccb.jhu.edu/software/tophat/downloads/tophat-"
1271                     version ".tar.gz"))
1272               (sha256
1273                (base32
1274                 "168zlzykq622zbgkh90a90f1bdgsxkscq2zxzbj8brq80hbjpyp7"))
1275               (patches (search-patches "tophat-build-with-later-seqan.patch"))
1276               (modules '((guix build utils)))
1277               (snippet
1278                '(begin
1279                   ;; Remove bundled SeqAn and samtools
1280                   (delete-file-recursively "src/SeqAn-1.3")
1281                   (delete-file-recursively "src/samtools-0.1.18")
1282                   #t))))
1283     (build-system gnu-build-system)
1284     (arguments
1285      '(#:parallel-build? #f ; not supported
1286        #:phases
1287        (modify-phases %standard-phases
1288          (add-after 'unpack 'use-system-samtools
1289            (lambda* (#:key inputs #:allow-other-keys)
1290              (substitute* "src/Makefile.in"
1291                (("(noinst_LIBRARIES = )\\$\\(SAMLIB\\)" _ prefix) prefix)
1292                (("\\$\\(SAMPROG\\): \\$\\(SAMLIB\\)") "")
1293                (("SAMPROG = samtools_0\\.1\\.18") "")
1294                (("\\$\\(samtools_0_1_18_SOURCES\\)") "")
1295                (("am__EXEEXT_1 = samtools_0\\.1\\.18\\$\\(EXEEXT\\)") ""))
1296              (substitute* '("src/common.cpp"
1297                             "src/tophat.py")
1298                (("samtools_0.1.18") (which "samtools")))
1299              (substitute* '("src/common.h"
1300                             "src/bam2fastx.cpp")
1301                (("#include \"bam.h\"") "#include <samtools/bam.h>")
1302                (("#include \"sam.h\"") "#include <samtools/sam.h>"))
1303              (substitute* '("src/bwt_map.h"
1304                             "src/map2gtf.h"
1305                             "src/align_status.h")
1306                (("#include <bam.h>") "#include <samtools/bam.h>")
1307                (("#include <sam.h>") "#include <samtools/sam.h>"))
1308              #t)))))
1309     (inputs
1310      `(("boost" ,boost)
1311        ("bowtie" ,bowtie)
1312        ("samtools" ,samtools-0.1)
1313        ("ncurses" ,ncurses)
1314        ("python" ,python-2)
1315        ("perl" ,perl)
1316        ("zlib" ,zlib)
1317        ("seqan" ,seqan)))
1318     (home-page "http://ccb.jhu.edu/software/tophat/index.shtml")
1319     (synopsis "Spliced read mapper for RNA-Seq data")
1320     (description
1321      "TopHat is a fast splice junction mapper for nucleotide sequence
1322 reads produced by the RNA-Seq method.  It aligns RNA-Seq reads to
1323 mammalian-sized genomes using the ultra high-throughput short read
1324 aligner Bowtie, and then analyzes the mapping results to identify
1325 splice junctions between exons.")
1326     ;; TopHat is released under the Boost Software License, Version 1.0
1327     ;; See https://github.com/infphilo/tophat/issues/11#issuecomment-121589893
1328     (license license:boost1.0)))
1329
1330 (define-public bwa
1331   (package
1332     (name "bwa")
1333     (version "0.7.17")
1334     (source (origin
1335               (method url-fetch)
1336               (uri (string-append
1337                     "https://github.com/lh3/bwa/releases/download/v"
1338                     version "/bwa-" version ".tar.bz2"))
1339               (sha256
1340                (base32
1341                 "1zfhv2zg9v1icdlq4p9ssc8k01mca5d1bd87w71py2swfi74s6yy"))))
1342     (build-system gnu-build-system)
1343     (arguments
1344      '(#:tests? #f ;no "check" target
1345        #:phases
1346        (modify-phases %standard-phases
1347          (replace 'install
1348            (lambda* (#:key outputs #:allow-other-keys)
1349              (let ((bin (string-append
1350                          (assoc-ref outputs "out") "/bin"))
1351                    (doc (string-append
1352                          (assoc-ref outputs "out") "/share/doc/bwa"))
1353                    (man (string-append
1354                          (assoc-ref outputs "out") "/share/man/man1")))
1355                (install-file "bwa" bin)
1356                (install-file "README.md" doc)
1357                (install-file "bwa.1" man))
1358              #t))
1359            ;; no "configure" script
1360           (delete 'configure))))
1361     (inputs `(("zlib" ,zlib)))
1362     ;; Non-portable SSE instructions are used so building fails on platforms
1363     ;; other than x86_64.
1364     (supported-systems '("x86_64-linux"))
1365     (home-page "http://bio-bwa.sourceforge.net/")
1366     (synopsis "Burrows-Wheeler sequence aligner")
1367     (description
1368      "BWA is a software package for mapping low-divergent sequences against a
1369 large reference genome, such as the human genome.  It consists of three
1370 algorithms: BWA-backtrack, BWA-SW and BWA-MEM.  The first algorithm is
1371 designed for Illumina sequence reads up to 100bp, while the rest two for
1372 longer sequences ranged from 70bp to 1Mbp.  BWA-MEM and BWA-SW share similar
1373 features such as long-read support and split alignment, but BWA-MEM, which is
1374 the latest, is generally recommended for high-quality queries as it is faster
1375 and more accurate.  BWA-MEM also has better performance than BWA-backtrack for
1376 70-100bp Illumina reads.")
1377     (license license:gpl3+)))
1378
1379 (define-public bwa-pssm
1380   (package (inherit bwa)
1381     (name "bwa-pssm")
1382     (version "0.5.11")
1383     (source (origin
1384               (method url-fetch)
1385               (uri (string-append "https://github.com/pkerpedjiev/bwa-pssm/"
1386                                   "archive/" version ".tar.gz"))
1387               (file-name (string-append name "-" version ".tar.gz"))
1388               (sha256
1389                (base32
1390                 "02p7mpbs4mlxmn84g2x4ghak638vbj4lqix2ipx5g84pz9bhdavg"))))
1391     (build-system gnu-build-system)
1392     (inputs
1393      `(("gdsl" ,gdsl)
1394        ("zlib" ,zlib)
1395        ("perl" ,perl)))
1396     (home-page "http://bwa-pssm.binf.ku.dk/")
1397     (synopsis "Burrows-Wheeler transform-based probabilistic short read mapper")
1398     (description
1399      "BWA-PSSM is a probabilistic short genomic sequence read aligner based on
1400 the use of @dfn{position specific scoring matrices} (PSSM).  Like many of the
1401 existing aligners it is fast and sensitive.  Unlike most other aligners,
1402 however, it is also adaptible in the sense that one can direct the alignment
1403 based on known biases within the data set.  It is coded as a modification of
1404 the original BWA alignment program and shares the genome index structure as
1405 well as many of the command line options.")
1406     (license license:gpl3+)))
1407
1408 (define-public python2-bx-python
1409   (package
1410     (name "python2-bx-python")
1411     (version "0.7.3")
1412     (source (origin
1413               (method url-fetch)
1414               (uri (pypi-uri "bx-python" version))
1415               (sha256
1416                (base32
1417                 "15z2w3bvnc0n4qmb9bd6d8ylc2h2nj883x2w9iixf4x3vki9b22i"))
1418               (modules '((guix build utils)))
1419               (snippet
1420                '(substitute* "setup.py"
1421                   ;; remove dependency on outdated "distribute" module
1422                   (("^from distribute_setup import use_setuptools") "")
1423                   (("^use_setuptools\\(\\)") "")))))
1424     (build-system python-build-system)
1425     (arguments
1426      `(#:tests? #f ;tests fail because test data are not included
1427        #:python ,python-2))
1428     (inputs
1429      `(("python-numpy" ,python2-numpy)
1430        ("zlib" ,zlib)))
1431     (native-inputs
1432      `(("python-nose" ,python2-nose)))
1433     (home-page "http://bitbucket.org/james_taylor/bx-python/")
1434     (synopsis "Tools for manipulating biological data")
1435     (description
1436      "bx-python provides tools for manipulating biological data, particularly
1437 multiple sequence alignments.")
1438     (license license:expat)))
1439
1440 (define-public python-pysam
1441   (package
1442     (name "python-pysam")
1443     (version "0.13.0")
1444     (source (origin
1445               (method url-fetch)
1446               ;; Test data is missing on PyPi.
1447               (uri (string-append
1448                     "https://github.com/pysam-developers/pysam/archive/v"
1449                     version ".tar.gz"))
1450               (file-name (string-append name "-" version ".tar.gz"))
1451               (sha256
1452                (base32
1453                 "0dzap2axin9cbbl0d825w294bpn00zagfm1sigamm4v2pm5bj9lp"))
1454               (modules '((guix build utils)))
1455               (snippet
1456                ;; Drop bundled htslib. TODO: Also remove samtools and bcftools.
1457                '(delete-file-recursively "htslib"))))
1458     (build-system python-build-system)
1459     (arguments
1460      `(#:modules ((ice-9 ftw)
1461                   (srfi srfi-26)
1462                   (guix build python-build-system)
1463                   (guix build utils))
1464        #:phases
1465        (modify-phases %standard-phases
1466          (add-before 'build 'set-flags
1467            (lambda* (#:key inputs #:allow-other-keys)
1468              (setenv "HTSLIB_MODE" "external")
1469              (setenv "HTSLIB_LIBRARY_DIR"
1470                      (string-append (assoc-ref inputs "htslib") "/lib"))
1471              (setenv "HTSLIB_INCLUDE_DIR"
1472                      (string-append (assoc-ref inputs "htslib") "/include"))
1473              (setenv "LDFLAGS" "-lncurses")
1474              (setenv "CFLAGS" "-D_CURSES_LIB=1")
1475              #t))
1476          (replace 'check
1477            (lambda* (#:key inputs outputs #:allow-other-keys)
1478              ;; Add first subdirectory of "build" directory to PYTHONPATH.
1479              (setenv "PYTHONPATH"
1480                      (string-append
1481                       (getenv "PYTHONPATH")
1482                       ":" (getcwd) "/build/"
1483                       (car (scandir "build"
1484                                     (negate (cut string-prefix? "." <>))))))
1485              ;; Step out of source dir so python does not import from CWD.
1486              (with-directory-excursion "tests"
1487                (setenv "HOME" "/tmp")
1488                (and (zero? (system* "make" "-C" "pysam_data"))
1489                     (zero? (system* "make" "-C" "cbcf_data"))
1490                     ;; Running nosetests without explicitly asking for a
1491                     ;; single process leads to a crash.  Running with multiple
1492                     ;; processes fails because the tests are not designed to
1493                     ;; run in parallel.
1494
1495                     ;; FIXME: tests keep timing out on some systems.
1496                     ;; (zero? (system* "nosetests" "-v"
1497                     ;;                 "--processes" "1"))
1498                     )))))))
1499     (propagated-inputs
1500      `(("htslib"            ,htslib))) ; Included from installed header files.
1501     (inputs
1502      `(("ncurses"           ,ncurses)
1503        ("zlib"              ,zlib)))
1504     (native-inputs
1505      `(("python-cython"     ,python-cython)
1506        ;; Dependencies below are are for tests only.
1507        ("samtools"          ,samtools)
1508        ("bcftools"          ,bcftools)
1509        ("python-nose"       ,python-nose)))
1510     (home-page "https://github.com/pysam-developers/pysam")
1511     (synopsis "Python bindings to the SAMtools C API")
1512     (description
1513      "Pysam is a Python module for reading and manipulating files in the
1514 SAM/BAM format.  Pysam is a lightweight wrapper of the SAMtools C API.  It
1515 also includes an interface for tabix.")
1516     (license license:expat)))
1517
1518 (define-public python2-pysam
1519   (package-with-python2 python-pysam))
1520
1521 (define-public python-twobitreader
1522   (package
1523     (name "python-twobitreader")
1524     (version "3.1.4")
1525     (source (origin
1526               (method url-fetch)
1527               (uri (pypi-uri "twobitreader" version))
1528               (sha256
1529                (base32
1530                 "1q8wnj2kga9nz1lwc4w7qv52smfm536hp6mc8w6s53lhyj0mpi22"))))
1531     (build-system python-build-system)
1532     (arguments
1533      '(;; Tests are not distributed in the PyPi release.
1534        ;; TODO Try building from the Git repo or asking the upstream maintainer
1535        ;; to distribute the tests on PyPi.
1536        #:tests? #f))
1537     (native-inputs
1538      `(("python-sphinx" ,python-sphinx)))
1539     (home-page "https://github.com/benjschiller/twobitreader")
1540     (synopsis "Python library for reading .2bit files")
1541     (description
1542      "twobitreader is a Python library for reading .2bit files as used by the
1543 UCSC genome browser.")
1544     (license license:artistic2.0)))
1545
1546 (define-public python2-twobitreader
1547   (package-with-python2 python-twobitreader))
1548
1549 (define-public python-plastid
1550   (package
1551     (name "python-plastid")
1552     (version "0.4.8")
1553     (source (origin
1554               (method url-fetch)
1555               (uri (pypi-uri "plastid" version))
1556               (sha256
1557                (base32
1558                 "0l24dd3q66if8yj042m4s0g95n6acn7im1imqd3p6h8ns43kxhj8"))))
1559     (build-system python-build-system)
1560     (arguments
1561      ;; Some test files are not included.
1562      `(#:tests? #f))
1563     (propagated-inputs
1564      `(("python-numpy" ,python-numpy)
1565        ("python-scipy" ,python-scipy)
1566        ("python-pandas" ,python-pandas)
1567        ("python-pysam" ,python-pysam)
1568        ("python-matplotlib" ,python-matplotlib)
1569        ("python-biopython" ,python-biopython)
1570        ("python-twobitreader" ,python-twobitreader)
1571        ("python-termcolor" ,python-termcolor)))
1572     (native-inputs
1573      `(("python-cython" ,python-cython)
1574        ("python-nose" ,python-nose)))
1575     (home-page "https://github.com/joshuagryphon/plastid")
1576     (synopsis "Python library for genomic analysis")
1577     (description
1578      "plastid is a Python library for genomic analysis – in particular,
1579 high-throughput sequencing data – with an emphasis on simplicity.")
1580     (license license:bsd-3)))
1581
1582 (define-public python2-plastid
1583   (package-with-python2 python-plastid))
1584
1585 (define-public cd-hit
1586   (package
1587     (name "cd-hit")
1588     (version "4.6.8")
1589     (source (origin
1590               (method url-fetch)
1591               (uri (string-append "https://github.com/weizhongli/cdhit"
1592                                   "/releases/download/V" version
1593                                   "/cd-hit-v" version
1594                                   "-2017-0621-source.tar.gz"))
1595               (sha256
1596                (base32
1597                 "1b4mwm2520ixjbw57sil20f9iixzw4bkdqqwgg1fc3pzm6rz4zmn"))))
1598     (build-system gnu-build-system)
1599     (arguments
1600      `(#:tests? #f ; there are no tests
1601        #:make-flags
1602        ;; Executables are copied directly to the PREFIX.
1603        (list (string-append "PREFIX=" (assoc-ref %outputs "out") "/bin"))
1604        #:phases
1605        (modify-phases %standard-phases
1606          ;; No "configure" script
1607          (delete 'configure)
1608          ;; Remove sources of non-determinism
1609          (add-after 'unpack 'be-timeless
1610            (lambda _
1611              (substitute* "cdhit-utility.c++"
1612                ((" \\(built on \" __DATE__ \"\\)") ""))
1613              (substitute* "cdhit-common.c++"
1614                (("__DATE__") "\"0\"")
1615                (("\", %s, \" __TIME__ \"\\\\n\", date") ""))
1616              #t))
1617          ;; The "install" target does not create the target directory.
1618          (add-before 'install 'create-target-dir
1619            (lambda* (#:key outputs #:allow-other-keys)
1620              (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
1621              #t)))))
1622     (inputs
1623      `(("perl" ,perl)))
1624     (home-page "http://weizhongli-lab.org/cd-hit/")
1625     (synopsis "Cluster and compare protein or nucleotide sequences")
1626     (description
1627      "CD-HIT is a program for clustering and comparing protein or nucleotide
1628 sequences.  CD-HIT is designed to be fast and handle extremely large
1629 databases.")
1630     ;; The manual says: "It can be copied under the GNU General Public License
1631     ;; version 2 (GPLv2)."
1632     (license license:gpl2)))
1633
1634 (define-public clipper
1635   (package
1636     (name "clipper")
1637     (version "1.1")
1638     (source (origin
1639               (method url-fetch)
1640               (uri (string-append
1641                     "https://github.com/YeoLab/clipper/archive/"
1642                     version ".tar.gz"))
1643               (file-name (string-append name "-" version ".tar.gz"))
1644               (sha256
1645                (base32
1646                 "0pflmsvhbf8izbgwhbhj1i7349sw1f55qpqj8ljmapp16hb0p0qi"))
1647               (modules '((guix build utils)))
1648               (snippet
1649                '(begin
1650                   ;; remove unnecessary setup dependency
1651                   (substitute* "setup.py"
1652                     (("setup_requires = .*") ""))
1653                   (for-each delete-file
1654                             '("clipper/src/peaks.so"
1655                               "clipper/src/readsToWiggle.so"))
1656                   (delete-file-recursively "dist/")
1657                   #t))))
1658     (build-system python-build-system)
1659     (arguments `(#:python ,python-2)) ; only Python 2 is supported
1660     (inputs
1661      `(("htseq" ,python2-htseq)
1662        ("python-pybedtools" ,python2-pybedtools)
1663        ("python-cython" ,python2-cython)
1664        ("python-scikit-learn" ,python2-scikit-learn)
1665        ("python-matplotlib" ,python2-matplotlib)
1666        ("python-pandas" ,python2-pandas)
1667        ("python-pysam" ,python2-pysam)
1668        ("python-numpy" ,python2-numpy)
1669        ("python-scipy" ,python2-scipy)))
1670     (native-inputs
1671      `(("python-mock" ,python2-mock)   ; for tests
1672        ("python-nose" ,python2-nose)   ; for tests
1673        ("python-pytz" ,python2-pytz))) ; for tests
1674     (home-page "https://github.com/YeoLab/clipper")
1675     (synopsis "CLIP peak enrichment recognition")
1676     (description
1677      "CLIPper is a tool to define peaks in CLIP-seq datasets.")
1678     (license license:gpl2)))
1679
1680 (define-public codingquarry
1681   (package
1682     (name "codingquarry")
1683     (version "2.0")
1684     (source (origin
1685               (method url-fetch)
1686               (uri (string-append
1687                     "mirror://sourceforge/codingquarry/CodingQuarry_v"
1688                     version ".tar.gz"))
1689               (sha256
1690                (base32
1691                 "0115hkjflsnfzn36xppwf9h9avfxlavr43djqmshkkzbgjzsz60i"))))
1692     (build-system gnu-build-system)
1693     (arguments
1694      '(#:tests? #f ; no "check" target
1695        #:phases
1696        (modify-phases %standard-phases
1697          (delete 'configure)
1698          (replace 'install
1699            (lambda* (#:key outputs #:allow-other-keys)
1700              (let* ((out (assoc-ref outputs "out"))
1701                     (bin (string-append out "/bin"))
1702                     (doc (string-append out "/share/doc/codingquarry")))
1703                (install-file "INSTRUCTIONS.pdf" doc)
1704                (copy-recursively "QuarryFiles"
1705                                  (string-append out "/QuarryFiles"))
1706                (install-file "CodingQuarry" bin)
1707                (install-file "CufflinksGTF_to_CodingQuarryGFF3.py" bin)))))))
1708     (inputs `(("openmpi" ,openmpi)))
1709     (native-search-paths
1710      (list (search-path-specification
1711             (variable "QUARRY_PATH")
1712             (files '("QuarryFiles")))))
1713     (native-inputs `(("python" ,python-2))) ; Only Python 2 is supported
1714     (synopsis "Fungal gene predictor")
1715     (description "CodingQuarry is a highly accurate, self-training GHMM fungal
1716 gene predictor designed to work with assembled, aligned RNA-seq transcripts.")
1717     (home-page "https://sourceforge.net/projects/codingquarry/")
1718     (license license:gpl3+)))
1719
1720 (define-public couger
1721   (package
1722     (name "couger")
1723     (version "1.8.2")
1724     (source (origin
1725               (method url-fetch)
1726               (uri (string-append
1727                     "http://couger.oit.duke.edu/static/assets/COUGER"
1728                     version ".zip"))
1729               (sha256
1730                (base32
1731                 "04p2b14nmhzxw5h72mpzdhalv21bx4w9b87z0wpw0xzxpysyncmq"))))
1732     (build-system gnu-build-system)
1733     (arguments
1734      `(#:tests? #f
1735        #:phases
1736        (modify-phases %standard-phases
1737          (delete 'configure)
1738          (delete 'build)
1739          (replace
1740           'install
1741           (lambda* (#:key outputs #:allow-other-keys)
1742             (let* ((out (assoc-ref outputs "out"))
1743                    (bin (string-append out "/bin")))
1744               (copy-recursively "src" (string-append out "/src"))
1745               (mkdir bin)
1746               ;; Add "src" directory to module lookup path.
1747               (substitute* "couger"
1748                 (("from argparse")
1749                  (string-append "import sys\nsys.path.append(\""
1750                                 out "\")\nfrom argparse")))
1751               (install-file "couger" bin))
1752             #t))
1753          (add-after
1754           'install 'wrap-program
1755           (lambda* (#:key inputs outputs #:allow-other-keys)
1756             ;; Make sure 'couger' runs with the correct PYTHONPATH.
1757             (let* ((out (assoc-ref outputs "out"))
1758                    (path (getenv "PYTHONPATH")))
1759               (wrap-program (string-append out "/bin/couger")
1760                 `("PYTHONPATH" ":" prefix (,path))))
1761             #t)))))
1762     (inputs
1763      `(("python" ,python-2)
1764        ("python2-pillow" ,python2-pillow)
1765        ("python2-numpy" ,python2-numpy)
1766        ("python2-scipy" ,python2-scipy)
1767        ("python2-matplotlib" ,python2-matplotlib)))
1768     (propagated-inputs
1769      `(("r-minimal" ,r-minimal)
1770        ("libsvm" ,libsvm)
1771        ("randomjungle" ,randomjungle)))
1772     (native-inputs
1773      `(("unzip" ,unzip)))
1774     (home-page "http://couger.oit.duke.edu")
1775     (synopsis "Identify co-factors in sets of genomic regions")
1776     (description
1777      "COUGER can be applied to any two sets of genomic regions bound by
1778 paralogous TFs (e.g., regions derived from ChIP-seq experiments) to identify
1779 putative co-factors that provide specificity to each TF.  The framework
1780 determines the genomic targets uniquely-bound by each TF, and identifies a
1781 small set of co-factors that best explain the in vivo binding differences
1782 between the two TFs.
1783
1784 COUGER uses classification algorithms (support vector machines and random
1785 forests) with features that reflect the DNA binding specificities of putative
1786 co-factors.  The features are generated either from high-throughput TF-DNA
1787 binding data (from protein binding microarray experiments), or from large
1788 collections of DNA motifs.")
1789     (license license:gpl3+)))
1790
1791 (define-public clustal-omega
1792   (package
1793     (name "clustal-omega")
1794     (version "1.2.4")
1795     (source (origin
1796               (method url-fetch)
1797               (uri (string-append "http://www.clustal.org/omega/clustal-omega-"
1798                                   version ".tar.gz"))
1799               (sha256
1800                (base32
1801                 "1vm30mzncwdv881vrcwg11vzvrsmwy4wg80j5i0lcfk6dlld50w6"))))
1802     (build-system gnu-build-system)
1803     (inputs
1804      `(("argtable" ,argtable)))
1805     (home-page "http://www.clustal.org/omega/")
1806     (synopsis "Multiple sequence aligner for protein and DNA/RNA")
1807     (description
1808      "Clustal-Omega is a general purpose multiple sequence alignment (MSA)
1809 program for protein and DNA/RNA.  It produces high quality MSAs and is capable
1810 of handling data-sets of hundreds of thousands of sequences in reasonable
1811 time.")
1812     (license license:gpl2+)))
1813
1814 (define-public crossmap
1815   (package
1816     (name "crossmap")
1817     (version "0.2.1")
1818     (source (origin
1819               (method url-fetch)
1820               (uri (string-append "mirror://sourceforge/crossmap/CrossMap-"
1821                                   version ".tar.gz"))
1822               (sha256
1823                (base32
1824                 "07y179f63d7qnzdvkqcziwk9bs3k4zhp81q392fp1hwszjdvy22f"))
1825               ;; This patch has been sent upstream already and is available
1826               ;; for download from Sourceforge, but it has not been merged.
1827               (patches (search-patches "crossmap-allow-system-pysam.patch"))
1828               (modules '((guix build utils)))
1829               ;; remove bundled copy of pysam
1830               (snippet
1831                '(delete-file-recursively "lib/pysam"))))
1832     (build-system python-build-system)
1833     (arguments
1834      `(#:python ,python-2
1835        #:phases
1836        (modify-phases %standard-phases
1837          (add-after 'unpack 'set-env
1838            (lambda _ (setenv "CROSSMAP_USE_SYSTEM_PYSAM" "1") #t)))))
1839     (inputs
1840      `(("python-numpy" ,python2-numpy)
1841        ("python-pysam" ,python2-pysam)
1842        ("zlib" ,zlib)))
1843     (native-inputs
1844      `(("python-cython" ,python2-cython)
1845        ("python-nose" ,python2-nose)))
1846     (home-page "http://crossmap.sourceforge.net/")
1847     (synopsis "Convert genome coordinates between assemblies")
1848     (description
1849      "CrossMap is a program for conversion of genome coordinates or annotation
1850 files between different genome assemblies.  It supports most commonly used
1851 file formats including SAM/BAM, Wiggle/BigWig, BED, GFF/GTF, VCF.")
1852     (license license:gpl2+)))
1853
1854 (define-public cutadapt
1855   (package
1856     (name "cutadapt")
1857     (version "1.14")
1858     (source (origin
1859               (method url-fetch)
1860               (uri (string-append
1861                     "https://github.com/marcelm/cutadapt/archive/v"
1862                     version ".tar.gz"))
1863               (file-name (string-append name "-" version ".tar.gz"))
1864               (sha256
1865                (base32
1866                 "16gbpiwy4m48vq2h5wqar3i8vr6vcj9gcl2qvqim19x6ya9dp8kd"))))
1867     (build-system python-build-system)
1868     (arguments
1869      `(#:phases
1870        (modify-phases %standard-phases
1871          ;; The tests must be run after installation.
1872          (delete 'check)
1873          (add-after 'install 'check
1874            (lambda* (#:key inputs outputs #:allow-other-keys)
1875              (setenv "PYTHONPATH"
1876                      (string-append
1877                       (getenv "PYTHONPATH")
1878                       ":" (assoc-ref outputs "out")
1879                       "/lib/python"
1880                       (string-take (string-take-right
1881                                     (assoc-ref inputs "python") 5) 3)
1882                       "/site-packages"))
1883              (zero? (system* "nosetests" "-P" "tests")))))))
1884     (inputs
1885      `(("python-xopen" ,python-xopen)))
1886     (native-inputs
1887      `(("python-cython" ,python-cython)
1888        ("python-nose" ,python-nose)))
1889     (home-page "https://cutadapt.readthedocs.io/en/stable/")
1890     (synopsis "Remove adapter sequences from nucleotide sequencing reads")
1891     (description
1892      "Cutadapt finds and removes adapter sequences, primers, poly-A tails and
1893 other types of unwanted sequence from high-throughput sequencing reads.")
1894     (license license:expat)))
1895
1896 (define-public libbigwig
1897   (package
1898     (name "libbigwig")
1899     (version "0.1.4")
1900     (source (origin
1901               (method url-fetch)
1902               (uri (string-append "https://github.com/dpryan79/libBigWig/"
1903                                   "archive/" version ".tar.gz"))
1904               (file-name (string-append name "-" version ".tar.gz"))
1905               (sha256
1906                (base32
1907                 "098rjh35pi4a9q83n8wiwvyzykjqj6l8q189p1xgfw4ghywdlvw1"))))
1908     (build-system gnu-build-system)
1909     (arguments
1910      `(#:test-target "test"
1911        #:make-flags
1912        (list "CC=gcc"
1913              (string-append "prefix=" (assoc-ref %outputs "out")))
1914        #:phases
1915        (modify-phases %standard-phases
1916          (delete 'configure)
1917          (add-before 'check 'disable-curl-test
1918            (lambda _
1919              (substitute* "Makefile"
1920                (("./test/testRemote.*") ""))
1921              #t))
1922          ;; This has been fixed with the upstream commit 4ff6959cd8a0, but
1923          ;; there has not yet been a release containing this change.
1924          (add-before 'install 'create-target-dirs
1925            (lambda* (#:key outputs #:allow-other-keys)
1926              (let ((out (assoc-ref outputs "out")))
1927                (mkdir-p (string-append out "/lib"))
1928                (mkdir-p (string-append out "/include"))
1929                #t))))))
1930     (inputs
1931      `(("zlib" ,zlib)
1932        ("curl" ,curl)))
1933     (native-inputs
1934      `(("doxygen" ,doxygen)))
1935     (home-page "https://github.com/dpryan79/libBigWig")
1936     (synopsis "C library for handling bigWig files")
1937     (description
1938      "This package provides a C library for parsing local and remote BigWig
1939 files.")
1940     (license license:expat)))
1941
1942 (define-public python-pybigwig
1943   (package
1944     (name "python-pybigwig")
1945     (version "0.2.5")
1946     (source (origin
1947               (method url-fetch)
1948               (uri (pypi-uri "pyBigWig" version))
1949               (sha256
1950                (base32
1951                 "0yrpdxg3y0sny25x4w22lv1k47jzccqjmg7j4bp0hywklvp0hg7d"))
1952               (modules '((guix build utils)))
1953               (snippet
1954                '(begin
1955                   ;; Delete bundled libBigWig sources
1956                   (delete-file-recursively "libBigWig")))))
1957     (build-system python-build-system)
1958     (arguments
1959      `(#:phases
1960        (modify-phases %standard-phases
1961          (add-after 'unpack 'link-with-libBigWig
1962            (lambda* (#:key inputs #:allow-other-keys)
1963              (substitute* "setup.py"
1964                (("libs=\\[") "libs=[\"BigWig\", "))
1965              #t)))))
1966     (inputs
1967      `(("libbigwig" ,libbigwig)
1968        ("zlib" ,zlib)
1969        ("curl" ,curl)))
1970     (home-page "https://github.com/dpryan79/pyBigWig")
1971     (synopsis "Access bigWig files in Python using libBigWig")
1972     (description
1973      "This package provides Python bindings to the libBigWig library for
1974 accessing bigWig files.")
1975     (license license:expat)))
1976
1977 (define-public python2-pybigwig
1978   (package-with-python2 python-pybigwig))
1979
1980 (define-public python-dendropy
1981   (package
1982     (name "python-dendropy")
1983     (version "4.2.0")
1984     (source
1985      (origin
1986        (method url-fetch)
1987        (uri (pypi-uri "DendroPy" version))
1988        (sha256
1989         (base32
1990          "15c7s3d5gf19ljsxvq5advaa752wfi7pwrdjyhzmg85hccyvp47p"))
1991        (patches (search-patches "python-dendropy-fix-tests.patch"))))
1992     (build-system python-build-system)
1993     (home-page "http://packages.python.org/DendroPy/")
1994     (synopsis "Library for phylogenetics and phylogenetic computing")
1995     (description
1996      "DendroPy is a library for phylogenetics and phylogenetic computing: reading,
1997 writing, simulation, processing and manipulation of phylogenetic
1998 trees (phylogenies) and characters.")
1999     (license license:bsd-3)
2000     (properties `((python2-variant . ,(delay python2-dendropy))))))
2001
2002 (define-public python2-dendropy
2003   (let ((base (package-with-python2 (strip-python2-variant python-dendropy))))
2004     (package
2005       (inherit base)
2006       (arguments
2007        `(#:python ,python-2
2008          #:phases
2009            (modify-phases %standard-phases
2010              (replace 'check
2011                ;; There is currently a test failure that only happens on some
2012                ;; systems, and only using "setup.py test"
2013                (lambda _ (zero? (system* "nosetests")))))))
2014       (native-inputs `(("python2-nose" ,python2-nose)
2015                        ,@(package-native-inputs base))))))
2016
2017 (define-public python-py2bit
2018   (package
2019     (name "python-py2bit")
2020     (version "0.2.1")
2021     (source
2022      (origin
2023        (method url-fetch)
2024        (uri (pypi-uri "py2bit" version))
2025        (sha256
2026         (base32
2027          "1cdf4qlmgwsh1f4k0wdv2sr8x9qn4366p0k3614vbd0fpqiarxrl"))))
2028     (build-system python-build-system)
2029     (home-page "https://github.com/dpryan79/py2bit")
2030     (synopsis "Access 2bit files using lib2bit")
2031     (description
2032      "This package provides Python bindings for lib2bit to access 2bit files
2033 with Python.")
2034     (license license:expat)))
2035
2036 (define-public deeptools
2037   (package
2038     (name "deeptools")
2039     (version "2.5.1")
2040     (source (origin
2041               (method url-fetch)
2042               (uri (string-append "https://github.com/deeptools/deepTools/"
2043                                   "archive/" version ".tar.gz"))
2044               (file-name (string-append name "-" version ".tar.gz"))
2045               (sha256
2046                (base32
2047                 "1q8i12l2gvk4n2s8lhyzwhh9g4qbc8lrk5l7maz00yvd5g6z5540"))))
2048     (build-system python-build-system)
2049     (inputs
2050      `(("python-scipy" ,python-scipy)
2051        ("python-numpy" ,python-numpy)
2052        ("python-numpydoc" ,python-numpydoc)
2053        ("python-matplotlib" ,python-matplotlib)
2054        ("python-pysam" ,python-pysam)
2055        ("python-py2bit" ,python-py2bit)
2056        ("python-pybigwig" ,python-pybigwig)))
2057     (native-inputs
2058      `(("python-mock" ,python-mock)   ;for tests
2059        ("python-nose" ,python-nose)   ;for tests
2060        ("python-pytz" ,python-pytz))) ;for tests
2061     (home-page "https://github.com/deeptools/deepTools")
2062     (synopsis "Tools for normalizing and visualizing deep-sequencing data")
2063     (description
2064      "DeepTools addresses the challenge of handling the large amounts of data
2065 that are now routinely generated from DNA sequencing centers.  To do so,
2066 deepTools contains useful modules to process the mapped reads data to create
2067 coverage files in standard bedGraph and bigWig file formats.  By doing so,
2068 deepTools allows the creation of normalized coverage files or the comparison
2069 between two files (for example, treatment and control).  Finally, using such
2070 normalized and standardized files, multiple visualizations can be created to
2071 identify enrichments with functional annotations of the genome.")
2072     (license license:gpl3+)))
2073
2074 (define-public diamond
2075   (package
2076     (name "diamond")
2077     (version "0.9.18")
2078     (source (origin
2079               (method url-fetch)
2080               (uri (string-append
2081                     "https://github.com/bbuchfink/diamond/archive/v"
2082                     version ".tar.gz"))
2083               (file-name (string-append name "-" version ".tar.gz"))
2084               (sha256
2085                (base32
2086                 "1vi2nddmy7knrv8gsprwqp6a40k63n3f2dfvx22ipjhrg9xir96f"))))
2087     (build-system cmake-build-system)
2088     (arguments
2089      '(#:tests? #f ; no "check" target
2090        #:phases
2091        (modify-phases %standard-phases
2092          (add-after 'unpack 'remove-native-compilation
2093            (lambda _
2094              (substitute* "CMakeLists.txt" (("-march=native") ""))
2095              #t)))))
2096     (inputs
2097      `(("zlib" ,zlib)))
2098     (home-page "https://github.com/bbuchfink/diamond")
2099     (synopsis "Accelerated BLAST compatible local sequence aligner")
2100     (description
2101      "DIAMOND is a BLAST-compatible local aligner for mapping protein and
2102 translated DNA query sequences against a protein reference database (BLASTP
2103 and BLASTX alignment mode).  The speedup over BLAST is up to 20,000 on short
2104 reads at a typical sensitivity of 90-99% relative to BLAST depending on the
2105 data and settings.")
2106     (license license:agpl3+)))
2107
2108 (define-public discrover
2109   (package
2110     (name "discrover")
2111     (version "1.6.0")
2112     (source
2113      (origin
2114        (method url-fetch)
2115        (uri (string-append "https://github.com/maaskola/discrover/archive/"
2116                            version ".tar.gz"))
2117        (file-name (string-append name "-" version ".tar.gz"))
2118        (sha256
2119         (base32
2120          "0rah9ja4m0rl5mldd6vag9rwrivw1zrqxssfq8qx64m7961fp68k"))))
2121     (build-system cmake-build-system)
2122     (arguments
2123      `(#:tests? #f                      ; there are no tests
2124        #:phases
2125        (modify-phases %standard-phases
2126          (add-after 'unpack 'add-missing-includes
2127            (lambda _
2128              (substitute* "src/executioninformation.hpp"
2129                (("#define EXECUTIONINFORMATION_HPP" line)
2130                 (string-append line "\n#include <random>")))
2131              (substitute* "src/plasma/fasta.hpp"
2132                (("#define FASTA_HPP" line)
2133                 (string-append line "\n#include <random>")))
2134              #t)))))
2135     (inputs
2136      `(("boost" ,boost)
2137        ("cairo" ,cairo)))
2138     (native-inputs
2139      `(("texlive" ,texlive)
2140        ("imagemagick" ,imagemagick)))
2141     (home-page "http://dorina.mdc-berlin.de/public/rajewsky/discrover/")
2142     (synopsis "Discover discriminative nucleotide sequence motifs")
2143     (description "Discrover is a motif discovery method to find binding sites
2144 of nucleic acid binding proteins.")
2145     (license license:gpl3+)))
2146
2147 (define-public eigensoft
2148   (let ((revision "1")
2149         (commit "b14d1e202e21e532536ff8004f0419cd5e259dc7"))
2150     (package
2151       (name "eigensoft")
2152       (version (string-append "6.1.2-"
2153                               revision "."
2154                               (string-take commit 9)))
2155       (source
2156        (origin
2157          (method git-fetch)
2158          (uri (git-reference
2159                (url "https://github.com/DReichLab/EIG.git")
2160                (commit commit)))
2161          (file-name (string-append "eigensoft-" commit "-checkout"))
2162          (sha256
2163           (base32
2164            "0f5m6k2j5c16xc3xbywcs989xyc26ncy1zfzp9j9n55n9r4xcaiq"))
2165          (modules '((guix build utils)))
2166          ;; Remove pre-built binaries.
2167          (snippet '(begin
2168                      (delete-file-recursively "bin")
2169                      (mkdir "bin")
2170                      #t))))
2171       (build-system gnu-build-system)
2172       (arguments
2173        `(#:tests? #f                    ; There are no tests.
2174          #:make-flags '("CC=gcc")
2175          #:phases
2176          (modify-phases %standard-phases
2177            ;; There is no configure phase, but the Makefile is in a
2178            ;; sub-directory.
2179            (replace 'configure
2180              (lambda _
2181                (chdir "src")
2182                ;; The link flags are incomplete.
2183                (substitute* "Makefile"
2184                  (("-lgsl") "-lgsl -lm -llapack -llapacke -lpthread"))
2185                #t))
2186            ;; The provided install target only copies executables to
2187            ;; the "bin" directory in the build root.
2188            (add-after 'install 'actually-install
2189              (lambda* (#:key outputs #:allow-other-keys)
2190                (let* ((out (assoc-ref outputs "out"))
2191                       (bin  (string-append out "/bin")))
2192                  (for-each (lambda (file)
2193                              (install-file file bin))
2194                            (find-files "../bin" ".*"))
2195                  #t))))))
2196       (inputs
2197        `(("gsl" ,gsl)
2198          ("lapack" ,lapack)
2199          ("openblas" ,openblas)
2200          ("perl" ,perl)
2201          ("gfortran" ,gfortran "lib")))
2202       (home-page "https://github.com/DReichLab/EIG")
2203       (synopsis "Tools for population genetics")
2204       (description "The EIGENSOFT package provides tools for population
2205 genetics and stratification correction.  EIGENSOFT implements methods commonly
2206 used in population genetics analyses such as PCA, computation of Tracy-Widom
2207 statistics, and finding related individuals in structured populations.  It
2208 comes with a built-in plotting script and supports multiple file formats and
2209 quantitative phenotypes.")
2210       ;; The license of the eigensoft tools is Expat, but since it's
2211       ;; linking with the GNU Scientific Library (GSL) the effective
2212       ;; license is the GPL.
2213       (license license:gpl3+))))
2214
2215 (define-public edirect
2216   (package
2217     (name "edirect")
2218     (version "4.10")
2219     (source (origin
2220               (method url-fetch)
2221               (uri (string-append "ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/"
2222                                   "versions/2016-05-03/edirect.tar.gz"))
2223               (sha256
2224                (base32
2225                 "15zsprak5yh8c1yrz4r1knmb5s8qcmdid4xdhkh3lqcv64l60hli"))))
2226     (build-system perl-build-system)
2227     (arguments
2228      `(#:tests? #f ;no "check" target
2229        #:phases
2230        (modify-phases %standard-phases
2231          (delete 'configure)
2232          (delete 'build)
2233          (replace 'install
2234                   (lambda* (#:key outputs #:allow-other-keys)
2235                     (let ((target (string-append (assoc-ref outputs "out")
2236                                                  "/bin")))
2237                       (mkdir-p target)
2238                       (install-file "edirect.pl" target)
2239                       #t)))
2240          (add-after
2241           'install 'wrap-program
2242           (lambda* (#:key inputs outputs #:allow-other-keys)
2243             ;; Make sure 'edirect.pl' finds all perl inputs at runtime.
2244             (let* ((out (assoc-ref outputs "out"))
2245                    (path (getenv "PERL5LIB")))
2246               (wrap-program (string-append out "/bin/edirect.pl")
2247                 `("PERL5LIB" ":" prefix (,path)))))))))
2248     (inputs
2249      `(("perl-html-parser" ,perl-html-parser)
2250        ("perl-encode-locale" ,perl-encode-locale)
2251        ("perl-file-listing" ,perl-file-listing)
2252        ("perl-html-tagset" ,perl-html-tagset)
2253        ("perl-html-tree" ,perl-html-tree)
2254        ("perl-http-cookies" ,perl-http-cookies)
2255        ("perl-http-date" ,perl-http-date)
2256        ("perl-http-message" ,perl-http-message)
2257        ("perl-http-negotiate" ,perl-http-negotiate)
2258        ("perl-lwp-mediatypes" ,perl-lwp-mediatypes)
2259        ("perl-lwp-protocol-https" ,perl-lwp-protocol-https)
2260        ("perl-net-http" ,perl-net-http)
2261        ("perl-uri" ,perl-uri)
2262        ("perl-www-robotrules" ,perl-www-robotrules)
2263        ("perl" ,perl)))
2264     (home-page "http://www.ncbi.nlm.nih.gov/books/NBK179288/")
2265     (synopsis "Tools for accessing the NCBI's set of databases")
2266     (description
2267      "Entrez Direct (EDirect) is a method for accessing the National Center
2268 for Biotechnology Information's (NCBI) set of interconnected
2269 databases (publication, sequence, structure, gene, variation, expression,
2270 etc.) from a terminal.  Functions take search terms from command-line
2271 arguments.  Individual operations are combined to build multi-step queries.
2272 Record retrieval and formatting normally complete the process.
2273
2274 EDirect also provides an argument-driven function that simplifies the
2275 extraction of data from document summaries or other results that are returned
2276 in structured XML format.  This can eliminate the need for writing custom
2277 software to answer ad hoc questions.")
2278     (license license:public-domain)))
2279
2280 (define-public exonerate
2281   (package
2282     (name "exonerate")
2283     (version "2.4.0")
2284     (source
2285      (origin
2286        (method url-fetch)
2287        (uri
2288         (string-append
2289          "http://ftp.ebi.ac.uk/pub/software/vertebrategenomics/exonerate/"
2290          "exonerate-" version ".tar.gz"))
2291        (sha256
2292         (base32
2293          "0hj0m9xygiqsdxvbg79wq579kbrx1mdrabi2bzqz2zn9qwfjcjgq"))))
2294     (build-system gnu-build-system)
2295     (arguments
2296      `(#:parallel-build? #f)) ; Building in parallel fails on some machines.
2297     (native-inputs
2298      `(("pkg-config" ,pkg-config)))
2299     (inputs
2300      `(("glib" ,glib)))
2301     (home-page
2302      "https://www.ebi.ac.uk/about/vertebrate-genomics/software/exonerate")
2303     (synopsis "Generic tool for biological sequence alignment")
2304     (description
2305      "Exonerate is a generic tool for pairwise sequence comparison.  It allows
2306 the alignment of sequences using a many alignment models, either exhaustive
2307 dynamic programming or a variety of heuristics.")
2308     (license license:gpl3)))
2309
2310 (define-public express
2311   (package
2312     (name "express")
2313     (version "1.5.1")
2314     (source (origin
2315               (method url-fetch)
2316               (uri
2317                (string-append
2318                 "http://bio.math.berkeley.edu/eXpress/downloads/express-"
2319                 version "/express-" version "-src.tgz"))
2320               (sha256
2321                (base32
2322                 "03rczxd0gjp2l1jxcmjfmf5j94j77zqyxa6x063zsc585nj40n0c"))))
2323     (build-system cmake-build-system)
2324     (arguments
2325      `(#:tests? #f ;no "check" target
2326        #:phases
2327        (modify-phases %standard-phases
2328          (add-after 'unpack 'use-shared-boost-libs-and-set-bamtools-paths
2329            (lambda* (#:key inputs #:allow-other-keys)
2330              (substitute* "CMakeLists.txt"
2331                (("set\\(Boost_USE_STATIC_LIBS ON\\)")
2332                 "set(Boost_USE_STATIC_LIBS OFF)")
2333                (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/bamtools/include")
2334                 (string-append (assoc-ref inputs "bamtools") "/include/bamtools")))
2335              (substitute* "src/CMakeLists.txt"
2336                (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/\\.\\./bamtools/lib")
2337                 (string-append (assoc-ref inputs "bamtools") "/lib/bamtools")))
2338              #t)))))
2339     (inputs
2340      `(("boost" ,boost)
2341        ("bamtools" ,bamtools)
2342        ("protobuf" ,protobuf)
2343        ("zlib" ,zlib)))
2344     (home-page "http://bio.math.berkeley.edu/eXpress")
2345     (synopsis "Streaming quantification for high-throughput genomic sequencing")
2346     (description
2347      "eXpress is a streaming tool for quantifying the abundances of a set of
2348 target sequences from sampled subsequences.  Example applications include
2349 transcript-level RNA-Seq quantification, allele-specific/haplotype expression
2350 analysis (from RNA-Seq), transcription factor binding quantification in
2351 ChIP-Seq, and analysis of metagenomic data.")
2352     (license license:artistic2.0)))
2353
2354 (define-public express-beta-diversity
2355   (package
2356    (name "express-beta-diversity")
2357    (version "1.0.7")
2358    (source (origin
2359              (method url-fetch)
2360              (uri
2361               (string-append
2362                "https://github.com/dparks1134/ExpressBetaDiversity/archive/v"
2363                version ".tar.gz"))
2364              (file-name (string-append name "-" version ".tar.gz"))
2365              (sha256
2366               (base32
2367                "1djvdlmqvjf6h0zq7w36y8cl5cli6rgj86x65znl48agnwmzxfxr"))))
2368    (build-system gnu-build-system)
2369    (arguments
2370     `(#:phases
2371       (modify-phases %standard-phases
2372         (delete 'configure)
2373         (add-before 'build 'enter-source (lambda _ (chdir "source") #t))
2374         (replace 'check
2375                  (lambda _ (zero? (system* "../bin/ExpressBetaDiversity"
2376                                            "-u"))))
2377         (add-after 'check 'exit-source (lambda _ (chdir "..") #t))
2378         (replace 'install
2379                  (lambda* (#:key outputs #:allow-other-keys)
2380                    (let ((bin (string-append (assoc-ref outputs "out")
2381                                              "/bin")))
2382                      (mkdir-p bin)
2383                      (install-file "scripts/convertToEBD.py" bin)
2384                      (install-file "bin/ExpressBetaDiversity" bin)
2385                      #t))))))
2386    (inputs
2387     `(("python" ,python-2)))
2388    (home-page "http://kiwi.cs.dal.ca/Software/ExpressBetaDiversity")
2389    (synopsis "Taxon- and phylogenetic-based beta diversity measures")
2390    (description
2391     "Express Beta Diversity (EBD) calculates ecological beta diversity
2392 (dissimilarity) measures between biological communities.  EBD implements a
2393 variety of diversity measures including those that make use of phylogenetic
2394 similarity of community members.")
2395    (license license:gpl3+)))
2396
2397 (define-public fasttree
2398   (package
2399    (name "fasttree")
2400    (version "2.1.10")
2401    (source (origin
2402              (method url-fetch)
2403              (uri (string-append
2404                    "http://www.microbesonline.org/fasttree/FastTree-"
2405                    version ".c"))
2406              (sha256
2407               (base32
2408                "0vcjdvy1j4m702vmak4svbfkrpcw63k7wymfksjp9a982zy8kjsl"))))
2409    (build-system gnu-build-system)
2410    (arguments
2411     `(#:tests? #f ; no "check" target
2412       #:phases
2413       (modify-phases %standard-phases
2414         (delete 'unpack)
2415         (delete 'configure)
2416         (replace 'build
2417           (lambda* (#:key source #:allow-other-keys)
2418             (and (zero? (system* "gcc"
2419                                  "-O3"
2420                                  "-finline-functions"
2421                                  "-funroll-loops"
2422                                  "-Wall"
2423                                  "-o"
2424                                  "FastTree"
2425                                  source
2426                                  "-lm"))
2427                  (zero? (system* "gcc"
2428                                  "-DOPENMP"
2429                                  "-fopenmp"
2430                                  "-O3"
2431                                  "-finline-functions"
2432                                  "-funroll-loops"
2433                                  "-Wall"
2434                                  "-o"
2435                                  "FastTreeMP"
2436                                  source
2437                                  "-lm")))))
2438         (replace 'install
2439           (lambda* (#:key outputs #:allow-other-keys)
2440             (let ((bin (string-append (assoc-ref outputs "out")
2441                                       "/bin")))
2442               (mkdir-p bin)
2443               (install-file "FastTree" bin)
2444               (install-file "FastTreeMP" bin)
2445               #t))))))
2446    (home-page "http://www.microbesonline.org/fasttree")
2447    (synopsis "Infers approximately-maximum-likelihood phylogenetic trees")
2448    (description
2449     "FastTree can handle alignments with up to a million of sequences in a
2450 reasonable amount of time and memory.  For large alignments, FastTree is
2451 100-1,000 times faster than PhyML 3.0 or RAxML 7.")
2452    (license license:gpl2+)))
2453
2454 (define-public fastx-toolkit
2455   (package
2456     (name "fastx-toolkit")
2457     (version "0.0.14")
2458     (source (origin
2459               (method url-fetch)
2460               (uri
2461                (string-append
2462                 "https://github.com/agordon/fastx_toolkit/releases/download/"
2463                 version "/fastx_toolkit-" version ".tar.bz2"))
2464               (sha256
2465                (base32
2466                 "01jqzw386873sr0pjp1wr4rn8fsga2vxs1qfmicvx1pjr72007wy"))))
2467     (build-system gnu-build-system)
2468     (inputs
2469      `(("libgtextutils" ,libgtextutils)))
2470     (native-inputs
2471      `(("pkg-config" ,pkg-config)))
2472     (home-page "http://hannonlab.cshl.edu/fastx_toolkit/")
2473     (synopsis "Tools for FASTA/FASTQ file preprocessing")
2474     (description
2475      "The FASTX-Toolkit is a collection of command line tools for Short-Reads
2476 FASTA/FASTQ files preprocessing.
2477
2478 Next-Generation sequencing machines usually produce FASTA or FASTQ files,
2479 containing multiple short-reads sequences.  The main processing of such
2480 FASTA/FASTQ files is mapping the sequences to reference genomes.  However, it
2481 is sometimes more productive to preprocess the files before mapping the
2482 sequences to the genome---manipulating the sequences to produce better mapping
2483 results.  The FASTX-Toolkit tools perform some of these preprocessing tasks.")
2484     (license license:agpl3+)))
2485
2486 (define-public flexbar
2487   (package
2488     (name "flexbar")
2489     (version "2.5")
2490     (source (origin
2491               (method url-fetch)
2492               (uri
2493                (string-append "mirror://sourceforge/flexbar/"
2494                               version "/flexbar_v" version "_src.tgz"))
2495               (sha256
2496                (base32
2497                 "13jaykc3y1x8y5nn9j8ljnb79s5y51kyxz46hdmvvjj6qhyympmf"))))
2498     (build-system cmake-build-system)
2499     (arguments
2500      `(#:configure-flags (list
2501                           (string-append "-DFLEXBAR_BINARY_DIR="
2502                                          (assoc-ref %outputs "out")
2503                                          "/bin/"))
2504        #:phases
2505        (modify-phases %standard-phases
2506          (replace 'check
2507            (lambda* (#:key outputs #:allow-other-keys)
2508              (setenv "PATH" (string-append
2509                              (assoc-ref outputs "out") "/bin:"
2510                              (getenv "PATH")))
2511              (chdir "../flexbar_v2.5_src/test")
2512              (zero? (system* "bash" "flexbar_validate.sh"))))
2513          (delete 'install))))
2514     (inputs
2515      `(("tbb" ,tbb)
2516        ("zlib" ,zlib)))
2517     (native-inputs
2518      `(("pkg-config" ,pkg-config)
2519        ("seqan" ,seqan)))
2520     (home-page "http://flexbar.sourceforge.net")
2521     (synopsis "Barcode and adapter removal tool for sequencing platforms")
2522     (description
2523      "Flexbar preprocesses high-throughput nucleotide sequencing data
2524 efficiently.  It demultiplexes barcoded runs and removes adapter sequences.
2525 Moreover, trimming and filtering features are provided.  Flexbar increases
2526 read mapping rates and improves genome and transcriptome assemblies.  It
2527 supports next-generation sequencing data in fasta/q and csfasta/q format from
2528 Illumina, Roche 454, and the SOLiD platform.")
2529     (license license:gpl3)))
2530
2531 (define-public fraggenescan
2532   (package
2533     (name "fraggenescan")
2534     (version "1.30")
2535     (source
2536      (origin
2537        (method url-fetch)
2538        (uri
2539         (string-append "mirror://sourceforge/fraggenescan/"
2540                        "FragGeneScan" version ".tar.gz"))
2541        (sha256
2542         (base32 "158dcnwczgcyhwm4qlx19sanrwgdpzf6bn2y57mbpx55lkgz1mzj"))))
2543     (build-system gnu-build-system)
2544     (arguments
2545      `(#:phases
2546        (modify-phases %standard-phases
2547          (delete 'configure)
2548          (add-before 'build 'patch-paths
2549            (lambda* (#:key outputs #:allow-other-keys)
2550              (let* ((out (string-append (assoc-ref outputs "out")))
2551                     (share (string-append out "/share/fraggenescan/")))
2552                (substitute* "run_FragGeneScan.pl"
2553                  (("system\\(\"rm")
2554                   (string-append "system(\"" (which "rm")))
2555                  (("system\\(\"mv")
2556                   (string-append "system(\"" (which "mv")))
2557                  (("\\\"awk") (string-append "\"" (which "awk")))
2558                  ;; This script and other programs expect the training files
2559                  ;; to be in the non-standard location bin/train/XXX. Change
2560                  ;; this to be share/fraggenescan/train/XXX instead.
2561                  (("^\\$train.file = \\$dir.*")
2562                   (string-append "$train_file = \""
2563                                  share
2564                                  "train/\".$FGS_train_file;")))
2565                (substitute* "run_hmm.c"
2566                  (("^  strcat\\(train_dir, \\\"train/\\\"\\);")
2567                   (string-append "  strcpy(train_dir, \"" share "/train/\");"))))
2568              #t))
2569          (replace 'build
2570            (lambda _ (and (zero? (system* "make" "clean"))
2571                           (zero? (system* "make" "fgs")))))
2572          (replace 'install
2573            (lambda* (#:key outputs #:allow-other-keys)
2574              (let* ((out (string-append (assoc-ref outputs "out")))
2575                     (bin (string-append out "/bin/"))
2576                     (share (string-append out "/share/fraggenescan/train")))
2577                (install-file "run_FragGeneScan.pl" bin)
2578                (install-file "FragGeneScan" bin)
2579                (copy-recursively "train" share))))
2580          (delete 'check)
2581          (add-after 'install 'post-install-check
2582            ;; In lieu of 'make check', run one of the examples and check the
2583            ;; output files gets created.
2584            (lambda* (#:key outputs #:allow-other-keys)
2585              (let* ((out (string-append (assoc-ref outputs "out")))
2586                     (bin (string-append out "/bin/"))
2587                     (frag (string-append bin "run_FragGeneScan.pl")))
2588                (and (zero? (system* frag ; Test complete genome.
2589                              "-genome=./example/NC_000913.fna"
2590                              "-out=./test2"
2591                              "-complete=1"
2592                              "-train=complete"))
2593                     (file-exists? "test2.faa")
2594                     (file-exists? "test2.ffn")
2595                     (file-exists? "test2.gff")
2596                     (file-exists? "test2.out")
2597                     (zero? (system* ; Test incomplete sequences.
2598                             frag
2599                             "-genome=./example/NC_000913-fgs.ffn"
2600                             "-out=out"
2601                             "-complete=0"
2602                             "-train=454_30")))))))))
2603     (inputs
2604      `(("perl" ,perl)
2605        ("python" ,python-2))) ;not compatible with python 3.
2606     (home-page "https://sourceforge.net/projects/fraggenescan/")
2607     (synopsis "Finds potentially fragmented genes in short reads")
2608     (description
2609      "FragGeneScan is a program for predicting bacterial and archaeal genes in
2610 short and error-prone DNA sequencing reads.  It can also be applied to predict
2611 genes in incomplete assemblies or complete genomes.")
2612     ;; GPL3+ according to private correspondense with the authors.
2613     (license license:gpl3+)))
2614
2615 (define-public fxtract
2616   (let ((util-commit "776ca85a18a47492af3794745efcb4a905113115"))
2617     (package
2618       (name "fxtract")
2619       (version "2.3")
2620       (source
2621        (origin
2622          (method url-fetch)
2623          (uri (string-append
2624                "https://github.com/ctSkennerton/fxtract/archive/"
2625                version ".tar.gz"))
2626          (file-name (string-append "ctstennerton-util-"
2627                                    (string-take util-commit 7)
2628                                    "-checkout"))
2629          (sha256
2630           (base32
2631            "0275cfdhis8517hm01is62062swmi06fxzifq7mr3knbbxjlaiwj"))))
2632       (build-system gnu-build-system)
2633       (arguments
2634        `(#:make-flags (list
2635                        (string-append "PREFIX=" (assoc-ref %outputs "out"))
2636                        "CC=gcc")
2637          #:test-target "fxtract_test"
2638          #:phases
2639          (modify-phases %standard-phases
2640            (delete 'configure)
2641            (add-before 'build 'copy-util
2642              (lambda* (#:key inputs #:allow-other-keys)
2643                (rmdir "util")
2644                (copy-recursively (assoc-ref inputs "ctskennerton-util") "util")
2645                #t))
2646            ;; Do not use make install as this requires additional dependencies.
2647            (replace 'install
2648              (lambda* (#:key outputs #:allow-other-keys)
2649                (let* ((out (assoc-ref outputs "out"))
2650                       (bin (string-append out"/bin")))
2651                  (install-file "fxtract" bin)
2652                  #t))))))
2653       (inputs
2654        `(("pcre" ,pcre)
2655          ("zlib" ,zlib)))
2656       (native-inputs
2657        ;; ctskennerton-util is licensed under GPL2.
2658        `(("ctskennerton-util"
2659           ,(origin
2660              (method git-fetch)
2661              (uri (git-reference
2662                    (url "https://github.com/ctSkennerton/util.git")
2663                    (commit util-commit)))
2664              (file-name (string-append
2665                          "ctstennerton-util-" util-commit "-checkout"))
2666              (sha256
2667               (base32
2668                "0cls1hd4vgj3f36fpzzg4xc77d6f3hpc60cbpfmn2gdr7ykzzad7"))))))
2669       (home-page "https://github.com/ctSkennerton/fxtract")
2670       (synopsis "Extract sequences from FASTA and FASTQ files")
2671       (description
2672        "Fxtract extracts sequences from a protein or nucleotide fastx (FASTA
2673 or FASTQ) file given a subsequence.  It uses a simple substring search for
2674 basic tasks but can change to using POSIX regular expressions, PCRE, hash
2675 lookups or multi-pattern searching as required.  By default fxtract looks in
2676 the sequence of each record but can also be told to look in the header,
2677 comment or quality sections.")
2678       ;; 'util' requires SSE instructions.
2679       (supported-systems '("x86_64-linux"))
2680       (license license:expat))))
2681
2682 (define-public gemma
2683   (package
2684     (name "gemma")
2685     (version "0.96")
2686     (source (origin
2687               (method url-fetch)
2688               (uri (string-append "https://github.com/xiangzhou/GEMMA/archive/v"
2689                                   version ".tar.gz"))
2690               (file-name (string-append name "-" version ".tar.gz"))
2691               (sha256
2692                (base32
2693                 "055ynn16gd12pf78n4vr2a9jlwsbwzajpdnf2y2yilg1krfff222"))
2694               (patches (search-patches "gemma-intel-compat.patch"))))
2695     (inputs
2696      `(("gsl" ,gsl)
2697        ("lapack" ,lapack)
2698        ("zlib" ,zlib)))
2699     (build-system gnu-build-system)
2700     (arguments
2701      `(#:make-flags
2702        '(,@(match (%current-system)
2703          ("x86_64-linux"
2704           '("FORCE_DYNAMIC=1"))
2705          ("i686-linux"
2706           '("FORCE_DYNAMIC=1" "FORCE_32BIT=1"))
2707          (_
2708           '("FORCE_DYNAMIC=1" "NO_INTEL_COMPAT=1"))))
2709        #:phases
2710        (modify-phases %standard-phases
2711          (delete 'configure)
2712          (add-before 'build 'bin-mkdir
2713           (lambda _
2714             (mkdir-p "bin")
2715             #t))
2716          (replace 'install
2717            (lambda* (#:key outputs #:allow-other-keys)
2718              (let ((out (assoc-ref outputs "out")))
2719                (install-file "bin/gemma"
2720                              (string-append
2721                               out "/bin")))
2722              #t)))
2723        #:tests? #f)) ; no tests included yet
2724     (home-page "https://github.com/xiangzhou/GEMMA")
2725     (synopsis "Tool for genome-wide efficient mixed model association")
2726     (description
2727      "Genome-wide Efficient Mixed Model Association (GEMMA) provides a
2728 standard linear mixed model resolver with application in genome-wide
2729 association studies (GWAS).")
2730     (license license:gpl3)))
2731
2732 (define-public grit
2733   (package
2734     (name "grit")
2735     (version "2.0.2")
2736     (source (origin
2737               (method url-fetch)
2738               (uri (string-append
2739                     "https://github.com/nboley/grit/archive/"
2740                     version ".tar.gz"))
2741               (file-name (string-append name "-" version ".tar.gz"))
2742               (sha256
2743                (base32
2744                 "157in84dj70wimbind3x7sy1whs3h57qfgcnj2s6lrd38fbrb7mj"))))
2745     (build-system python-build-system)
2746     (arguments
2747      `(#:python ,python-2
2748        #:phases
2749        (modify-phases %standard-phases
2750          (add-after 'unpack 'generate-from-cython-sources
2751            (lambda* (#:key inputs outputs #:allow-other-keys)
2752              ;; Delete these C files to force fresh generation from pyx sources.
2753              (delete-file "grit/sparsify_support_fns.c")
2754              (delete-file "grit/call_peaks_support_fns.c")
2755              (substitute* "setup.py"
2756                (("Cython.Setup") "Cython.Build")
2757                ;; Add numpy include path to fix compilation
2758                (("pyx\", \\]")
2759                 (string-append "pyx\", ], include_dirs = ['"
2760                                (assoc-ref inputs "python-numpy")
2761                                "/lib/python2.7/site-packages/numpy/core/include/"
2762                                "']")))
2763              #t)))))
2764     (inputs
2765      `(("python-scipy" ,python2-scipy)
2766        ("python-numpy" ,python2-numpy)
2767        ("python-pysam" ,python2-pysam)
2768        ("python-networkx" ,python2-networkx)))
2769     (native-inputs
2770      `(("python-cython" ,python2-cython)))
2771     (home-page "http://grit-bio.org")
2772     (synopsis "Tool for integrative analysis of RNA-seq type assays")
2773     (description
2774      "GRIT is designed to use RNA-seq, TES, and TSS data to build and quantify
2775 full length transcript models.  When none of these data sources are available,
2776 GRIT can be run by providing a candidate set of TES or TSS sites.  In
2777 addition, GRIT can merge in reference junctions and gene boundaries.  GRIT can
2778 also be run in quantification mode, where it uses a provided GTF file and just
2779 estimates transcript expression.")
2780     (license license:gpl3+)))
2781
2782 (define-public hisat
2783   (package
2784     (name "hisat")
2785     (version "0.1.4")
2786     (source (origin
2787               (method url-fetch)
2788               (uri (string-append
2789                     "http://ccb.jhu.edu/software/hisat/downloads/hisat-"
2790                     version "-beta-source.zip"))
2791               (sha256
2792                (base32
2793                 "1k381ydranqxp09yf2y7w1d0chz5d59vb6jchi89hbb0prq19lk5"))))
2794     (build-system gnu-build-system)
2795     (arguments
2796      `(#:tests? #f ;no check target
2797        #:make-flags '("allall"
2798                       ;; Disable unsupported `popcnt' instructions on
2799                       ;; architectures other than x86_64
2800                       ,@(if (string-prefix? "x86_64"
2801                                             (or (%current-target-system)
2802                                                 (%current-system)))
2803                             '()
2804                             '("POPCNT_CAPABILITY=0")))
2805        #:phases
2806        (modify-phases %standard-phases
2807          (add-after 'unpack 'patch-sources
2808            (lambda _
2809              ;; XXX Cannot use snippet because zip files are not supported
2810              (substitute* "Makefile"
2811                (("^CC = .*$") "CC = gcc")
2812                (("^CPP = .*$") "CPP = g++")
2813                ;; replace BUILD_HOST and BUILD_TIME for deterministic build
2814                (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
2815                (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\""))
2816              (substitute* '("hisat-build" "hisat-inspect")
2817                (("/usr/bin/env") (which "env")))
2818              #t))
2819          (replace 'install
2820            (lambda* (#:key outputs #:allow-other-keys)
2821              (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
2822                (for-each (lambda (file)
2823                            (install-file file bin))
2824                          (find-files
2825                           "."
2826                           "hisat(-(build|align|inspect)(-(s|l)(-debug)*)*)*$")))
2827              #t))
2828          (delete 'configure))))
2829     (native-inputs
2830      `(("unzip" ,unzip)))
2831     (inputs
2832      `(("perl" ,perl)
2833        ("python" ,python)
2834        ("zlib" ,zlib)))
2835     ;; Non-portable SSE instructions are used so building fails on platforms
2836     ;; other than x86_64.
2837     (supported-systems '("x86_64-linux"))
2838     (home-page "http://ccb.jhu.edu/software/hisat/index.shtml")
2839     (synopsis "Hierarchical indexing for spliced alignment of transcripts")
2840     (description
2841      "HISAT is a fast and sensitive spliced alignment program for mapping
2842 RNA-seq reads.  In addition to one global FM index that represents a whole
2843 genome, HISAT uses a large set of small FM indexes that collectively cover the
2844 whole genome.  These small indexes (called local indexes) combined with
2845 several alignment strategies enable effective alignment of RNA-seq reads, in
2846 particular, reads spanning multiple exons.")
2847     (license license:gpl3+)))
2848
2849 (define-public hisat2
2850   (package
2851     (name "hisat2")
2852     (version "2.0.5")
2853     (source
2854      (origin
2855        (method url-fetch)
2856        ;; FIXME: a better source URL is
2857        ;; (string-append "ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2"
2858        ;;                "/downloads/hisat2-" version "-source.zip")
2859        ;; with hash "0lywnr8kijwsc2aw10dwxic0n0yvip6fl3rjlvc8zzwahamy4x7g"
2860        ;; but it is currently unavailable.
2861        (uri "https://github.com/infphilo/hisat2/archive/cba6e8cb.tar.gz")
2862        (file-name (string-append name "-" version ".tar.gz"))
2863        (sha256
2864         (base32
2865          "1mf2hdsyv7cd97xm9mp9a4qws02yrj95y6w6f6cdwnq0klp81r50"))))
2866     (build-system gnu-build-system)
2867     (arguments
2868      `(#:tests? #f                      ; no check target
2869        #:make-flags (list "CC=gcc" "CXX=g++" "allall")
2870        #:modules ((guix build gnu-build-system)
2871                   (guix build utils)
2872                   (srfi srfi-26))
2873        #:phases
2874        (modify-phases %standard-phases
2875          (add-after 'unpack 'make-deterministic
2876            (lambda _
2877              (substitute* "Makefile"
2878                (("`date`") "0"))
2879              #t))
2880          (delete 'configure)
2881          (replace 'install
2882            (lambda* (#:key outputs #:allow-other-keys)
2883              (let* ((out (assoc-ref outputs "out"))
2884                     (bin (string-append out "/bin/"))
2885                     (doc (string-append out "/share/doc/hisat2/")))
2886                (for-each
2887                 (cut install-file <> bin)
2888                 (find-files "."
2889                             "hisat2(-(build|align|inspect)(-(s|l)(-debug)*)*)*$"))
2890                (mkdir-p doc)
2891                (install-file "doc/manual.inc.html" doc))
2892              #t)))))
2893     (native-inputs
2894      `(("unzip" ,unzip)                 ; needed for archive from ftp
2895        ("perl" ,perl)
2896        ("pandoc" ,ghc-pandoc)))         ; for documentation
2897     (home-page "http://ccb.jhu.edu/software/hisat2/index.shtml")
2898     (synopsis "Graph-based alignment of genomic sequencing reads")
2899     (description "HISAT2 is a fast and sensitive alignment program for mapping
2900 next-generation sequencing reads (both DNA and RNA) to a population of human
2901 genomes (as well as to a single reference genome).  In addition to using one
2902 global @dfn{graph FM} (GFM) index that represents a population of human
2903 genomes, HISAT2 uses a large set of small GFM indexes that collectively cover
2904 the whole genome.  These small indexes, combined with several alignment
2905 strategies, enable rapid and accurate alignment of sequencing reads.  This new
2906 indexing scheme is called a @dfn{Hierarchical Graph FM index} (HGFM).")
2907     ;; HISAT2 contains files from Bowtie2, which is released under
2908     ;; GPLv2 or later.  The HISAT2 source files are released under
2909     ;; GPLv3 or later.
2910     (license license:gpl3+)))
2911
2912 (define-public hmmer
2913   (package
2914     (name "hmmer")
2915     (version "3.1b2")
2916     (source
2917      (origin
2918        (method url-fetch)
2919        (uri (string-append
2920              "http://eddylab.org/software/hmmer"
2921              (version-major version) "/"
2922              version "/hmmer-" version ".tar.gz"))
2923        (sha256
2924         (base32
2925          "0djmgc0pfli0jilfx8hql1axhwhqxqb8rxg2r5rg07aw73sfs5nx"))
2926        (patches (search-patches "hmmer-remove-cpu-specificity.patch"))))
2927     (build-system gnu-build-system)
2928     (native-inputs `(("perl" ,perl)))
2929     (home-page "http://hmmer.org/")
2930     (synopsis "Biosequence analysis using profile hidden Markov models")
2931     (description
2932      "HMMER is used for searching sequence databases for homologs of protein
2933 sequences, and for making protein sequence alignments.  It implements methods
2934 using probabilistic models called profile hidden Markov models (profile
2935 HMMs).")
2936     (license (list license:gpl3+
2937                    ;; The bundled library 'easel' is distributed
2938                    ;; under The Janelia Farm Software License.
2939                    (license:non-copyleft
2940                     "file://easel/LICENSE"
2941                     "See easel/LICENSE in the distribution.")))))
2942
2943 (define-public htseq
2944   (package
2945     (name "htseq")
2946     (version "0.9.1")
2947     (source (origin
2948               (method url-fetch)
2949               (uri (pypi-uri "HTSeq" version))
2950               (sha256
2951                (base32
2952                 "11flgb1381xdhk43bzbfm3vhnszkpqg6jk76rpa5xd1zbrvvlnxg"))))
2953     (build-system python-build-system)
2954     (native-inputs
2955      `(("python-cython" ,python-cython)))
2956     ;; Numpy needs to be propagated when htseq is used as a Python library.
2957     (propagated-inputs
2958      `(("python-numpy" ,python-numpy)))
2959     (inputs
2960      `(("python-pysam" ,python-pysam)
2961        ("python-matplotlib" ,python-matplotlib)))
2962     (home-page "http://www-huber.embl.de/users/anders/HTSeq/")
2963     (synopsis "Analysing high-throughput sequencing data with Python")
2964     (description
2965      "HTSeq is a Python package that provides infrastructure to process data
2966 from high-throughput sequencing assays.")
2967     (license license:gpl3+)))
2968
2969 (define-public python2-htseq
2970   (package-with-python2 htseq))
2971
2972 (define-public java-htsjdk
2973   (package
2974     (name "java-htsjdk")
2975     (version "2.3.0") ; last version without build dependency on gradle
2976     (source (origin
2977               (method url-fetch)
2978               (uri (string-append
2979                     "https://github.com/samtools/htsjdk/archive/"
2980                     version ".tar.gz"))
2981               (file-name (string-append name "-" version ".tar.gz"))
2982               (sha256
2983                (base32
2984                 "1ibhzzxsfc38nqyk9r8zqj6blfc1kh26iirypd4q6n90hs2m6nyq"))
2985               (modules '((guix build utils)))
2986               (snippet
2987                ;; Delete pre-built binaries
2988                '(begin
2989                   (delete-file-recursively "lib")
2990                   (mkdir-p "lib")
2991                   #t))))
2992     (build-system ant-build-system)
2993     (arguments
2994      `(#:tests? #f ; test require Internet access
2995        #:jdk ,icedtea-8
2996        #:make-flags
2997        (list (string-append "-Ddist=" (assoc-ref %outputs "out")
2998                             "/share/java/htsjdk/"))
2999        #:build-target "all"
3000        #:phases
3001        (modify-phases %standard-phases
3002          ;; The build phase also installs the jars
3003          (delete 'install))))
3004     (inputs
3005      `(("java-ngs" ,java-ngs)
3006        ("java-snappy-1" ,java-snappy-1)
3007        ("java-commons-compress" ,java-commons-compress)
3008        ("java-commons-logging-minimal" ,java-commons-logging-minimal)
3009        ("java-commons-jexl-2" ,java-commons-jexl-2)
3010        ("java-xz" ,java-xz)))
3011     (native-inputs
3012      `(("java-testng" ,java-testng)))
3013     (home-page "http://samtools.github.io/htsjdk/")
3014     (synopsis "Java API for high-throughput sequencing data (HTS) formats")
3015     (description
3016      "HTSJDK is an implementation of a unified Java library for accessing
3017 common file formats, such as SAM and VCF, used for high-throughput
3018 sequencing (HTS) data.  There are also an number of useful utilities for
3019 manipulating HTS data.")
3020     (license license:expat)))
3021
3022 ;; This version matches java-htsjdk 2.3.0.  Later versions also require a more
3023 ;; recent version of java-htsjdk, which depends on gradle.
3024 (define-public java-picard
3025   (package
3026     (name "java-picard")
3027     (version "2.3.0")
3028     (source (origin
3029               (method git-fetch)
3030               (uri (git-reference
3031                     (url "https://github.com/broadinstitute/picard.git")
3032                     (commit version)))
3033               (file-name (string-append "java-picard-" version "-checkout"))
3034               (sha256
3035                (base32
3036                 "1ll7mf4r3by92w2nhlmpa591xd1f46xlkwh59mq6fvbb5pdwzvx6"))
3037               (modules '((guix build utils)))
3038               (snippet
3039                '(begin
3040                   ;; Delete pre-built binaries.
3041                   (delete-file-recursively "lib")
3042                   (mkdir-p "lib")
3043                   (substitute* "build.xml"
3044                     ;; Remove build-time dependency on git.
3045                     (("failifexecutionfails=\"true\"")
3046                      "failifexecutionfails=\"false\"")
3047                     ;; Use our htsjdk.
3048                     (("depends=\"compile-htsjdk, ")
3049                      "depends=\"")
3050                     (("depends=\"compile-htsjdk-tests, ")
3051                      "depends=\"")
3052                     ;; Build picard-lib.jar before building picard.jar
3053                     (("name=\"picard-jar\" depends=\"" line)
3054                      (string-append line "picard-lib-jar, ")))
3055                   #t))))
3056     (build-system ant-build-system)
3057     (arguments
3058      `(#:build-target "picard-jar"
3059        #:test-target "test"
3060        ;; Tests require jacoco:coverage.
3061        #:tests? #f
3062        #:make-flags
3063        (list (string-append "-Dhtsjdk_lib_dir="
3064                             (assoc-ref %build-inputs "java-htsjdk")
3065                             "/share/java/htsjdk/")
3066              "-Dhtsjdk-classes=dist/tmp"
3067              (string-append "-Dhtsjdk-version="
3068                             ,(package-version java-htsjdk)))
3069        #:jdk ,icedtea-8
3070        #:phases
3071        (modify-phases %standard-phases
3072          (add-after 'unpack 'use-our-htsjdk
3073            (lambda* (#:key inputs #:allow-other-keys)
3074              (substitute* "build.xml"
3075                (("\\$\\{htsjdk\\}/lib")
3076                 (string-append (assoc-ref inputs "java-htsjdk")
3077                                "/share/java/htsjdk/")))
3078              #t))
3079          (add-after 'unpack 'make-test-target-independent
3080            (lambda* (#:key inputs #:allow-other-keys)
3081              (substitute* "build.xml"
3082                (("name=\"test\" depends=\"compile, ")
3083                 "name=\"test\" depends=\""))
3084              #t))
3085          (replace 'install (install-jars "dist")))))
3086     (inputs
3087      `(("java-htsjdk" ,java-htsjdk)
3088        ("java-guava" ,java-guava)))
3089     (native-inputs
3090      `(("java-testng" ,java-testng)))
3091     (home-page "http://broadinstitute.github.io/picard/")
3092     (synopsis "Tools for manipulating high-throughput sequencing data and formats")
3093     (description "Picard is a set of Java command line tools for manipulating
3094 high-throughput sequencing (HTS) data and formats.  Picard is implemented
3095 using the HTSJDK Java library to support accessing file formats that are
3096 commonly used for high-throughput sequencing data such as SAM, BAM, CRAM and
3097 VCF.")
3098     (license license:expat)))
3099
3100 ;; This is the last version of Picard to provide net.sf.samtools
3101 (define-public java-picard-1.113
3102   (package (inherit java-picard)
3103     (name "java-picard")
3104     (version "1.113")
3105     (source (origin
3106               (method git-fetch)
3107               (uri (git-reference
3108                     (url "https://github.com/broadinstitute/picard.git")
3109                     (commit version)))
3110               (file-name (string-append "java-picard-" version "-checkout"))
3111               (sha256
3112                (base32
3113                 "0lkpvin2fz3hhly4l02kk56fqy8lmlgyzr9kmvljk6ry6l1hw973"))
3114               (modules '((guix build utils)))
3115               (snippet
3116                '(begin
3117                   ;; Delete pre-built binaries.
3118                   (delete-file-recursively "lib")
3119                   (mkdir-p "lib")
3120                   #t))))
3121     (build-system ant-build-system)
3122     (arguments
3123      `(#:build-target "picard-jar"
3124        #:test-target "test"
3125        ;; FIXME: the class path at test time is wrong.
3126        ;; [testng] Error: A JNI error has occurred, please check your installation and try again
3127        ;; [testng] Exception in thread "main" java.lang.NoClassDefFoundError: com/beust/jcommander/ParameterException
3128        #:tests? #f
3129        #:jdk ,icedtea-8
3130        ;; This is only used for tests.
3131        #:make-flags
3132        (list "-Dsamjdk.intel_deflater_so_path=lib/jni/libIntelDeflater.so")
3133        #:phases
3134        (modify-phases %standard-phases
3135          ;; Do not use bundled ant bzip2.
3136          (add-after 'unpack 'use-ant-bzip
3137            (lambda* (#:key inputs #:allow-other-keys)
3138              (substitute* "build.xml"
3139                (("\\$\\{lib\\}/apache-ant-1.8.2-bzip2.jar")
3140                 (string-append (assoc-ref inputs "ant")
3141                                "/lib/ant.jar")))
3142              #t))
3143          (add-after 'unpack 'make-test-target-independent
3144            (lambda* (#:key inputs #:allow-other-keys)
3145              (substitute* "build.xml"
3146                (("name=\"test\" depends=\"compile, ")
3147                 "name=\"test\" depends=\"compile-tests, ")
3148                (("name=\"compile\" depends=\"compile-src, compile-tests\"")
3149                 "name=\"compile\" depends=\"compile-src\""))
3150              #t))
3151          (add-after 'unpack 'fix-deflater-path
3152            (lambda* (#:key outputs #:allow-other-keys)
3153              (substitute* "src/java/net/sf/samtools/Defaults.java"
3154                (("getStringProperty\\(\"intel_deflater_so_path\", null\\)")
3155                 (string-append "getStringProperty(\"intel_deflater_so_path\", \""
3156                                (assoc-ref outputs "out")
3157                                "/lib/jni/libIntelDeflater.so"
3158                                "\")")))
3159              #t))
3160          ;; Build the deflater library, because we've previously deleted the
3161          ;; pre-built one.  This can only be built with access to the JDK
3162          ;; sources.
3163          (add-after 'build 'build-jni
3164            (lambda* (#:key inputs #:allow-other-keys)
3165              (mkdir-p "lib/jni")
3166              (mkdir-p "jdk-src")
3167              (and (zero? (system* "tar" "--strip-components=1" "-C" "jdk-src"
3168                                   "-xf" (assoc-ref inputs "jdk-src")))
3169                   (zero? (system* "javah" "-jni"
3170                                   "-classpath" "classes"
3171                                   "-d" "lib/"
3172                                   "net.sf.samtools.util.zip.IntelDeflater"))
3173                   (with-directory-excursion "src/c/inteldeflater"
3174                     (zero? (system* "gcc" "-I../../../lib" "-I."
3175                                     (string-append "-I" (assoc-ref inputs "jdk")
3176                                                    "/include/linux")
3177                                     "-I../../../jdk-src/src/share/native/common/"
3178                                     "-I../../../jdk-src/src/solaris/native/common/"
3179                                     "-c" "-O3" "-fPIC" "IntelDeflater.c"))
3180                     (zero? (system* "gcc" "-shared"
3181                                     "-o" "../../../lib/jni/libIntelDeflater.so"
3182                                     "IntelDeflater.o" "-lz" "-lstdc++"))))))
3183          ;; We can only build everything else after building the JNI library.
3184          (add-after 'build-jni 'build-rest
3185            (lambda* (#:key make-flags #:allow-other-keys)
3186              (zero? (apply system* `("ant" "all" ,@make-flags)))))
3187          (add-before 'build 'set-JAVA6_HOME
3188            (lambda _
3189              (setenv "JAVA6_HOME" (getenv "JAVA_HOME"))
3190              #t))
3191          (replace 'install (install-jars "dist"))
3192          (add-after 'install 'install-jni-lib
3193            (lambda* (#:key outputs #:allow-other-keys)
3194              (let ((jni (string-append (assoc-ref outputs "out")
3195                                        "/lib/jni")))
3196                (mkdir-p jni)
3197                (install-file "lib/jni/libIntelDeflater.so" jni)
3198                #t))))))
3199     (inputs
3200      `(("java-snappy-1" ,java-snappy-1)
3201        ("java-commons-jexl-2" ,java-commons-jexl-2)
3202        ("java-cofoja" ,java-cofoja)
3203        ("ant" ,ant) ; for bzip2 support at runtime
3204        ("zlib" ,zlib)))
3205     (native-inputs
3206      `(("ant-apache-bcel" ,ant-apache-bcel)
3207        ("ant-junit" ,ant-junit)
3208        ("java-testng" ,java-testng)
3209        ("java-commons-bcel" ,java-commons-bcel)
3210        ("java-jcommander" ,java-jcommander)
3211        ("jdk" ,icedtea-8 "jdk")
3212        ("jdk-src" ,(car (assoc-ref (package-native-inputs icedtea-8) "jdk-drop")))))))
3213
3214 (define-public htslib
3215   (package
3216     (name "htslib")
3217     (version "1.6")
3218     (source (origin
3219               (method url-fetch)
3220               (uri (string-append
3221                     "https://github.com/samtools/htslib/releases/download/"
3222                     version "/htslib-" version ".tar.bz2"))
3223               (sha256
3224                (base32
3225                 "1jsca3hg4rbr6iqq6imkj4lsvgl8g9768bcmny3hlff2w25vx24m"))))
3226     (build-system gnu-build-system)
3227     (arguments
3228      `(#:phases
3229        (modify-phases %standard-phases
3230          (add-after
3231           'unpack 'patch-tests
3232           (lambda _
3233             (substitute* "test/test.pl"
3234               (("/bin/bash") (which "bash")))
3235             #t)))))
3236     (inputs
3237      `(("openssl" ,openssl)
3238        ("curl" ,curl)
3239        ("zlib" ,zlib)))
3240     (native-inputs
3241      `(("perl" ,perl)))
3242     (home-page "http://www.htslib.org")
3243     (synopsis "C library for reading/writing high-throughput sequencing data")
3244     (description
3245      "HTSlib is a C library for reading/writing high-throughput sequencing
3246 data.  It also provides the @command{bgzip}, @command{htsfile}, and
3247 @command{tabix} utilities.")
3248     ;; Files under cram/ are released under the modified BSD license;
3249     ;; the rest is released under the Expat license
3250     (license (list license:expat license:bsd-3))))
3251
3252 ;; This package should be removed once no packages rely upon it.
3253 (define htslib-1.3
3254   (package
3255     (inherit htslib)
3256     (version "1.3.1")
3257     (source (origin
3258               (method url-fetch)
3259               (uri (string-append
3260                     "https://github.com/samtools/htslib/releases/download/"
3261                     version "/htslib-" version ".tar.bz2"))
3262               (sha256
3263                (base32
3264                 "1rja282fwdc25ql6izkhdyh8ppw8x2fs0w0js78zgkmqjlikmma9"))))))
3265
3266 (define-public idr
3267   (package
3268     (name "idr")
3269     (version "2.0.3")
3270     (source (origin
3271               (method url-fetch)
3272               (uri (string-append
3273                     "https://github.com/nboley/idr/archive/"
3274                     version ".tar.gz"))
3275               (file-name (string-append name "-" version ".tar.gz"))
3276               (sha256
3277                (base32
3278                 "1rjdly6daslw66r43g9md8znizlscn1sphycqyldzsidkc4vxqv3"))
3279               ;; Delete generated C code.
3280               (snippet
3281                '(begin (delete-file "idr/inv_cdf.c") #t))))
3282     (build-system python-build-system)
3283     ;; There is only one test ("test_inv_cdf.py") and it tests features that
3284     ;; are no longer part of this package.  It also asserts False, which
3285     ;; causes the tests to always fail.
3286     (arguments `(#:tests? #f))
3287     (propagated-inputs
3288      `(("python-scipy" ,python-scipy)
3289        ("python-sympy" ,python-sympy)
3290        ("python-numpy" ,python-numpy)
3291        ("python-matplotlib" ,python-matplotlib)))
3292     (native-inputs
3293      `(("python-cython" ,python-cython)))
3294     (home-page "https://github.com/nboley/idr")
3295     (synopsis "Tool to measure the irreproducible discovery rate (IDR)")
3296     (description
3297      "The IDR (Irreproducible Discovery Rate) framework is a unified approach
3298 to measure the reproducibility of findings identified from replicate
3299 experiments and provide highly stable thresholds based on reproducibility.")
3300     (license license:gpl2+)))
3301
3302 (define-public jellyfish
3303   (package
3304     (name "jellyfish")
3305     (version "2.2.7")
3306     (source (origin
3307               (method url-fetch)
3308               (uri (string-append "https://github.com/gmarcais/Jellyfish/"
3309                                   "releases/download/v" version
3310                                   "/jellyfish-" version ".tar.gz"))
3311               (sha256
3312                (base32
3313                 "1a1iwq9pq54k2m9ypvwl5s0bqfl64gwh9dx5af9i382ajas2016q"))))
3314     (build-system gnu-build-system)
3315     (outputs '("out"      ;for library
3316                "ruby"     ;for Ruby bindings
3317                "python")) ;for Python bindings
3318     (arguments
3319      `(#:configure-flags
3320        (list (string-append "--enable-ruby-binding="
3321                             (assoc-ref %outputs "ruby"))
3322              (string-append "--enable-python-binding="
3323                             (assoc-ref %outputs "python")))
3324        #:phases
3325        (modify-phases %standard-phases
3326          (add-before 'check 'set-SHELL-variable
3327            (lambda _
3328              ;; generator_manager.hpp either uses /bin/sh or $SHELL
3329              ;; to run tests.
3330              (setenv "SHELL" (which "bash"))
3331              #t)))))
3332     (native-inputs
3333      `(("bc" ,bc)
3334        ("time" ,time)
3335        ("ruby" ,ruby)
3336        ("python" ,python-2)
3337        ("pkg-config" ,pkg-config)))
3338     (inputs
3339      `(("htslib" ,htslib)))
3340     (synopsis "Tool for fast counting of k-mers in DNA")
3341     (description
3342      "Jellyfish is a tool for fast, memory-efficient counting of k-mers in
3343 DNA.  A k-mer is a substring of length k, and counting the occurrences of all
3344 such substrings is a central step in many analyses of DNA sequence.  Jellyfish
3345 is a command-line program that reads FASTA and multi-FASTA files containing
3346 DNA sequences.  It outputs its k-mer counts in a binary format, which can be
3347 translated into a human-readable text format using the @code{jellyfish dump}
3348 command, or queried for specific k-mers with @code{jellyfish query}.")
3349     (home-page "http://www.genome.umd.edu/jellyfish.html")
3350     ;; From their website: JELLYFISH runs on 64-bit Intel-compatible processors
3351     (supported-systems '("x86_64-linux"))
3352     ;; The combined work is published under the GPLv3 or later.  Individual
3353     ;; files such as lib/jsoncpp.cpp are released under the Expat license.
3354     (license (list license:gpl3+ license:expat))))
3355
3356 (define-public khmer
3357   (package
3358     (name "khmer")
3359     (version "2.0")
3360     (source
3361      (origin
3362        (method url-fetch)
3363        (uri (pypi-uri "khmer" version))
3364        (sha256
3365         (base32
3366          "0wb05shqh77v00256qlm68vbbx3kl76fyzihszbz5nhanl4ni33a"))
3367        (patches (search-patches "khmer-use-libraries.patch"))))
3368     (build-system python-build-system)
3369     (arguments
3370      `(#:phases
3371        (modify-phases %standard-phases
3372          (add-after 'unpack 'set-paths
3373            (lambda* (#:key inputs outputs #:allow-other-keys)
3374              ;; Delete bundled libraries.
3375              (delete-file-recursively "third-party/zlib")
3376              (delete-file-recursively "third-party/bzip2")
3377              ;; Replace bundled seqan.
3378              (let* ((seqan-all "third-party/seqan")
3379                     (seqan-include (string-append
3380                                     seqan-all "/core/include")))
3381                (delete-file-recursively seqan-all)
3382                (copy-recursively (string-append (assoc-ref inputs "seqan")
3383                                                 "/include/seqan")
3384                           (string-append seqan-include "/seqan")))
3385              ;; We do not replace the bundled MurmurHash as the canonical
3386              ;; repository for this code 'SMHasher' is unsuitable for
3387              ;; providing a library.  See
3388              ;; https://lists.gnu.org/archive/html/guix-devel/2016-06/msg00977.html
3389              #t))
3390          (add-after 'unpack 'set-cc
3391            (lambda _
3392              (setenv "CC" "gcc")
3393              #t))
3394          ;; It is simpler to test after installation.
3395          (delete 'check)
3396          (add-after 'install 'post-install-check
3397            (lambda* (#:key inputs outputs #:allow-other-keys)
3398              (let ((out (assoc-ref outputs "out")))
3399                (setenv "PATH"
3400                        (string-append
3401                         (getenv "PATH")
3402                         ":"
3403                         (assoc-ref outputs "out")
3404                         "/bin"))
3405                (setenv "PYTHONPATH"
3406                        (string-append
3407                         (getenv "PYTHONPATH")
3408                         ":"
3409                         out
3410                         "/lib/python"
3411                         (string-take (string-take-right
3412                                       (assoc-ref inputs "python") 5) 3)
3413                         "/site-packages"))
3414                (with-directory-excursion "build"
3415                  (zero? (system* "nosetests" "khmer" "--attr"
3416                                  "!known_failing")))))))))
3417     (native-inputs
3418      `(("seqan" ,seqan)
3419        ("python-nose" ,python-nose)))
3420     (inputs
3421      `(("zlib" ,zlib)
3422        ("bzip2" ,bzip2)
3423        ("python-screed" ,python-screed)
3424        ("python-bz2file" ,python-bz2file)
3425        ;; Tests fail when gcc-5 is used for compilation.  Use gcc-4.9 at least
3426        ;; until the next version of khmer (likely 2.1) is released.
3427        ("gcc" ,gcc-4.9)))
3428     (home-page "https://khmer.readthedocs.org/")
3429     (synopsis "K-mer counting, filtering and graph traversal library")
3430     (description "The khmer software is a set of command-line tools for
3431 working with DNA shotgun sequencing data from genomes, transcriptomes,
3432 metagenomes and single cells.  Khmer can make de novo assemblies faster, and
3433 sometimes better.  Khmer can also identify and fix problems with shotgun
3434 data.")
3435     ;; When building on i686, armhf and mips64el, we get the following error:
3436     ;; error: ['khmer', 'khmer.tests', 'oxli'] require 64-bit operating system
3437     (supported-systems '("x86_64-linux" "aarch64-linux"))
3438     (license license:bsd-3)))
3439
3440 (define-public kaiju
3441   (package
3442     (name "kaiju")
3443     (version "1.5.0")
3444     (source (origin
3445               (method url-fetch)
3446               (uri (string-append
3447                     "https://github.com/bioinformatics-centre/kaiju/archive/v"
3448                     version ".tar.gz"))
3449               (file-name (string-append name "-" version ".tar.gz"))
3450               (sha256
3451                (base32
3452                 "0afbfalfw9y39bkwnqjrh9bghs118ws1pzj5h8l0nblgn3mbjdks"))))
3453     (build-system gnu-build-system)
3454     (arguments
3455      `(#:tests? #f ; There are no tests.
3456        #:phases
3457        (modify-phases %standard-phases
3458          (delete 'configure)
3459          (add-before 'build 'move-to-src-dir
3460            (lambda _ (chdir "src") #t))
3461          (replace 'install
3462            (lambda* (#:key inputs outputs #:allow-other-keys)
3463              (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
3464                (mkdir-p bin)
3465                (chdir "..")
3466                (copy-recursively "bin" bin)
3467                (copy-recursively "util" bin))
3468              #t)))))
3469     (inputs
3470      `(("perl" ,perl)))
3471     (home-page "http://kaiju.binf.ku.dk/")
3472     (synopsis "Fast and sensitive taxonomic classification for metagenomics")
3473     (description "Kaiju is a program for sensitive taxonomic classification
3474 of high-throughput sequencing reads from metagenomic whole genome sequencing
3475 experiments.")
3476     (license license:gpl3+)))
3477
3478 (define-public macs
3479   (package
3480     (name "macs")
3481     (version "2.1.0.20151222")
3482     (source (origin
3483               (method url-fetch)
3484               (uri (pypi-uri "MACS2" version))
3485               (sha256
3486                (base32
3487                 "1r2hcz6irhcq7lwbafjks98jbn34hv05avgbdjnp6w6mlfjkf8x5"))))
3488     (build-system python-build-system)
3489     (arguments
3490      `(#:python ,python-2 ; only compatible with Python 2.7
3491        #:tests? #f)) ; no test target
3492     (inputs
3493      `(("python-numpy" ,python2-numpy)))
3494     (home-page "https://github.com/taoliu/MACS/")
3495     (synopsis "Model based analysis for ChIP-Seq data")
3496     (description
3497      "MACS is an implementation of a ChIP-Seq analysis algorithm for
3498 identifying transcript factor binding sites named Model-based Analysis of
3499 ChIP-Seq (MACS).  MACS captures the influence of genome complexity to evaluate
3500 the significance of enriched ChIP regions and it improves the spatial
3501 resolution of binding sites through combining the information of both
3502 sequencing tag position and orientation.")
3503     (license license:bsd-3)))
3504
3505 (define-public mafft
3506   (package
3507     (name "mafft")
3508     (version "7.313")
3509     (source (origin
3510               (method url-fetch)
3511               (uri (string-append
3512                     "https://mafft.cbrc.jp/alignment/software/mafft-" version
3513                     "-without-extensions-src.tgz"))
3514               (file-name (string-append name "-" version ".tgz"))
3515               (sha256
3516                (base32
3517                 "0r83qmg2if8mi6jyx3xdf8ar2gcxl7r9nmj98jr7lxym97v61a2k"))))
3518     (build-system gnu-build-system)
3519     (arguments
3520      `(#:tests? #f ; no automated tests, though there are tests in the read me
3521        #:make-flags (let ((out (assoc-ref %outputs "out")))
3522                       (list (string-append "PREFIX=" out)
3523                             (string-append "BINDIR="
3524                                            (string-append out "/bin"))))
3525        #:phases
3526        (modify-phases %standard-phases
3527          (add-after 'unpack 'enter-dir
3528            (lambda _ (chdir "core") #t))
3529          (add-after 'enter-dir 'patch-makefile
3530            (lambda _
3531              ;; on advice from the MAFFT authors, there is no need to
3532              ;; distribute mafft-profile, mafft-distance, or
3533              ;; mafft-homologs.rb as they are too "specialised".
3534              (substitute* "Makefile"
3535                ;; remove mafft-homologs.rb from SCRIPTS
3536                (("^SCRIPTS = mafft mafft-homologs.rb")
3537                 "SCRIPTS = mafft")
3538                ;; remove mafft-homologs from MANPAGES
3539                (("^MANPAGES = mafft.1 mafft-homologs.1")
3540                 "MANPAGES = mafft.1")
3541                ;; remove mafft-distance from PROGS
3542                (("^PROGS = dvtditr dndfast7 dndblast sextet5 mafft-distance")
3543                 "PROGS = dvtditr dndfast7 dndblast sextet5")
3544                ;; remove mafft-profile from PROGS
3545                (("splittbfast disttbfast tbfast mafft-profile 2cl mccaskillwrap")
3546                 "splittbfast disttbfast tbfast f2cl mccaskillwrap")
3547                (("^rm -f mafft-profile mafft-profile.exe") "#")
3548                (("^rm -f mafft-distance mafft-distance.exe") ")#")
3549                ;; do not install MAN pages in libexec folder
3550                (("^\t\\$\\(INSTALL\\) -m 644 \\$\\(MANPAGES\\) \
3551 \\$\\(DESTDIR\\)\\$\\(LIBDIR\\)") "#"))
3552              #t))
3553          (add-after 'enter-dir 'patch-paths
3554            (lambda* (#:key inputs #:allow-other-keys)
3555              (substitute* '("pairash.c"
3556                             "mafft.tmpl")
3557                (("perl") (which "perl"))
3558                (("([\"`| ])awk" _ prefix)
3559                 (string-append prefix (which "awk")))
3560                (("grep") (which "grep")))
3561              #t))
3562          (delete 'configure)
3563          (add-after 'install 'wrap-programs
3564            (lambda* (#:key outputs #:allow-other-keys)
3565              (let* ((out (assoc-ref outputs "out"))
3566                     (bin (string-append out "/bin"))
3567                     (path (string-append
3568                            (assoc-ref %build-inputs "coreutils") "/bin:")))
3569                (for-each (lambda (file)
3570                            (wrap-program file
3571                              `("PATH" ":" prefix (,path))))
3572                          (find-files bin)))
3573              #t)))))
3574     (inputs
3575      `(("perl" ,perl)
3576        ("ruby" ,ruby)
3577        ("gawk" ,gawk)
3578        ("grep" ,grep)
3579        ("coreutils" ,coreutils)))
3580     (home-page "http://mafft.cbrc.jp/alignment/software/")
3581     (synopsis "Multiple sequence alignment program")
3582     (description
3583      "MAFFT offers a range of multiple alignment methods for nucleotide and
3584 protein sequences.  For instance, it offers L-INS-i (accurate; for alignment
3585 of <~200 sequences) and FFT-NS-2 (fast; for alignment of <~30,000
3586 sequences).")
3587     (license (license:non-copyleft
3588               "http://mafft.cbrc.jp/alignment/software/license.txt"
3589               "BSD-3 with different formatting"))))
3590
3591 (define-public mash
3592   (package
3593     (name "mash")
3594     (version "1.1.1")
3595     (source (origin
3596               (method url-fetch)
3597               (uri (string-append
3598                     "https://github.com/marbl/mash/archive/v"
3599                     version ".tar.gz"))
3600               (file-name (string-append name "-" version ".tar.gz"))
3601               (sha256
3602                (base32
3603                 "08znbvqq5xknfhmpp3wcj574zvi4p7i8zifi67c9qw9a6ikp42fj"))
3604               (modules '((guix build utils)))
3605               (snippet
3606                ;; Delete bundled kseq.
3607                ;; TODO: Also delete bundled murmurhash and open bloom filter.
3608                '(delete-file "src/mash/kseq.h"))))
3609     (build-system gnu-build-system)
3610     (arguments
3611      `(#:tests? #f ; No tests.
3612        #:configure-flags
3613        (list
3614         (string-append "--with-capnp=" (assoc-ref %build-inputs "capnproto"))
3615         (string-append "--with-gsl=" (assoc-ref %build-inputs "gsl")))
3616        #:make-flags (list "CC=gcc")
3617        #:phases
3618        (modify-phases %standard-phases
3619          (add-after 'unpack 'fix-includes
3620            (lambda _
3621              (substitute* '("src/mash/Sketch.cpp" "src/mash/CommandFind.cpp")
3622                (("^#include \"kseq\\.h\"")
3623                 "#include \"htslib/kseq.h\""))
3624              #t))
3625          (add-after 'fix-includes 'autoconf
3626            (lambda _ (zero? (system* "autoconf")))))))
3627     (native-inputs
3628      `(("autoconf" ,autoconf)
3629        ;; Capnproto and htslib are statically embedded in the final
3630        ;; application. Therefore we also list their licenses, below.
3631        ("capnproto" ,capnproto)
3632        ("htslib" ,htslib)))
3633     (inputs
3634      `(("gsl" ,gsl)
3635        ("zlib" ,zlib)))
3636     (supported-systems '("x86_64-linux"))
3637     (home-page "https://mash.readthedocs.io")
3638     (synopsis "Fast genome and metagenome distance estimation using MinHash")
3639     (description "Mash is a fast sequence distance estimator that uses the
3640 MinHash algorithm and is designed to work with genomes and metagenomes in the
3641 form of assemblies or reads.")
3642     (license (list license:bsd-3          ; Mash
3643                    license:expat          ; HTSlib and capnproto
3644                    license:public-domain  ; MurmurHash 3
3645                    license:cpl1.0))))     ; Open Bloom Filter
3646
3647 (define-public metabat
3648   (package
3649     (name "metabat")
3650     (version "2.12.1")
3651     (source
3652      (origin
3653        (method url-fetch)
3654        (uri (string-append "https://bitbucket.org/berkeleylab/metabat/get/v"
3655                            version ".tar.gz"))
3656        (file-name (string-append name "-" version ".tar.gz"))
3657        (sha256
3658         (base32
3659          "1hmvdalz3zj5sqqklg0l4npjdv37cv2hsdi1al9iby2ndxjs1b73"))
3660        (patches (search-patches "metabat-fix-compilation.patch"))))
3661     (build-system scons-build-system)
3662     (arguments
3663      `(#:scons ,scons-python2
3664        #:scons-flags
3665        (list (string-append "PREFIX=" (assoc-ref %outputs "out"))
3666              (string-append "BOOST_ROOT=" (assoc-ref %build-inputs "boost")))
3667        #:tests? #f ;; Tests are run during the build phase.
3668        #:phases
3669        (modify-phases %standard-phases
3670          (add-after 'unpack 'fix-includes
3671            (lambda _
3672              (substitute* "src/BamUtils.h"
3673                (("^#include \"bam/bam\\.h\"")
3674                 "#include \"samtools/bam.h\"")
3675                (("^#include \"bam/sam\\.h\"")
3676                 "#include \"samtools/sam.h\""))
3677              (substitute* "src/KseqReader.h"
3678                (("^#include \"bam/kseq\\.h\"")
3679                 "#include \"htslib/kseq.h\""))
3680              #t))
3681          (add-after 'unpack 'fix-scons
3682            (lambda* (#:key inputs #:allow-other-keys)
3683              (substitute* "SConstruct"
3684                (("^htslib_dir += 'samtools'")
3685                 (string-append "htslib_dir = '"
3686                                (assoc-ref inputs "htslib")
3687                                "'"))
3688                (("^samtools_dir = 'samtools'")
3689                 (string-append "samtools_dir = '"
3690                                (assoc-ref inputs "samtools")
3691                                "'"))
3692                (("^findStaticOrShared\\('bam', hts_lib")
3693                 (string-append "findStaticOrShared('bam', '"
3694                                (assoc-ref inputs "samtools")
3695                                "/lib'"))
3696                ;; Do not distribute README.
3697                (("^env\\.Install\\(idir_prefix, 'README\\.md'\\)") ""))
3698              #t)))))
3699     (inputs
3700      `(("zlib" ,zlib)
3701        ("perl" ,perl)
3702        ("samtools" ,samtools)
3703        ("htslib" ,htslib)
3704        ("boost" ,boost)))
3705     (home-page "https://bitbucket.org/berkeleylab/metabat")
3706     (synopsis
3707      "Reconstruction of single genomes from complex microbial communities")
3708     (description
3709      "Grouping large genomic fragments assembled from shotgun metagenomic
3710 sequences to deconvolute complex microbial communities, or metagenome binning,
3711 enables the study of individual organisms and their interactions.  MetaBAT is
3712 an automated metagenome binning software, which integrates empirical
3713 probabilistic distances of genome abundance and tetranucleotide frequency.")
3714     ;; The source code contains inline assembly.
3715     (supported-systems '("x86_64-linux" "i686-linux"))
3716     (license (license:non-copyleft "file://license.txt"
3717                                    "See license.txt in the distribution."))))
3718
3719 (define-public minced
3720   (package
3721     (name "minced")
3722     (version "0.2.0")
3723     (source (origin
3724               (method url-fetch)
3725               (uri (string-append
3726                     "https://github.com/ctSkennerton/minced/archive/"
3727                     version ".tar.gz"))
3728               (file-name (string-append name "-" version ".tar.gz"))
3729               (sha256
3730                (base32
3731                 "0wxmlsapxfpxfd3ps9636h7i2xy6la8i42mwh0j2lsky63h63jp1"))))
3732     (build-system gnu-build-system)
3733     (arguments
3734      `(#:test-target "test"
3735        #:phases
3736        (modify-phases %standard-phases
3737          (delete 'configure)
3738          (add-before 'check 'fix-test
3739            (lambda _
3740              ;; Fix test for latest version.
3741              (substitute* "t/Aquifex_aeolicus_VF5.expected"
3742                (("minced:0.1.6") "minced:0.2.0"))
3743              #t))
3744          (replace 'install ; No install target.
3745            (lambda* (#:key inputs outputs #:allow-other-keys)
3746              (let* ((out (assoc-ref outputs "out"))
3747                     (bin (string-append out "/bin"))
3748                     (wrapper (string-append bin "/minced")))
3749                ;; Minced comes with a wrapper script that tries to figure out where
3750                ;; it is located before running the JAR. Since these paths are known
3751                ;; to us, we build our own wrapper to avoid coreutils dependency.
3752                (install-file "minced.jar" bin)
3753                (with-output-to-file wrapper
3754                  (lambda _
3755                    (display
3756                     (string-append
3757                      "#!" (assoc-ref inputs "bash") "/bin/sh\n\n"
3758                      (assoc-ref inputs "jre") "/bin/java -jar "
3759                      bin "/minced.jar \"$@\"\n"))))
3760                (chmod wrapper #o555)))))))
3761     (native-inputs
3762      `(("jdk" ,icedtea "jdk")))
3763     (inputs
3764      `(("bash" ,bash)
3765        ("jre" ,icedtea "out")))
3766     (home-page "https://github.com/ctSkennerton/minced")
3767     (synopsis "Mining CRISPRs in Environmental Datasets")
3768     (description
3769      "MinCED is a program to find Clustered Regularly Interspaced Short
3770 Palindromic Repeats (CRISPRs) in DNA sequences.  It can be used for
3771 unassembled metagenomic reads, but is mainly designed for full genomes and
3772 assembled metagenomic sequence.")
3773     (license license:gpl3+)))
3774
3775 (define-public miso
3776   (package
3777     (name "miso")
3778     (version "0.5.4")
3779     (source (origin
3780               (method url-fetch)
3781               (uri (pypi-uri "misopy" version))
3782               (sha256
3783                (base32
3784                 "1z3x0vd8ma7pdrnywj7i3kgwl89sdkwrrn62zl7r5calqaq2hyip"))
3785               (modules '((guix build utils)))
3786               (snippet
3787                '(substitute* "setup.py"
3788                   ;; Use setuptools, or else the executables are not
3789                   ;; installed.
3790                   (("distutils.core") "setuptools")
3791                   ;; use "gcc" instead of "cc" for compilation
3792                   (("^defines")
3793                    "cc.set_executables(
3794 compiler='gcc',
3795 compiler_so='gcc',
3796 linker_exe='gcc',
3797 linker_so='gcc -shared'); defines")))))
3798     (build-system python-build-system)
3799     (arguments
3800      `(#:python ,python-2 ; only Python 2 is supported
3801        #:tests? #f)) ; no "test" target
3802     (inputs
3803      `(("samtools" ,samtools)
3804        ("python-numpy" ,python2-numpy)
3805        ("python-pysam" ,python2-pysam)
3806        ("python-scipy" ,python2-scipy)
3807        ("python-matplotlib" ,python2-matplotlib)))
3808     (native-inputs
3809      `(("python-mock" ,python2-mock)   ;for tests
3810        ("python-pytz" ,python2-pytz))) ;for tests
3811     (home-page "http://genes.mit.edu/burgelab/miso/index.html")
3812     (synopsis "Mixture of Isoforms model for RNA-Seq isoform quantitation")
3813     (description
3814      "MISO (Mixture-of-Isoforms) is a probabilistic framework that quantitates
3815 the expression level of alternatively spliced genes from RNA-Seq data, and
3816 identifies differentially regulated isoforms or exons across samples.  By
3817 modeling the generative process by which reads are produced from isoforms in
3818 RNA-Seq, the MISO model uses Bayesian inference to compute the probability
3819 that a read originated from a particular isoform.")
3820     (license license:gpl2)))
3821
3822 (define-public muscle
3823   (package
3824     (name "muscle")
3825     (version "3.8.1551")
3826     (source (origin
3827               (method url-fetch/tarbomb)
3828               (uri (string-append
3829                     "http://www.drive5.com/muscle/muscle_src_"
3830                     version ".tar.gz"))
3831               (sha256
3832                (base32
3833                 "0bj8kj7sdizy3987zx6w7axihk40fk8rn76mpbqqjcnd64i5a367"))))
3834     (build-system gnu-build-system)
3835     (arguments
3836      `(#:make-flags (list "LDLIBS = -lm")
3837        #:phases
3838        (modify-phases %standard-phases
3839          (delete 'configure)
3840          (replace 'check
3841            ;; There are no tests, so just test if it runs.
3842            (lambda _ (zero? (system* "./muscle" "-version"))))
3843          (replace 'install
3844            (lambda* (#:key outputs #:allow-other-keys)
3845              (let* ((out (assoc-ref outputs "out"))
3846                     (bin (string-append out "/bin")))
3847                (install-file "muscle" bin)))))))
3848     (home-page "http://www.drive5.com/muscle")
3849     (synopsis "Multiple sequence alignment program")
3850     (description
3851      "MUSCLE aims to be a fast and accurate multiple sequence alignment
3852 program for nucleotide and protein sequences.")
3853     ;; License information found in 'muscle -h' and usage.cpp.
3854     (license license:public-domain)))
3855
3856 (define-public newick-utils
3857   ;; There are no recent releases so we package from git.
3858   (let ((commit "da121155a977197cab9fbb15953ca1b40b11eb87"))
3859     (package
3860       (name "newick-utils")
3861       (version (string-append "1.6-1." (string-take commit 8)))
3862       (source (origin
3863                 (method git-fetch)
3864                 (uri (git-reference
3865                       (url "https://github.com/tjunier/newick_utils.git")
3866                       (commit commit)))
3867                 (file-name (string-append name "-" version "-checkout"))
3868                 (sha256
3869                  (base32
3870                   "1hkw21rq1mwf7xp0rmbb2gqc0i6p11108m69i7mr7xcjl268pxnb"))))
3871     (build-system gnu-build-system)
3872     (arguments
3873      `(#:phases
3874        (modify-phases %standard-phases
3875          (add-after 'unpack 'autoconf
3876            (lambda _ (zero? (system* "autoreconf" "-vif")))))))
3877     (inputs
3878      ;; XXX: TODO: Enable Lua and Guile bindings.
3879      ;; https://github.com/tjunier/newick_utils/issues/13
3880      `(("libxml2" ,libxml2)
3881        ("flex" ,flex)
3882        ("bison" ,bison)))
3883     (native-inputs
3884      `(("autoconf" ,autoconf)
3885        ("automake" ,automake)
3886        ("libtool" ,libtool)))
3887     (synopsis "Programs for working with newick format phylogenetic trees")
3888     (description
3889      "Newick-utils is a suite of utilities for processing phylogenetic trees
3890 in Newick format.  Functions include re-rooting, extracting subtrees,
3891 trimming, pruning, condensing, drawing (ASCII graphics or SVG).")
3892     (home-page "https://github.com/tjunier/newick_utils")
3893     (license license:bsd-3))))
3894
3895 (define-public orfm
3896   (package
3897     (name "orfm")
3898     (version "0.7.1")
3899     (source (origin
3900               (method url-fetch)
3901               (uri (string-append
3902                     "https://github.com/wwood/OrfM/releases/download/v"
3903                     version "/orfm-" version ".tar.gz"))
3904               (sha256
3905                (base32
3906                 "16iigyr2gd8x0imzkk1dr3k5xsds9bpmwg31ayvjg0f4pir9rwqr"))))
3907     (build-system gnu-build-system)
3908     (inputs `(("zlib" ,zlib)))
3909     (native-inputs
3910      `(("ruby-bio-commandeer" ,ruby-bio-commandeer)
3911        ("ruby-rspec" ,ruby-rspec)
3912        ("ruby" ,ruby)))
3913     (synopsis "Simple and not slow open reading frame (ORF) caller")
3914     (description
3915      "An ORF caller finds stretches of DNA that, when translated, are not
3916 interrupted by stop codons.  OrfM finds and prints these ORFs.")
3917     (home-page "https://github.com/wwood/OrfM")
3918     (license license:lgpl3+)))
3919
3920 (define-public pplacer
3921   (let ((commit "g807f6f3"))
3922     (package
3923       (name "pplacer")
3924       ;; The commit should be updated with each version change.
3925       (version "1.1.alpha19")
3926       (source
3927        (origin
3928          (method url-fetch)
3929          (uri (string-append "https://github.com/matsen/pplacer/archive/v"
3930                              version ".tar.gz"))
3931          (file-name (string-append name "-" version ".tar.gz"))
3932          (sha256
3933           (base32 "0z1lnd2s8sh6kpzg106wzbh2szw7h0hvq8syd5a6wv4rmyyz6x0f"))))
3934       (build-system ocaml-build-system)
3935       (arguments
3936        `(#:ocaml ,ocaml-4.01
3937          #:findlib ,ocaml4.01-findlib
3938          #:modules ((guix build ocaml-build-system)
3939                     (guix build utils)
3940                     (ice-9 ftw))
3941          #:phases
3942          (modify-phases %standard-phases
3943            (delete 'configure)
3944            (add-after 'unpack 'replace-bundled-cddlib
3945              (lambda* (#:key inputs #:allow-other-keys)
3946                (let* ((cddlib-src (assoc-ref inputs "cddlib-src"))
3947                       (local-dir "cddlib_guix"))
3948                  (mkdir local-dir)
3949                  (with-directory-excursion local-dir
3950                    (system* "tar" "xvf" cddlib-src))
3951                  (let ((cddlib-src-folder
3952                         (string-append local-dir "/"
3953                                        (list-ref (scandir local-dir) 2)
3954                                        "/lib-src")))
3955                    (for-each
3956                     (lambda (file)
3957                       (copy-file file
3958                                  (string-append "cdd_src/" (basename file))))
3959                     (find-files cddlib-src-folder ".*[ch]$")))
3960                  #t)))
3961            (add-after 'unpack 'fix-makefile
3962              (lambda _
3963                ;; Remove system calls to 'git'.
3964                (substitute* "Makefile"
3965                  (("^DESCRIPT:=pplacer-.*")
3966                   (string-append
3967                    "DESCRIPT:=pplacer-$(shell uname)-v" ,version "\n")))
3968                (substitute* "myocamlbuild.ml"
3969                  (("git describe --tags --long .*\\\" with")
3970                   (string-append
3971                    "echo -n v" ,version "-" ,commit "\" with")))
3972                #t))
3973            (replace 'install
3974              (lambda* (#:key outputs #:allow-other-keys)
3975                (let* ((out (assoc-ref outputs "out"))
3976                       (bin (string-append out "/bin")))
3977                  (copy-recursively "bin" bin))
3978                #t)))))
3979       (native-inputs
3980        `(("zlib" ,zlib)
3981          ("gsl" ,gsl)
3982          ("ocaml-ounit" ,ocaml4.01-ounit)
3983          ("ocaml-batteries" ,ocaml4.01-batteries)
3984          ("ocaml-camlzip" ,ocaml4.01-camlzip)
3985          ("ocaml-csv" ,ocaml4.01-csv)
3986          ("ocaml-sqlite3" ,ocaml4.01-sqlite3)
3987          ("ocaml-xmlm" ,ocaml4.01-xmlm)
3988          ("ocaml-mcl" ,ocaml4.01-mcl)
3989          ("ocaml-gsl" ,ocaml4.01-gsl)
3990          ("cddlib-src" ,(package-source cddlib))))
3991       (propagated-inputs
3992        `(("pplacer-scripts" ,pplacer-scripts)))
3993       (synopsis "Phylogenetic placement of biological sequences")
3994       (description
3995        "Pplacer places query sequences on a fixed reference phylogenetic tree
3996 to maximize phylogenetic likelihood or posterior probability according to a
3997 reference alignment.  Pplacer is designed to be fast, to give useful
3998 information about uncertainty, and to offer advanced visualization and
3999 downstream analysis.")
4000       (home-page "http://matsen.fhcrc.org/pplacer")
4001       (license license:gpl3))))
4002
4003 ;; This package is installed alongside 'pplacer'.  It is a separate package so
4004 ;; that it can use the python-build-system for the scripts that are
4005 ;; distributed alongside the main OCaml binaries.
4006 (define pplacer-scripts
4007   (package
4008     (inherit pplacer)
4009     (name "pplacer-scripts")
4010     (build-system python-build-system)
4011     (arguments
4012      `(#:python ,python-2
4013        #:phases
4014        (modify-phases %standard-phases
4015          (add-after 'unpack 'enter-scripts-dir
4016            (lambda _ (chdir "scripts")))
4017          (replace 'check
4018            (lambda _
4019              (zero? (system* "python" "-m" "unittest" "discover" "-v"))))
4020          (add-after 'install 'wrap-executables
4021            (lambda* (#:key inputs outputs #:allow-other-keys)
4022              (let* ((out (assoc-ref outputs "out"))
4023                     (bin (string-append out "/bin")))
4024                (let ((path (string-append
4025                             (assoc-ref inputs "hmmer") "/bin:"
4026                             (assoc-ref inputs "infernal") "/bin")))
4027                  (display path)
4028                  (wrap-program (string-append bin "/refpkg_align.py")
4029                    `("PATH" ":" prefix (,path))))
4030                (let ((path (string-append
4031                             (assoc-ref inputs "hmmer") "/bin")))
4032                  (wrap-program (string-append bin "/hrefpkg_query.py")
4033                    `("PATH" ":" prefix (,path)))))
4034              #t)))))
4035     (inputs
4036      `(("infernal" ,infernal)
4037        ("hmmer" ,hmmer)))
4038     (propagated-inputs
4039      `(("python-biopython" ,python2-biopython)
4040        ("taxtastic" ,taxtastic)))
4041     (synopsis "Pplacer Python scripts")))
4042
4043 (define-public python2-pbcore
4044   (package
4045     (name "python2-pbcore")
4046     (version "1.2.10")
4047     (source (origin
4048               (method url-fetch)
4049               (uri (pypi-uri "pbcore" version))
4050               (sha256
4051                (base32
4052                 "1kjmv891d6qbpp4shhhvkl02ff4q5xlpnls2513sm2cjcrs52f1i"))))
4053     (build-system python-build-system)
4054     (arguments `(#:python ,python-2)) ; pbcore requires Python 2.7
4055     (propagated-inputs
4056      `(("python-cython" ,python2-cython)
4057        ("python-numpy" ,python2-numpy)
4058        ("python-pysam" ,python2-pysam)
4059        ("python-h5py" ,python2-h5py)))
4060     (native-inputs
4061      `(("python-nose" ,python2-nose)
4062        ("python-sphinx" ,python2-sphinx)
4063        ("python-pyxb" ,python2-pyxb)))
4064     (home-page "http://pacificbiosciences.github.io/pbcore/")
4065     (synopsis "Library for reading and writing PacBio data files")
4066     (description
4067      "The pbcore package provides Python APIs for interacting with PacBio data
4068 files and writing bioinformatics applications.")
4069     (license license:bsd-3)))
4070
4071 (define-public python2-warpedlmm
4072   (package
4073     (name "python2-warpedlmm")
4074     (version "0.21")
4075     (source
4076      (origin
4077        (method url-fetch)
4078        (uri (string-append
4079              "https://pypi.python.org/packages/source/W/WarpedLMM/WarpedLMM-"
4080              version ".zip"))
4081        (sha256
4082         (base32
4083          "1agfz6zqa8nc6cw47yh0s3y14gkpa9wqazwcj7mwwj3ffnw39p3j"))))
4084     (build-system python-build-system)
4085     (arguments
4086      `(#:python ,python-2))  ; requires Python 2.7
4087     (propagated-inputs
4088      `(("python-scipy" ,python2-scipy)
4089        ("python-numpy" ,python2-numpy)
4090        ("python-matplotlib" ,python2-matplotlib)
4091        ("python-fastlmm" ,python2-fastlmm)
4092        ("python-pandas" ,python2-pandas)
4093        ("python-pysnptools" ,python2-pysnptools)))
4094     (native-inputs
4095      `(("python-mock" ,python2-mock)
4096        ("python-nose" ,python2-nose)
4097        ("unzip" ,unzip)))
4098     (home-page "https://github.com/PMBio/warpedLMM")
4099     (synopsis "Implementation of warped linear mixed models")
4100     (description
4101      "WarpedLMM is a Python implementation of the warped linear mixed model,
4102 which automatically learns an optimal warping function (or transformation) for
4103 the phenotype as it models the data.")
4104     (license license:asl2.0)))
4105
4106 (define-public pbtranscript-tofu
4107   (let ((commit "8f5467fe6a4472bcfb4226c8720993c8507adfe4"))
4108     (package
4109       (name "pbtranscript-tofu")
4110       (version (string-append "2.2.3." (string-take commit 7)))
4111       (source (origin
4112                 (method git-fetch)
4113                 (uri (git-reference
4114                       (url "https://github.com/PacificBiosciences/cDNA_primer.git")
4115                       (commit commit)))
4116                 (file-name (string-append name "-" version "-checkout"))
4117                 (sha256
4118                  (base32
4119                   "1lgnpi35ihay42qx0b6yl3kkgra723i413j33kvs0kvs61h82w0f"))
4120                 (modules '((guix build utils)))
4121                 (snippet
4122                  '(begin
4123                     ;; remove bundled Cython sources
4124                     (delete-file "pbtranscript-tofu/pbtranscript/Cython-0.20.1.tar.gz")
4125                     #t))))
4126       (build-system python-build-system)
4127       (arguments
4128        `(#:python ,python-2
4129          ;; FIXME: Tests fail with "No such file or directory:
4130          ;; pbtools/pbtranscript/modified_bx_intervals/intersection_unique.so"
4131          #:tests? #f
4132          #:phases
4133          (modify-phases %standard-phases
4134            (add-after 'unpack 'enter-directory
4135             (lambda _
4136               (chdir "pbtranscript-tofu/pbtranscript/")
4137               #t))
4138            ;; With setuptools version 18.0 and later this setup.py hack causes
4139            ;; a build error, so we disable it.
4140            (add-after 'enter-directory 'patch-setuppy
4141             (lambda _
4142               (substitute* "setup.py"
4143                 (("if 'setuptools.extension' in sys.modules:")
4144                  "if False:"))
4145               #t)))))
4146       (inputs
4147        `(("python-numpy" ,python2-numpy)
4148          ("python-bx-python" ,python2-bx-python)
4149          ("python-networkx" ,python2-networkx)
4150          ("python-scipy" ,python2-scipy)
4151          ("python-pbcore" ,python2-pbcore)
4152          ("python-h5py" ,python2-h5py)))
4153       (native-inputs
4154        `(("python-cython" ,python2-cython)
4155          ("python-nose" ,python2-nose)))
4156       (home-page "https://github.com/PacificBiosciences/cDNA_primer")
4157       (synopsis "Analyze transcriptome data generated with the Iso-Seq protocol")
4158       (description
4159        "pbtranscript-tofu contains scripts to analyze transcriptome data
4160 generated using the PacBio Iso-Seq protocol.")
4161       (license license:bsd-3))))
4162
4163 (define-public prank
4164   (package
4165     (name "prank")
4166     (version "150803")
4167     (source (origin
4168               (method url-fetch)
4169               (uri (string-append
4170                     "http://wasabiapp.org/download/prank/prank.source."
4171                     version ".tgz"))
4172               (sha256
4173                (base32
4174                 "0am4z94fs3w2n5xpfls9zda61vq7qqz4q2i7b9hlsxz5q4j3kfm4"))))
4175     (build-system gnu-build-system)
4176     (arguments
4177      `(#:phases
4178        (modify-phases %standard-phases
4179          (add-after 'unpack 'enter-src-dir
4180             (lambda _
4181               (chdir "src")
4182               #t))
4183          (add-after 'unpack 'remove-m64-flag
4184            ;; Prank will build with the correct 'bit-ness' without this flag
4185            ;; and this allows building on 32-bit machines.
4186            (lambda _ (substitute* "src/Makefile"
4187                                   (("-m64") ""))
4188              #t))
4189          (delete 'configure)
4190          (replace 'install
4191            (lambda* (#:key outputs #:allow-other-keys)
4192              (let* ((out (assoc-ref outputs "out"))
4193                     (bin (string-append out "/bin"))
4194                     (man (string-append out "/share/man/man1"))
4195                     (path (string-append
4196                            (assoc-ref %build-inputs "mafft") "/bin:"
4197                            (assoc-ref %build-inputs "exonerate") "/bin:"
4198                            (assoc-ref %build-inputs "bppsuite") "/bin")))
4199                (install-file "prank" bin)
4200                (wrap-program (string-append bin "/prank")
4201                  `("PATH" ":" prefix (,path)))
4202                (install-file "prank.1" man))
4203              #t)))))
4204     (inputs
4205      `(("mafft" ,mafft)
4206        ("exonerate" ,exonerate)
4207        ("bppsuite" ,bppsuite)))
4208     (home-page "http://wasabiapp.org/software/prank/")
4209     (synopsis "Probabilistic multiple sequence alignment program")
4210     (description
4211      "PRANK is a probabilistic multiple sequence alignment program for DNA,
4212 codon and amino-acid sequences.  It is based on a novel algorithm that treats
4213 insertions correctly and avoids over-estimation of the number of deletion
4214 events.  In addition, PRANK borrows ideas from maximum likelihood methods used
4215 in phylogenetics and correctly takes into account the evolutionary distances
4216 between sequences.  Lastly, PRANK allows for defining a potential structure
4217 for sequences to be aligned and then, simultaneously with the alignment,
4218 predicts the locations of structural units in the sequences.")
4219     (license license:gpl2+)))
4220
4221 (define-public proteinortho
4222   (package
4223     (name "proteinortho")
4224     (version "5.16b")
4225     (source
4226      (origin
4227       (method url-fetch)
4228       (uri
4229        (string-append
4230         "http://www.bioinf.uni-leipzig.de/Software/proteinortho/proteinortho_v"
4231         version "_src.tar.gz"))
4232       (sha256
4233        (base32
4234         "1wl0dawpssqwfjvr651r4wlww8hhjin8nba6xh71ks7sbypx886j"))))
4235     (build-system gnu-build-system)
4236     (arguments
4237      `(#:test-target "test"
4238        #:phases
4239        (modify-phases %standard-phases
4240          (replace 'configure
4241            ;; There is no configure script, so we modify the Makefile directly.
4242            (lambda* (#:key outputs #:allow-other-keys)
4243              (substitute* "Makefile"
4244                (("INSTALLDIR=.*")
4245                 (string-append
4246                  "INSTALLDIR=" (assoc-ref outputs "out") "/bin\n")))
4247              #t))
4248          (add-before 'install 'make-install-directory
4249            ;; The install directory is not created during 'make install'.
4250            (lambda* (#:key outputs #:allow-other-keys)
4251              (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
4252              #t))
4253          (add-after 'install 'wrap-programs
4254            (lambda* (#:key inputs outputs #:allow-other-keys)
4255              (let* ((path (getenv "PATH"))
4256                     (out (assoc-ref outputs "out"))
4257                     (binary (string-append out "/bin/proteinortho5.pl")))
4258                (wrap-program binary `("PATH" ":" prefix (,path))))
4259              #t)))))
4260     (inputs
4261      `(("perl" ,perl)
4262        ("python" ,python-2)
4263        ("blast+" ,blast+)))
4264     (home-page "http://www.bioinf.uni-leipzig.de/Software/proteinortho")
4265     (synopsis "Detect orthologous genes across species")
4266     (description
4267      "Proteinortho is a tool to detect orthologous genes across different
4268 species.  For doing so, it compares similarities of given gene sequences and
4269 clusters them to find significant groups.  The algorithm was designed to handle
4270 large-scale data and can be applied to hundreds of species at once.")
4271     (license license:gpl2+)))
4272
4273 (define-public pyicoteo
4274   (package
4275     (name "pyicoteo")
4276     (version "2.0.7")
4277     (source
4278      (origin
4279        (method url-fetch)
4280        (uri (string-append "https://bitbucket.org/regulatorygenomicsupf/"
4281                            "pyicoteo/get/v" version ".tar.bz2"))
4282        (file-name (string-append name "-" version ".tar.bz2"))
4283        (sha256
4284         (base32
4285          "0d6087f29xp8wxwlj111c3sylli98n0l8ry58c51ixzq0zfm50wa"))))
4286     (build-system python-build-system)
4287     (arguments
4288      `(#:python ,python-2 ; does not work with Python 3
4289        #:tests? #f))      ; there are no tests
4290     (inputs
4291      `(("python2-matplotlib" ,python2-matplotlib)))
4292     (home-page "https://bitbucket.org/regulatorygenomicsupf/pyicoteo")
4293     (synopsis "Analyze high-throughput genetic sequencing data")
4294     (description
4295      "Pyicoteo is a suite of tools for the analysis of high-throughput genetic
4296 sequencing data.  It works with genomic coordinates.  There are currently six
4297 different command-line tools:
4298
4299 @enumerate
4300 @item pyicoregion: for generating exploratory regions automatically;
4301 @item pyicoenrich: for differential enrichment between two conditions;
4302 @item pyicoclip: for calling CLIP-Seq peaks without a control;
4303 @item pyicos: for genomic coordinates manipulation;
4304 @item pyicoller: for peak calling on punctuated ChIP-Seq;
4305 @item pyicount: to count how many reads from N experiment files overlap in a
4306   region file;
4307 @item pyicotrocol: to combine operations from pyicoteo.
4308 @end enumerate\n")
4309     (license license:gpl3+)))
4310
4311 (define-public prodigal
4312   (package
4313     (name "prodigal")
4314     (version "2.6.3")
4315     (source (origin
4316               (method url-fetch)
4317               (uri (string-append
4318                     "https://github.com/hyattpd/Prodigal/archive/v"
4319                     version ".tar.gz"))
4320               (file-name (string-append name "-" version ".tar.gz"))
4321               (sha256
4322                (base32
4323                 "17srxkqd3jc77xk15pfbgg1a9xahqg7337w95mrsia7mpza4l2c9"))))
4324     (build-system gnu-build-system)
4325     (arguments
4326      `(#:tests? #f ;no check target
4327        #:make-flags (list (string-append "INSTALLDIR="
4328                                          (assoc-ref %outputs "out")
4329                                          "/bin"))
4330        #:phases
4331        (modify-phases %standard-phases
4332          (delete 'configure))))
4333     (home-page "http://prodigal.ornl.gov")
4334     (synopsis "Protein-coding gene prediction for Archaea and Bacteria")
4335     (description
4336      "Prodigal runs smoothly on finished genomes, draft genomes, and
4337 metagenomes, providing gene predictions in GFF3, Genbank, or Sequin table
4338 format.  It runs quickly, in an unsupervised fashion, handles gaps, handles
4339 partial genes, and identifies translation initiation sites.")
4340     (license license:gpl3+)))
4341
4342 (define-public roary
4343   (package
4344     (name "roary")
4345     (version "3.11.0")
4346     (source
4347      (origin
4348        (method url-fetch)
4349        (uri (string-append
4350              "mirror://cpan/authors/id/A/AJ/AJPAGE/Bio-Roary-"
4351              version ".tar.gz"))
4352        (sha256
4353         (base32
4354          "10lw78x1xzvn7xzvnmh4bm3cak3ah5cssapl0yidvhaj1f44h29i"))))
4355     (build-system perl-build-system)
4356     (arguments
4357      `(#:phases
4358        (modify-phases %standard-phases
4359          (delete 'configure)
4360          (delete 'build)
4361          (replace 'check
4362            (lambda _
4363              ;; The tests are not run by default, so we run each test file
4364              ;; directly.
4365              (setenv "PATH" (string-append (getcwd) "/bin" ":"
4366                                            (getenv "PATH")))
4367              (setenv "PERL5LIB" (string-append (getcwd) "/lib" ":"
4368                                                (getenv "PERL5LIB")))
4369              (zero? (length (filter (lambda (file)
4370                                       (display file)(display "\n")
4371                                       (not (zero? (system* "perl" file))))
4372                                     (find-files "t" ".*\\.t$"))))))
4373          (replace 'install
4374            ;; There is no 'install' target in the Makefile.
4375            (lambda* (#:key outputs #:allow-other-keys)
4376              (let* ((out (assoc-ref outputs "out"))
4377                     (bin (string-append out "/bin"))
4378                     (perl (string-append out "/lib/perl5/site_perl"))
4379                     (roary-plots "contrib/roary_plots"))
4380                (mkdir-p bin)
4381                (mkdir-p perl)
4382                (copy-recursively "bin" bin)
4383                (copy-recursively "lib" perl)
4384                #t)))
4385          (add-after 'install 'wrap-programs
4386            (lambda* (#:key inputs outputs #:allow-other-keys)
4387              (let* ((out (assoc-ref outputs "out"))
4388                     (perl5lib (getenv "PERL5LIB"))
4389                     (path (getenv "PATH")))
4390                (for-each (lambda (prog)
4391                            (let ((binary (string-append out "/" prog)))
4392                              (wrap-program binary
4393                                `("PERL5LIB" ":" prefix
4394                                  (,(string-append perl5lib ":" out
4395                                                   "/lib/perl5/site_perl"))))
4396                              (wrap-program binary
4397                                `("PATH" ":" prefix
4398                                  (,(string-append path ":" out "/bin"))))))
4399                          (find-files "bin" ".*[^R]$"))
4400                (let ((file
4401                       (string-append out "/bin/roary-create_pan_genome_plots.R"))
4402                      (r-site-lib (getenv "R_LIBS_SITE"))
4403                      (coreutils-path
4404                       (string-append (assoc-ref inputs "coreutils") "/bin")))
4405                  (wrap-program file
4406                    `("R_LIBS_SITE" ":" prefix
4407                      (,(string-append r-site-lib ":" out "/site-library/"))))
4408                  (wrap-program file
4409                    `("PATH" ":" prefix
4410                      (,(string-append coreutils-path ":" out "/bin"))))))
4411              #t)))))
4412     (native-inputs
4413      `(("perl-env-path" ,perl-env-path)
4414        ("perl-test-files" ,perl-test-files)
4415        ("perl-test-most" ,perl-test-most)
4416        ("perl-test-output" ,perl-test-output)))
4417     (inputs
4418      `(("perl-array-utils" ,perl-array-utils)
4419        ("bioperl" ,bioperl-minimal)
4420        ("perl-digest-md5-file" ,perl-digest-md5-file)
4421        ("perl-exception-class" ,perl-exception-class)
4422        ("perl-file-find-rule" ,perl-file-find-rule)
4423        ("perl-file-grep" ,perl-file-grep)
4424        ("perl-file-slurper" ,perl-file-slurper)
4425        ("perl-file-which" ,perl-file-which)
4426        ("perl-graph" ,perl-graph)
4427        ("perl-graph-readwrite" ,perl-graph-readwrite)
4428        ("perl-log-log4perl" ,perl-log-log4perl)
4429        ("perl-moose" ,perl-moose)
4430        ("perl-perlio-utf8_strict" ,perl-perlio-utf8_strict)
4431        ("perl-text-csv" ,perl-text-csv)
4432        ("bedtools" ,bedtools)
4433        ("cd-hit" ,cd-hit)
4434        ("blast+" ,blast+)
4435        ("mcl" ,mcl)
4436        ("parallel" ,parallel)
4437        ("prank" ,prank)
4438        ("mafft" ,mafft)
4439        ("fasttree" ,fasttree)
4440        ("grep" ,grep)
4441        ("sed" ,sed)
4442        ("gawk" ,gawk)
4443        ("r-minimal" ,r-minimal)
4444        ("r-ggplot2" ,r-ggplot2)
4445        ("coreutils" ,coreutils)))
4446     (home-page "http://sanger-pathogens.github.io/Roary")
4447     (synopsis "High speed stand-alone pan genome pipeline")
4448     (description
4449      "Roary is a high speed stand alone pan genome pipeline, which takes
4450 annotated assemblies in GFF3 format (produced by the Prokka program) and
4451 calculates the pan genome.  Using a standard desktop PC, it can analyse
4452 datasets with thousands of samples, without compromising the quality of the
4453 results.  128 samples can be analysed in under 1 hour using 1 GB of RAM and a
4454 single processor.  Roary is not intended for metagenomics or for comparing
4455 extremely diverse sets of genomes.")
4456     (license license:gpl3)))
4457
4458 (define-public raxml
4459   (package
4460     (name "raxml")
4461     (version "8.2.10")
4462     (source
4463      (origin
4464        (method url-fetch)
4465        (uri
4466         (string-append
4467          "https://github.com/stamatak/standard-RAxML/archive/v"
4468          version ".tar.gz"))
4469        (file-name (string-append name "-" version ".tar.gz"))
4470        (sha256
4471         (base32
4472          "13s7aspfdcfr6asynwdg1x6vznys6pzap5f8wsffbnnwpkkg9ya8"))))
4473     (build-system gnu-build-system)
4474     (arguments
4475      `(#:tests? #f ; There are no tests.
4476        ;; Use 'standard' Makefile rather than SSE or AVX ones.
4477        #:make-flags (list "-f" "Makefile.HYBRID.gcc")
4478        #:phases
4479        (modify-phases %standard-phases
4480          (delete 'configure)
4481          (replace 'install
4482            (lambda* (#:key outputs #:allow-other-keys)
4483              (let* ((out (assoc-ref outputs "out"))
4484                     (bin (string-append out "/bin"))
4485                     (executable "raxmlHPC-HYBRID"))
4486                (install-file executable bin)
4487                (symlink (string-append bin "/" executable) "raxml"))
4488              #t)))))
4489     (inputs
4490      `(("openmpi" ,openmpi)))
4491     (home-page "http://sco.h-its.org/exelixis/web/software/raxml/index.html")
4492     (synopsis "Randomized Axelerated Maximum Likelihood phylogenetic trees")
4493     (description
4494      "RAxML is a tool for phylogenetic analysis and post-analysis of large
4495 phylogenies.")
4496     ;; The source includes x86 specific code
4497     (supported-systems '("x86_64-linux" "i686-linux"))
4498     (license license:gpl2+)))
4499
4500 (define-public rsem
4501   (package
4502     (name "rsem")
4503     (version "1.2.20")
4504     (source
4505      (origin
4506        (method url-fetch)
4507        (uri
4508         (string-append "http://deweylab.biostat.wisc.edu/rsem/src/rsem-"
4509                        version ".tar.gz"))
4510        (sha256
4511         (base32 "0nzdc0j0hjllhsd5f2xli95dafm3nawskigs140xzvjk67xh0r9q"))
4512        (patches (search-patches "rsem-makefile.patch"))
4513        (modules '((guix build utils)))
4514        (snippet
4515         '(begin
4516            ;; remove bundled copy of boost
4517            (delete-file-recursively "boost")
4518            #t))))
4519     (build-system gnu-build-system)
4520     (arguments
4521      `(#:tests? #f ;no "check" target
4522        #:phases
4523        (modify-phases %standard-phases
4524          ;; No "configure" script.
4525          ;; Do not build bundled samtools library.
4526          (replace 'configure
4527                   (lambda _
4528                     (substitute* "Makefile"
4529                       (("^all : sam/libbam.a") "all : "))
4530                     #t))
4531          (replace 'install
4532                   (lambda* (#:key outputs #:allow-other-keys)
4533                     (let* ((out (string-append (assoc-ref outputs "out")))
4534                            (bin (string-append out "/bin/"))
4535                            (perl (string-append out "/lib/perl5/site_perl")))
4536                       (mkdir-p bin)
4537                       (mkdir-p perl)
4538                       (for-each (lambda (file)
4539                                   (install-file file bin))
4540                                 (find-files "." "rsem-.*"))
4541                       (install-file "rsem_perl_utils.pm" perl))
4542                     #t))
4543          (add-after
4544           'install 'wrap-program
4545           (lambda* (#:key outputs #:allow-other-keys)
4546             (let ((out (assoc-ref outputs "out")))
4547               (for-each (lambda (prog)
4548                           (wrap-program (string-append out "/bin/" prog)
4549                             `("PERL5LIB" ":" prefix
4550                               (,(string-append out "/lib/perl5/site_perl")))))
4551                         '("rsem-plot-transcript-wiggles"
4552                           "rsem-calculate-expression"
4553                           "rsem-generate-ngvector"
4554                           "rsem-run-ebseq"
4555                           "rsem-prepare-reference")))
4556             #t)))))
4557     (inputs
4558      `(("boost" ,boost)
4559        ("ncurses" ,ncurses)
4560        ("r-minimal" ,r-minimal)
4561        ("perl" ,perl)
4562        ("samtools" ,samtools-0.1)
4563        ("zlib" ,zlib)))
4564     (home-page "http://deweylab.biostat.wisc.edu/rsem/")
4565     (synopsis "Estimate gene expression levels from RNA-Seq data")
4566     (description
4567      "RSEM is a software package for estimating gene and isoform expression
4568 levels from RNA-Seq data.  The RSEM package provides a user-friendly
4569 interface, supports threads for parallel computation of the EM algorithm,
4570 single-end and paired-end read data, quality scores, variable-length reads and
4571 RSPD estimation.  In addition, it provides posterior mean and 95% credibility
4572 interval estimates for expression levels.  For visualization, it can generate
4573 BAM and Wiggle files in both transcript-coordinate and genomic-coordinate.")
4574     (license license:gpl3+)))
4575
4576 (define-public rseqc
4577   (package
4578     (name "rseqc")
4579     (version "2.6.1")
4580     (source
4581      (origin
4582        (method url-fetch)
4583        (uri
4584         (string-append "mirror://sourceforge/rseqc/"
4585                        "RSeQC-" version ".tar.gz"))
4586        (sha256
4587         (base32 "15ly0254yi032qzkdplg00q144qfdsd986gh62829rl5bkxhj330"))
4588        (modules '((guix build utils)))
4589        (snippet
4590         '(begin
4591            ;; remove bundled copy of pysam
4592            (delete-file-recursively "lib/pysam")
4593            (substitute* "setup.py"
4594              ;; remove dependency on outdated "distribute" module
4595              (("^from distribute_setup import use_setuptools") "")
4596              (("^use_setuptools\\(\\)") "")
4597              ;; do not use bundled copy of pysam
4598              (("^have_pysam = False") "have_pysam = True"))))))
4599     (build-system python-build-system)
4600     (arguments `(#:python ,python-2))
4601     (inputs
4602      `(("python-cython" ,python2-cython)
4603        ("python-pysam" ,python2-pysam)
4604        ("python-numpy" ,python2-numpy)
4605        ("zlib" ,zlib)))
4606     (native-inputs
4607      `(("python-nose" ,python2-nose)))
4608     (home-page "http://rseqc.sourceforge.net/")
4609     (synopsis "RNA-seq quality control package")
4610     (description
4611      "RSeQC provides a number of modules that can comprehensively evaluate
4612 high throughput sequence data, especially RNA-seq data.  Some basic modules
4613 inspect sequence quality, nucleotide composition bias, PCR bias and GC bias,
4614 while RNA-seq specific modules evaluate sequencing saturation, mapped reads
4615 distribution, coverage uniformity, strand specificity, etc.")
4616     (license license:gpl3+)))
4617
4618 (define-public seek
4619   ;; There are no release tarballs.  According to the installation
4620   ;; instructions at http://seek.princeton.edu/installation.jsp, the latest
4621   ;; stable release is identified by this changeset ID.
4622   (let ((changeset "2329130")
4623         (revision "1"))
4624     (package
4625       (name "seek")
4626       (version (string-append "0-" revision "." changeset))
4627       (source (origin
4628                 (method hg-fetch)
4629                 (uri (hg-reference
4630                       (url "https://bitbucket.org/libsleipnir/sleipnir")
4631                       (changeset changeset)))
4632                 (sha256
4633                  (base32
4634                   "0qrvilwh18dpbhkf92qvxbmay0j75ra3jg2wrhz67gf538zzphsx"))))
4635       (build-system gnu-build-system)
4636       (arguments
4637        `(#:modules ((srfi srfi-1)
4638                     (guix build gnu-build-system)
4639                     (guix build utils))
4640          #:phases
4641          (let ((dirs '("SeekMiner"
4642                        "SeekEvaluator"
4643                        "SeekPrep"
4644                        "Distancer"
4645                        "Data2DB"
4646                        "PCL2Bin")))
4647            (modify-phases %standard-phases
4648              (replace 'bootstrap
4649                (lambda _
4650                  (invoke "bash" "gen_auto")))
4651              (add-after 'build 'build-additional-tools
4652                (lambda* (#:key make-flags #:allow-other-keys)
4653                  (every (lambda (dir)
4654                           (with-directory-excursion (string-append "tools/" dir)
4655                             (zero? (apply system* "make" make-flags))))
4656                         dirs)))
4657              (add-after 'install 'install-additional-tools
4658                (lambda* (#:key make-flags #:allow-other-keys)
4659                  (fold (lambda (dir result)
4660                          (with-directory-excursion (string-append "tools/" dir)
4661                            (and result
4662                                 (zero? (apply system*
4663                                               `("make" ,@make-flags "install"))))))
4664                        #t dirs)))))))
4665       (inputs
4666        `(("gsl" ,gsl)
4667          ("boost" ,boost)
4668          ("libsvm" ,libsvm)
4669          ("readline" ,readline)
4670          ("gengetopt" ,gengetopt)
4671          ("log4cpp" ,log4cpp)))
4672       (native-inputs
4673        `(("autoconf" ,autoconf)
4674          ("automake" ,automake)
4675          ("perl" ,perl)))
4676       (home-page "http://seek.princeton.edu")
4677       (synopsis "Gene co-expression search engine")
4678       (description
4679        "SEEK is a computational gene co-expression search engine.  SEEK provides
4680 biologists with a way to navigate the massive human expression compendium that
4681 now contains thousands of expression datasets.  SEEK returns a robust ranking
4682 of co-expressed genes in the biological area of interest defined by the user's
4683 query genes.  It also prioritizes thousands of expression datasets according
4684 to the user's query of interest.")
4685       (license license:cc-by3.0))))
4686
4687 (define-public samtools
4688   (package
4689     (name "samtools")
4690     (version "1.5")
4691     (source
4692      (origin
4693        (method url-fetch)
4694        (uri
4695         (string-append "mirror://sourceforge/samtools/samtools/"
4696                        version "/samtools-" version ".tar.bz2"))
4697        (sha256
4698         (base32
4699          "1xidmv0jmfy7l0kb32hdnlshcxgzi1hmygvig0cqrq1fhckdlhl5"))))
4700     (build-system gnu-build-system)
4701     (arguments
4702      `(#:modules ((ice-9 ftw)
4703                   (ice-9 regex)
4704                   (guix build gnu-build-system)
4705                   (guix build utils))
4706        #:make-flags (list (string-append "prefix=" (assoc-ref %outputs "out")))
4707        #:configure-flags (list "--with-ncurses" "--with-htslib=system")
4708        #:phases
4709        (modify-phases %standard-phases
4710          (add-after 'unpack 'patch-tests
4711            (lambda _
4712              (substitute* "test/test.pl"
4713                ;; The test script calls out to /bin/bash
4714                (("/bin/bash") (which "bash")))
4715              #t))
4716          (add-after 'install 'install-library
4717            (lambda* (#:key outputs #:allow-other-keys)
4718              (let ((lib (string-append (assoc-ref outputs "out") "/lib")))
4719                (install-file "libbam.a" lib)
4720                #t)))
4721          (add-after 'install 'install-headers
4722            (lambda* (#:key outputs #:allow-other-keys)
4723              (let ((include (string-append (assoc-ref outputs "out")
4724                                            "/include/samtools/")))
4725                (for-each (lambda (file)
4726                            (install-file file include))
4727                          (scandir "." (lambda (name) (string-match "\\.h$" name))))
4728                #t))))))
4729     (native-inputs `(("pkg-config" ,pkg-config)))
4730     (inputs
4731      `(("htslib" ,htslib)
4732        ("ncurses" ,ncurses)
4733        ("perl" ,perl)
4734        ("python" ,python)
4735        ("zlib" ,zlib)))
4736     (home-page "http://samtools.sourceforge.net")
4737     (synopsis "Utilities to efficiently manipulate nucleotide sequence alignments")
4738     (description
4739      "Samtools implements various utilities for post-processing nucleotide
4740 sequence alignments in the SAM, BAM, and CRAM formats, including indexing,
4741 variant calling (in conjunction with bcftools), and a simple alignment
4742 viewer.")
4743     (license license:expat)))
4744
4745 (define-public samtools-0.1
4746   ;; This is the most recent version of the 0.1 line of samtools.  The input
4747   ;; and output formats differ greatly from that used and produced by samtools
4748   ;; 1.x and is still used in many bioinformatics pipelines.
4749   (package (inherit samtools)
4750     (version "0.1.19")
4751     (source
4752      (origin
4753        (method url-fetch)
4754        (uri
4755         (string-append "mirror://sourceforge/samtools/samtools/"
4756                        version "/samtools-" version ".tar.bz2"))
4757        (sha256
4758         (base32 "1m33xsfwz0s8qi45lylagfllqg7fphf4dr0780rsvw75av9wk06h"))))
4759     (arguments
4760      `(#:tests? #f ;no "check" target
4761        ,@(substitute-keyword-arguments (package-arguments samtools)
4762            ((#:make-flags flags)
4763             `(cons "LIBCURSES=-lncurses" ,flags))
4764            ((#:phases phases)
4765             `(modify-phases ,phases
4766                (replace 'install
4767                  (lambda* (#:key outputs #:allow-other-keys)
4768                    (let ((bin (string-append
4769                                (assoc-ref outputs "out") "/bin")))
4770                      (mkdir-p bin)
4771                      (install-file "samtools" bin)
4772                      #t)))
4773                (delete 'patch-tests)
4774                (delete 'configure))))))))
4775
4776 (define-public mosaik
4777   (let ((commit "5c25216d3522d6a33e53875cd76a6d65001e4e67"))
4778     (package
4779       (name "mosaik")
4780       (version "2.2.30")
4781       (source (origin
4782                 ;; There are no release tarballs nor tags.
4783                 (method git-fetch)
4784                 (uri (git-reference
4785                       (url "https://github.com/wanpinglee/MOSAIK.git")
4786                       (commit commit)))
4787                 (file-name (string-append name "-" version))
4788                 (sha256
4789                  (base32
4790                   "17gj3s07cm77r41z92awh0bim7w7q7fbn0sf5nkqmcm1vw052qgw"))))
4791       (build-system gnu-build-system)
4792       (arguments
4793        `(#:tests? #f ; no tests
4794          #:make-flags (list "CC=gcc")
4795          #:phases
4796          (modify-phases %standard-phases
4797            (replace 'configure
4798                     (lambda _ (chdir "src") #t))
4799            (replace 'install
4800                     (lambda* (#:key outputs #:allow-other-keys)
4801                       (let ((bin (string-append (assoc-ref outputs "out")
4802                                                 "/bin")))
4803                         (mkdir-p bin)
4804                         (copy-recursively "../bin" bin)
4805                         #t))))))
4806       (inputs
4807        `(("perl" ,perl)
4808          ("zlib" ,zlib)))
4809       (supported-systems '("x86_64-linux"))
4810       (home-page "https://github.com/wanpinglee/MOSAIK")
4811       (synopsis "Map nucleotide sequence reads to reference genomes")
4812       (description
4813        "MOSAIK is a program for mapping second and third-generation sequencing
4814 reads to a reference genome.  MOSAIK can align reads generated by all the
4815 major sequencing technologies, including Illumina, Applied Biosystems SOLiD,
4816 Roche 454, Ion Torrent and Pacific BioSciences SMRT.")
4817       ;; MOSAIK is released under the GPLv2+ with the exception of third-party
4818       ;; code released into the public domain:
4819       ;; 1. fastlz by Ariya Hidayat - http://www.fastlz.org/
4820       ;; 2. MD5 implementation - RSA Data Security, RFC 1321
4821       (license (list license:gpl2+ license:public-domain)))))
4822
4823 (define-public ngs-sdk
4824   (package
4825     (name "ngs-sdk")
4826     (version "1.3.0")
4827     (source
4828      (origin
4829        (method url-fetch)
4830        (uri
4831         (string-append "https://github.com/ncbi/ngs/archive/"
4832                        version ".tar.gz"))
4833        (file-name (string-append name "-" version ".tar.gz"))
4834        (sha256
4835         (base32
4836          "1wiyf4c6nm2j87pv015cbi0qny5byf3pbvcw3likifz5dl56ag40"))))
4837     (build-system gnu-build-system)
4838     (arguments
4839      `(#:parallel-build? #f ; not supported
4840        #:tests? #f ; no "check" target
4841        #:phases
4842        (modify-phases %standard-phases
4843          (replace 'configure
4844            (lambda* (#:key outputs #:allow-other-keys)
4845              (let ((out (assoc-ref outputs "out")))
4846                ;; Allow 'konfigure.perl' to find 'package.prl'.
4847                (setenv "PERL5LIB"
4848                        (string-append ".:" (getenv "PERL5LIB")))
4849
4850                ;; The 'configure' script doesn't recognize things like
4851                ;; '--enable-fast-install'.
4852                (zero? (system* "./configure"
4853                                (string-append "--build-prefix=" (getcwd) "/build")
4854                                (string-append "--prefix=" out))))))
4855          (add-after 'unpack 'enter-dir
4856            (lambda _ (chdir "ngs-sdk") #t)))))
4857     (native-inputs `(("perl" ,perl)))
4858     ;; According to the test
4859     ;;   unless ($MARCH =~ /x86_64/i || $MARCH =~ /i?86/i)
4860     ;; in ngs-sdk/setup/konfigure.perl
4861     (supported-systems '("i686-linux" "x86_64-linux"))
4862     (home-page "https://github.com/ncbi/ngs")
4863     (synopsis "API for accessing Next Generation Sequencing data")
4864     (description
4865      "NGS is a domain-specific API for accessing reads, alignments and pileups
4866 produced from Next Generation Sequencing.  The API itself is independent from
4867 any particular back-end implementation, and supports use of multiple back-ends
4868 simultaneously.")
4869     (license license:public-domain)))
4870
4871 (define-public java-ngs
4872   (package (inherit ngs-sdk)
4873     (name "java-ngs")
4874     (arguments
4875      `(,@(substitute-keyword-arguments
4876              `(#:modules ((guix build gnu-build-system)
4877                           (guix build utils)
4878                           (srfi srfi-1)
4879                           (srfi srfi-26))
4880                          ,@(package-arguments ngs-sdk))
4881            ((#:phases phases)
4882             `(modify-phases ,phases
4883                (replace 'enter-dir (lambda _ (chdir "ngs-java") #t)))))))
4884     (inputs
4885      `(("jdk" ,icedtea "jdk")
4886        ("ngs-sdk" ,ngs-sdk)))
4887     (synopsis "Java bindings for NGS SDK")))
4888
4889 (define-public ncbi-vdb
4890   (package
4891     (name "ncbi-vdb")
4892     (version "2.8.2")
4893     (source
4894      (origin
4895        (method url-fetch)
4896        (uri
4897         (string-append "https://github.com/ncbi/ncbi-vdb/archive/"
4898                        version ".tar.gz"))
4899        (file-name (string-append name "-" version ".tar.gz"))
4900        (sha256
4901         (base32
4902          "1acn4bv81mfl137qnbn9995mjjhwd36pm0b7qli1iw5skrxa9j8m"))))
4903     (build-system gnu-build-system)
4904     (arguments
4905      `(#:parallel-build? #f ; not supported
4906        #:tests? #f ; no "check" target
4907        #:phases
4908        (modify-phases %standard-phases
4909          (add-before 'configure 'set-perl-search-path
4910            (lambda _
4911              ;; Work around "dotless @INC" build failure.
4912              (setenv "PERL5LIB"
4913                      (string-append (getcwd) "/setup:"
4914                                     (getenv "PERL5LIB")))
4915              #t))
4916          (replace 'configure
4917            (lambda* (#:key inputs outputs #:allow-other-keys)
4918              (let ((out (assoc-ref outputs "out")))
4919                ;; Override include path for libmagic
4920                (substitute* "setup/package.prl"
4921                  (("name => 'magic', Include => '/usr/include'")
4922                   (string-append "name=> 'magic', Include => '"
4923                                  (assoc-ref inputs "libmagic")
4924                                  "/include" "'")))
4925
4926                ;; Install kdf5 library (needed by sra-tools)
4927                (substitute* "build/Makefile.install"
4928                  (("LIBRARIES_TO_INSTALL =")
4929                   "LIBRARIES_TO_INSTALL = kdf5.$(VERSION_LIBX) kdf5.$(VERSION_SHLX)"))
4930
4931                (substitute* "build/Makefile.env"
4932                  (("CFLAGS      =" prefix)
4933                   (string-append prefix "-msse2 ")))
4934
4935                ;; Override search path for ngs-java
4936                (substitute* "setup/package.prl"
4937                  (("/usr/local/ngs/ngs-java")
4938                   (assoc-ref inputs "java-ngs")))
4939
4940                ;; The 'configure' script doesn't recognize things like
4941                ;; '--enable-fast-install'.
4942                (zero? (system*
4943                        "./configure"
4944                        (string-append "--build-prefix=" (getcwd) "/build")
4945                        (string-append "--prefix=" (assoc-ref outputs "out"))
4946                        (string-append "--debug")
4947                        (string-append "--with-xml2-prefix="
4948                                       (assoc-ref inputs "libxml2"))
4949                        (string-append "--with-ngs-sdk-prefix="
4950                                       (assoc-ref inputs "ngs-sdk"))
4951                        (string-append "--with-hdf5-prefix="
4952                                       (assoc-ref inputs "hdf5")))))))
4953          (add-after 'install 'install-interfaces
4954            (lambda* (#:key outputs #:allow-other-keys)
4955              ;; Install interface libraries.  On i686 the interface libraries
4956              ;; are installed to "linux/gcc/i386", so we need to use the Linux
4957              ;; architecture name ("i386") instead of the target system prefix
4958              ;; ("i686").
4959              (mkdir (string-append (assoc-ref outputs "out") "/ilib"))
4960              (copy-recursively (string-append "build/ncbi-vdb/linux/gcc/"
4961                                               ,(system->linux-architecture
4962                                                 (or (%current-target-system)
4963                                                     (%current-system)))
4964                                               "/rel/ilib")
4965                                (string-append (assoc-ref outputs "out")
4966                                               "/ilib"))
4967              ;; Install interface headers
4968              (copy-recursively "interfaces"
4969                                (string-append (assoc-ref outputs "out")
4970                                               "/include"))
4971              #t))
4972          ;; These files are needed by sra-tools.
4973          (add-after 'install 'install-configuration-files
4974            (lambda* (#:key outputs #:allow-other-keys)
4975              (let ((target (string-append (assoc-ref outputs "out") "/kfg")))
4976                (mkdir target)
4977                (install-file "libs/kfg/default.kfg" target)
4978                (install-file "libs/kfg/certs.kfg" target))
4979              #t)))))
4980     (inputs
4981      `(("libxml2" ,libxml2)
4982        ("ngs-sdk" ,ngs-sdk)
4983        ("java-ngs" ,java-ngs)
4984        ("libmagic" ,file)
4985        ("hdf5" ,hdf5)))
4986     (native-inputs `(("perl" ,perl)))
4987     ;; NCBI-VDB requires SSE capability.
4988     (supported-systems '("i686-linux" "x86_64-linux"))
4989     (home-page "https://github.com/ncbi/ncbi-vdb")
4990     (synopsis "Database engine for genetic information")
4991     (description
4992      "The NCBI-VDB library implements a highly compressed columnar data
4993 warehousing engine that is most often used to store genetic information.
4994 Databases are stored in a portable image within the file system, and can be
4995 accessed/downloaded on demand across HTTP.")
4996     (license license:public-domain)))
4997
4998 (define-public plink
4999   (package
5000     (name "plink")
5001     (version "1.07")
5002     (source
5003      (origin
5004        (method url-fetch)
5005        (uri (string-append
5006              "http://pngu.mgh.harvard.edu/~purcell/plink/dist/plink-"
5007              version "-src.zip"))
5008        (sha256
5009         (base32 "0as8gxm4pjyc8dxmm1sl873rrd7wn5qs0l29nqfnl31x8i467xaa"))
5010        (patches (search-patches "plink-1.07-unclobber-i.patch"
5011                                 "plink-endian-detection.patch"))))
5012     (build-system gnu-build-system)
5013     (arguments
5014      '(#:tests? #f ;no "check" target
5015        #:make-flags (list (string-append "LIB_LAPACK="
5016                                          (assoc-ref %build-inputs "lapack")
5017                                          "/lib/liblapack.so")
5018                           "WITH_LAPACK=1"
5019                           "FORCE_DYNAMIC=1"
5020                           ;; disable phoning home
5021                           "WITH_WEBCHECK=")
5022        #:phases
5023        (modify-phases %standard-phases
5024          ;; no "configure" script
5025          (delete 'configure)
5026          (replace 'install
5027                   (lambda* (#:key outputs #:allow-other-keys)
5028                     (let ((bin (string-append (assoc-ref outputs "out")
5029                                               "/bin/")))
5030                       (install-file "plink" bin)
5031                       #t))))))
5032     (inputs
5033      `(("zlib" ,zlib)
5034        ("lapack" ,lapack)))
5035     (native-inputs
5036      `(("unzip" ,unzip)))
5037     (home-page "http://pngu.mgh.harvard.edu/~purcell/plink/")
5038     (synopsis "Whole genome association analysis toolset")
5039     (description
5040      "PLINK is a whole genome association analysis toolset, designed to
5041 perform a range of basic, large-scale analyses in a computationally efficient
5042 manner.  The focus of PLINK is purely on analysis of genotype/phenotype data,
5043 so there is no support for steps prior to this (e.g. study design and
5044 planning, generating genotype or CNV calls from raw data).  Through
5045 integration with gPLINK and Haploview, there is some support for the
5046 subsequent visualization, annotation and storage of results.")
5047     ;; Code is released under GPLv2, except for fisher.h, which is under
5048     ;; LGPLv2.1+
5049     (license (list license:gpl2 license:lgpl2.1+))))
5050
5051 (define-public plink-ng
5052   (package (inherit plink)
5053     (name "plink-ng")
5054     (version "1.90b4")
5055     (source
5056      (origin
5057        (method url-fetch)
5058        (uri (string-append "https://github.com/chrchang/plink-ng/archive/v"
5059                            version ".tar.gz"))
5060        (file-name (string-append name "-" version ".tar.gz"))
5061        (sha256
5062         (base32 "09ixrds009aczjswxr2alcb774mksq5g0v78dgjjn1h4dky0kf9a"))))
5063     (build-system gnu-build-system)
5064     (arguments
5065      '(#:tests? #f ;no "check" target
5066        #:make-flags (list "BLASFLAGS=-llapack -lopenblas"
5067                           "CFLAGS=-Wall -O2 -DDYNAMIC_ZLIB=1"
5068                           "ZLIB=-lz"
5069                           "-f" "Makefile.std")
5070        #:phases
5071        (modify-phases %standard-phases
5072          (add-after 'unpack 'chdir
5073            (lambda _ (chdir "1.9") #t))
5074          (delete 'configure) ; no "configure" script
5075          (replace 'install
5076                   (lambda* (#:key outputs #:allow-other-keys)
5077                     (let ((bin (string-append (assoc-ref outputs "out")
5078                                               "/bin/")))
5079                       (install-file "plink" bin)
5080                       #t))))))
5081     (inputs
5082      `(("zlib" ,zlib)
5083        ("lapack" ,lapack)
5084        ("openblas" ,openblas)))
5085     (home-page "https://www.cog-genomics.org/plink/")
5086     (license license:gpl3+)))
5087
5088 (define-public smithlab-cpp
5089   (let ((revision "1")
5090         (commit "728a097bec88c6f4b8528b685932049e660eff2e"))
5091     (package
5092       (name "smithlab-cpp")
5093       (version (string-append "0." revision "." (string-take commit 7)))
5094       (source (origin
5095                 (method git-fetch)
5096                 (uri (git-reference
5097                       (url "https://github.com/smithlabcode/smithlab_cpp.git")
5098                       (commit commit)))
5099                 (file-name (string-append name "-" version "-checkout"))
5100                 (sha256
5101                  (base32
5102                   "0d476lmj312xk77kr9fzrv7z1bv96yfyx0w7y62ycmnfbx32ll74"))))
5103       (build-system gnu-build-system)
5104       (arguments
5105        `(#:modules ((guix build gnu-build-system)
5106                     (guix build utils)
5107                     (srfi srfi-26))
5108          #:tests? #f ;no "check" target
5109          #:phases
5110          (modify-phases %standard-phases
5111            (add-after 'unpack 'use-samtools-headers
5112             (lambda _
5113               (substitute* '("SAM.cpp"
5114                              "SAM.hpp")
5115                 (("sam.h") "samtools/sam.h"))
5116               #t))
5117            (replace 'install
5118             (lambda* (#:key outputs #:allow-other-keys)
5119               (let* ((out     (assoc-ref outputs "out"))
5120                      (lib     (string-append out "/lib"))
5121                      (include (string-append out "/include/smithlab-cpp")))
5122                 (mkdir-p lib)
5123                 (mkdir-p include)
5124                 (for-each (cut install-file <> lib)
5125                           (find-files "." "\\.o$"))
5126                 (for-each (cut install-file <> include)
5127                           (find-files "." "\\.hpp$")))
5128               #t))
5129            (delete 'configure))))
5130       (inputs
5131        `(("samtools" ,samtools-0.1)
5132          ("zlib" ,zlib)))
5133       (home-page "https://github.com/smithlabcode/smithlab_cpp")
5134       (synopsis "C++ helper library for functions used in Smith lab projects")
5135       (description
5136        "Smithlab CPP is a C++ library that includes functions used in many of
5137 the Smith lab bioinformatics projects, such as a wrapper around Samtools data
5138 structures, classes for genomic regions, mapped sequencing reads, etc.")
5139       (license license:gpl3+))))
5140
5141 (define-public preseq
5142   (package
5143     (name "preseq")
5144     (version "2.0")
5145     (source (origin
5146               (method url-fetch)
5147               (uri (string-append "https://github.com/smithlabcode/"
5148                                   "preseq/archive/v" version ".tar.gz"))
5149               (file-name (string-append name "-" version ".tar.gz"))
5150               (sha256
5151                (base32 "08r684l50pnxjpvmhzjgqq56yv9rfw90k8vx0nsrnrzk8mf9hsdq"))
5152               (modules '((guix build utils)))
5153               (snippet
5154                ;; Remove bundled samtools.
5155                '(delete-file-recursively "samtools"))))
5156     (build-system gnu-build-system)
5157     (arguments
5158      `(#:tests? #f ;no "check" target
5159        #:phases
5160        (modify-phases %standard-phases
5161          (delete 'configure))
5162        #:make-flags
5163        (list (string-append "PREFIX="
5164                             (assoc-ref %outputs "out"))
5165              (string-append "LIBBAM="
5166                             (assoc-ref %build-inputs "samtools")
5167                             "/lib/libbam.a")
5168              (string-append "SMITHLAB_CPP="
5169                             (assoc-ref %build-inputs "smithlab-cpp")
5170                             "/lib")
5171              "PROGS=preseq"
5172              "INCLUDEDIRS=$(SMITHLAB_CPP)/../include/smithlab-cpp $(SAMTOOLS_DIR)")))
5173     (inputs
5174      `(("gsl" ,gsl)
5175        ("samtools" ,samtools-0.1)
5176        ("smithlab-cpp" ,smithlab-cpp)
5177        ("zlib" ,zlib)))
5178     (home-page "http://smithlabresearch.org/software/preseq/")
5179     (synopsis "Program for analyzing library complexity")
5180     (description
5181      "The preseq package is aimed at predicting and estimating the complexity
5182 of a genomic sequencing library, equivalent to predicting and estimating the
5183 number of redundant reads from a given sequencing depth and how many will be
5184 expected from additional sequencing using an initial sequencing experiment.
5185 The estimates can then be used to examine the utility of further sequencing,
5186 optimize the sequencing depth, or to screen multiple libraries to avoid low
5187 complexity samples.")
5188     (license license:gpl3+)))
5189
5190 (define-public python-screed
5191   (package
5192     (name "python-screed")
5193     (version "0.9")
5194     (source
5195      (origin
5196        (method url-fetch)
5197        (uri (pypi-uri "screed" version))
5198        (sha256
5199         (base32
5200          "18czszp9fkx3j6jr7y5kp6dfialscgddk05mw1zkhh2zhn0jd8i0"))))
5201     (build-system python-build-system)
5202     (arguments
5203      `(#:phases
5204        (modify-phases %standard-phases
5205          (replace 'check
5206            (lambda _
5207              (setenv "PYTHONPATH"
5208                      (string-append (getenv "PYTHONPATH") ":."))
5209              (zero? (system* "nosetests" "--attr" "!known_failing")))))))
5210     (native-inputs
5211      `(("python-nose" ,python-nose)))
5212     (inputs
5213      `(("python-bz2file" ,python-bz2file)))
5214     (home-page "https://github.com/dib-lab/screed/")
5215     (synopsis "Short read sequence database utilities")
5216     (description "Screed parses FASTA and FASTQ files and generates databases.
5217 Values such as sequence name, sequence description, sequence quality and the
5218 sequence itself can be retrieved from these databases.")
5219     (license license:bsd-3)))
5220
5221 (define-public python2-screed
5222   (package-with-python2 python-screed))
5223
5224 (define-public sra-tools
5225   (package
5226     (name "sra-tools")
5227     (version "2.8.2-1")
5228     (source
5229      (origin
5230        (method url-fetch)
5231        (uri
5232         (string-append "https://github.com/ncbi/sra-tools/archive/"
5233                        version ".tar.gz"))
5234        (file-name (string-append name "-" version ".tar.gz"))
5235        (sha256
5236         (base32
5237          "1camsijmvv2s45mb4iyf44ghl4gkd4rl0viphpcgl3ccchy32a0g"))))
5238     (build-system gnu-build-system)
5239     (arguments
5240      `(#:parallel-build? #f ; not supported
5241        #:tests? #f ; no "check" target
5242        #:make-flags
5243        (list (string-append "DEFAULT_CRT="
5244                             (assoc-ref %build-inputs "ncbi-vdb")
5245                             "/kfg/certs.kfg")
5246              (string-append "DEFAULT_KFG="
5247                             (assoc-ref %build-inputs "ncbi-vdb")
5248                             "/kfg/default.kfg")
5249              (string-append "VDB_LIBDIR="
5250                             (assoc-ref %build-inputs "ncbi-vdb")
5251                             ,(if (string-prefix? "x86_64"
5252                                                  (or (%current-target-system)
5253                                                      (%current-system)))
5254                                  "/lib64"
5255                                  "/lib32")))
5256        #:phases
5257        (modify-phases %standard-phases
5258          (add-before 'configure 'set-perl-search-path
5259            (lambda _
5260              ;; Work around "dotless @INC" build failure.
5261              (setenv "PERL5LIB"
5262                      (string-append (getcwd) "/setup:"
5263                                     (getenv "PERL5LIB")))
5264              #t))
5265          (replace 'configure
5266            (lambda* (#:key inputs outputs #:allow-other-keys)
5267              ;; The build system expects a directory containing the sources and
5268              ;; raw build output of ncbi-vdb, including files that are not
5269              ;; installed.  Since we are building against an installed version of
5270              ;; ncbi-vdb, the following modifications are needed.
5271              (substitute* "setup/konfigure.perl"
5272                ;; Make the configure script look for the "ilib" directory of
5273                ;; "ncbi-vdb" without first checking for the existence of a
5274                ;; matching library in its "lib" directory.
5275                (("^            my \\$f = File::Spec->catdir\\(\\$libdir, \\$lib\\);")
5276                 "my $f = File::Spec->catdir($ilibdir, $ilib);")
5277                ;; Look for interface libraries in ncbi-vdb's "ilib" directory.
5278                (("my \\$ilibdir = File::Spec->catdir\\(\\$builddir, 'ilib'\\);")
5279                 "my $ilibdir = File::Spec->catdir($dir, 'ilib');"))
5280
5281              ;; Dynamic linking
5282              (substitute* "tools/copycat/Makefile"
5283                (("smagic-static") "lmagic"))
5284
5285              ;; The 'configure' script doesn't recognize things like
5286              ;; '--enable-fast-install'.
5287              (zero? (system*
5288                      "./configure"
5289                      (string-append "--build-prefix=" (getcwd) "/build")
5290                      (string-append "--prefix=" (assoc-ref outputs "out"))
5291                      (string-append "--debug")
5292                      (string-append "--with-fuse-prefix="
5293                                     (assoc-ref inputs "fuse"))
5294                      (string-append "--with-magic-prefix="
5295                                     (assoc-ref inputs "libmagic"))
5296                      ;; TODO: building with libxml2 fails with linker errors
5297                      ;; (string-append "--with-xml2-prefix="
5298                      ;;                (assoc-ref inputs "libxml2"))
5299                      (string-append "--with-ncbi-vdb-sources="
5300                                     (assoc-ref inputs "ncbi-vdb"))
5301                      (string-append "--with-ncbi-vdb-build="
5302                                     (assoc-ref inputs "ncbi-vdb"))
5303                      (string-append "--with-ngs-sdk-prefix="
5304                                     (assoc-ref inputs "ngs-sdk"))
5305                      (string-append "--with-hdf5-prefix="
5306                                     (assoc-ref inputs "hdf5"))))))
5307          ;; This version of sra-tools fails to build with glibc because of a
5308          ;; naming conflict.  glibc-2.25/include/bits/mathcalls.h already
5309          ;; contains a definition of "canonicalize", so we rename it.
5310          ;;
5311          ;; See upstream bug report:
5312          ;; https://github.com/ncbi/sra-tools/issues/67
5313          (add-after 'unpack 'patch-away-glibc-conflict
5314            (lambda _
5315              (substitute* "tools/bam-loader/bam.c"
5316                (("canonicalize\\(" line)
5317                 (string-append "sra_tools_" line)))
5318              #t)))))
5319     (native-inputs `(("perl" ,perl)))
5320     (inputs
5321      `(("ngs-sdk" ,ngs-sdk)
5322        ("ncbi-vdb" ,ncbi-vdb)
5323        ("libmagic" ,file)
5324        ("fuse" ,fuse)
5325        ("hdf5" ,hdf5)
5326        ("zlib" ,zlib)))
5327     (home-page "http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software")
5328     (synopsis "Tools and libraries for reading and writing sequencing data")
5329     (description
5330      "The SRA Toolkit from NCBI is a collection of tools and libraries for
5331 reading of sequencing files from the Sequence Read Archive (SRA) database and
5332 writing files into the .sra format.")
5333     (license license:public-domain)))
5334
5335 (define-public seqan
5336   (package
5337     (name "seqan")
5338     (version "1.4.2")
5339     (source (origin
5340               (method url-fetch)
5341               (uri (string-append "http://packages.seqan.de/seqan-library/"
5342                                   "seqan-library-" version ".tar.bz2"))
5343               (sha256
5344                (base32
5345                 "05s3wrrwn50f81aklfm65i4a749zag1vr8z03k21xm0pdxy47yvp"))))
5346     ;; The documentation is 7.8MB and the includes are 3.6MB heavy, so it
5347     ;; makes sense to split the outputs.
5348     (outputs '("out" "doc"))
5349     (build-system trivial-build-system)
5350     (arguments
5351      `(#:modules ((guix build utils))
5352        #:builder
5353        (begin
5354          (use-modules (guix build utils))
5355          (let ((tar  (assoc-ref %build-inputs "tar"))
5356                (bzip (assoc-ref %build-inputs "bzip2"))
5357                (out  (assoc-ref %outputs "out"))
5358                (doc  (assoc-ref %outputs "doc")))
5359            (setenv "PATH" (string-append tar "/bin:" bzip "/bin"))
5360            (system* "tar" "xvf" (assoc-ref %build-inputs "source"))
5361            (chdir (string-append "seqan-library-" ,version))
5362            (copy-recursively "include" (string-append out "/include"))
5363            (copy-recursively "share"  (string-append doc "/share"))))))
5364     (native-inputs
5365      `(("source" ,source)
5366        ("tar" ,tar)
5367        ("bzip2" ,bzip2)))
5368     (home-page "http://www.seqan.de")
5369     (synopsis "Library for nucleotide sequence analysis")
5370     (description
5371      "SeqAn is a C++ library of efficient algorithms and data structures for
5372 the analysis of sequences with the focus on biological data.  It contains
5373 algorithms and data structures for string representation and their
5374 manipulation, online and indexed string search, efficient I/O of
5375 bioinformatics file formats, sequence alignment, and more.")
5376     (license license:bsd-3)))
5377
5378 (define-public seqmagick
5379   (package
5380     (name "seqmagick")
5381     (version "0.7.0")
5382     (source
5383      (origin
5384        (method url-fetch)
5385        (uri (pypi-uri "seqmagick" version))
5386        (sha256
5387         (base32
5388          "12bfyp8nqi0hd36rmj450aygafp01qy3hkbvlwn3bk39pyjjkgg5"))))
5389     (build-system python-build-system)
5390     (inputs
5391      `(("python-biopython" ,python-biopython)))
5392     (native-inputs
5393      `(("python-nose" ,python-nose)))
5394     (home-page "https://github.com/fhcrc/seqmagick")
5395     (synopsis "Tools for converting and modifying sequence files")
5396     (description
5397      "Bioinformaticians often have to convert sequence files between formats
5398 and do little manipulations on them, and it's not worth writing scripts for
5399 that.  Seqmagick is a utility to expose the file format conversion in
5400 BioPython in a convenient way.  Instead of having a big mess of scripts, there
5401 is one that takes arguments.")
5402     (license license:gpl3)))
5403
5404 (define-public seqtk
5405   (package
5406     (name "seqtk")
5407     (version "1.2")
5408     (source (origin
5409               (method url-fetch)
5410               (uri (string-append
5411                     "https://github.com/lh3/seqtk/archive/v"
5412                     version ".tar.gz"))
5413               (file-name (string-append name "-" version ".tar.gz"))
5414               (sha256
5415                (base32
5416                 "0ywdyzpmfiz2wp6ampbzqg4y8bj450nfgqarpamg045b8mk32lxx"))
5417                             (modules '((guix build utils)))
5418               (snippet
5419                '(begin
5420                   ;; Remove extraneous header files, as is done in the seqtk
5421                   ;; master branch.
5422                   (for-each (lambda (file) (delete-file file))
5423                             (list "ksort.h" "kstring.h" "kvec.h"))
5424                   #t))))
5425     (build-system gnu-build-system)
5426     (arguments
5427      `(#:phases
5428        (modify-phases %standard-phases
5429          (delete 'configure)
5430          (replace 'check
5431            ;; There are no tests, so we just run a sanity check.
5432            (lambda _ (zero? (system* "./seqtk" "seq"))))
5433          (replace 'install
5434            (lambda* (#:key outputs #:allow-other-keys)
5435              (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5436                (install-file "seqtk" bin)))))))
5437     (inputs
5438      `(("zlib" ,zlib)))
5439     (home-page "https://github.com/lh3/seqtk")
5440     (synopsis "Toolkit for processing biological sequences in FASTA/Q format")
5441     (description
5442      "Seqtk is a fast and lightweight tool for processing sequences in the
5443 FASTA or FASTQ format.  It parses both FASTA and FASTQ files which can be
5444 optionally compressed by gzip.")
5445       (license license:expat)))
5446
5447 (define-public snap-aligner
5448   (package
5449     (name "snap-aligner")
5450     (version "1.0beta.18")
5451     (source (origin
5452               (method url-fetch)
5453               (uri (string-append
5454                     "https://github.com/amplab/snap/archive/v"
5455                     version ".tar.gz"))
5456               (file-name (string-append name "-" version ".tar.gz"))
5457               (sha256
5458                (base32
5459                 "1vnsjwv007k1fl1q7d681kbwn6bc66cgw6h16hym6gvyy71qv2ly"))))
5460     (build-system gnu-build-system)
5461     (arguments
5462      '(#:phases
5463        (modify-phases %standard-phases
5464          (delete 'configure)
5465          (replace 'check (lambda _ (zero? (system* "./unit_tests"))))
5466          (replace 'install
5467            (lambda* (#:key outputs #:allow-other-keys)
5468              (let* ((out (assoc-ref outputs "out"))
5469                     (bin (string-append out "/bin")))
5470                (install-file "snap-aligner" bin)
5471                (install-file "SNAPCommand" bin)
5472                #t))))))
5473     (native-inputs
5474      `(("zlib" ,zlib)))
5475     (home-page "http://snap.cs.berkeley.edu/")
5476     (synopsis "Short read DNA sequence aligner")
5477     (description
5478      "SNAP is a fast and accurate aligner for short DNA reads.  It is
5479 optimized for modern read lengths of 100 bases or higher, and takes advantage
5480 of these reads to align data quickly through a hash-based indexing scheme.")
5481     ;; 32-bit systems are not supported by the unpatched code.
5482     ;; Following the bug reports https://github.com/amplab/snap/issues/68 and
5483     ;; https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=812378 we see that
5484     ;; systems without a lot of memory cannot make good use of this program.
5485     (supported-systems '("x86_64-linux"))
5486     (license license:asl2.0)))
5487
5488 (define-public sortmerna
5489   (package
5490     (name "sortmerna")
5491     (version "2.1b")
5492     (source
5493      (origin
5494        (method url-fetch)
5495        (uri (string-append
5496              "https://github.com/biocore/sortmerna/archive/"
5497              version ".tar.gz"))
5498        (file-name (string-append name "-" version ".tar.gz"))
5499        (sha256
5500         (base32
5501          "1ghaghvd82af9j5adavxh77g7hm247d1r69m3fbi6f1jdivj5ldk"))))
5502     (build-system gnu-build-system)
5503     (outputs '("out"      ;for binaries
5504                "db"))     ;for sequence databases
5505     (arguments
5506      `(#:phases
5507        (modify-phases %standard-phases
5508          (replace 'install
5509            (lambda* (#:key outputs #:allow-other-keys)
5510              (let* ((out   (assoc-ref outputs "out"))
5511                     (bin   (string-append out "/bin"))
5512                     (db    (assoc-ref outputs "db"))
5513                     (share
5514                      (string-append db "/share/sortmerna/rRNA_databases")))
5515                (install-file "sortmerna" bin)
5516                (install-file "indexdb_rna" bin)
5517                (for-each (lambda (file)
5518                            (install-file file share))
5519                          (find-files "rRNA_databases" ".*fasta"))
5520                #t))))))
5521     (inputs
5522      `(("zlib" ,zlib)))
5523     (home-page "http://bioinfo.lifl.fr/RNA/sortmerna")
5524     (synopsis "Biological sequence analysis tool for NGS reads")
5525     (description
5526      "SortMeRNA is a biological sequence analysis tool for filtering, mapping
5527 and operational taxonomic unit (OTU) picking of next generation
5528 sequencing (NGS) reads.  The core algorithm is based on approximate seeds and
5529 allows for fast and sensitive analyses of nucleotide sequences.  The main
5530 application of SortMeRNA is filtering rRNA from metatranscriptomic data.")
5531     ;; The source includes x86 specific code
5532     (supported-systems '("x86_64-linux" "i686-linux"))
5533     (license license:lgpl3)))
5534
5535 (define-public star
5536   (package
5537     (name "star")
5538     (version "2.5.3a")
5539     (source (origin
5540               (method url-fetch)
5541               (uri (string-append "https://github.com/alexdobin/STAR/archive/"
5542                                   version ".tar.gz"))
5543               (file-name (string-append name "-" version ".tar.gz"))
5544               (sha256
5545                (base32
5546                 "013wirlz8lllgjyagl48l75n1isxyabqb3sj7qlsl0x1rmvqw99a"))
5547               (modules '((guix build utils)))
5548               (snippet
5549                '(begin
5550                   (substitute* "source/Makefile"
5551                     (("/bin/rm") "rm"))
5552                   ;; Remove pre-built binaries and bundled htslib sources.
5553                   (delete-file-recursively "bin/MacOSX_x86_64")
5554                   (delete-file-recursively "bin/Linux_x86_64")
5555                   (delete-file-recursively "bin/Linux_x86_64_static")
5556                   (delete-file-recursively "source/htslib")
5557                   #t))))
5558     (build-system gnu-build-system)
5559     (arguments
5560      '(#:tests? #f ;no check target
5561        #:make-flags '("STAR")
5562        #:phases
5563        (modify-phases %standard-phases
5564          (add-after 'unpack 'enter-source-dir
5565            (lambda _ (chdir "source") #t))
5566          (add-after 'enter-source-dir 'make-reproducible
5567            (lambda _
5568              (substitute* "Makefile"
5569                (("(COMPILATION_TIME_PLACE=\")(.*)(\")" _ pre mid post)
5570                 (string-append pre "Built with Guix" post)))))
5571          (add-after 'enter-source-dir 'do-not-use-bundled-htslib
5572            (lambda _
5573              (substitute* "Makefile"
5574                (("(Depend.list: \\$\\(SOURCES\\) parametersDefault\\.xxd) htslib"
5575                  _ prefix) prefix))
5576              (substitute* '("BAMfunctions.cpp"
5577                             "signalFromBAM.h"
5578                             "bam_cat.h"
5579                             "bam_cat.c"
5580                             "STAR.cpp"
5581                             "bamRemoveDuplicates.cpp")
5582                (("#include \"htslib/([^\"]+\\.h)\"" _ header)
5583                 (string-append "#include <" header ">")))
5584              (substitute* "IncludeDefine.h"
5585                (("\"htslib/(htslib/[^\"]+.h)\"" _ header)
5586                 (string-append "<" header ">")))
5587              #t))
5588          (replace 'install
5589            (lambda* (#:key outputs #:allow-other-keys)
5590              (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5591                (install-file "STAR" bin))
5592              #t))
5593          (delete 'configure))))
5594     (native-inputs
5595      `(("xxd" ,xxd)))
5596     (inputs
5597      `(("htslib" ,htslib)
5598        ("zlib" ,zlib)))
5599     (home-page "https://github.com/alexdobin/STAR")
5600     (synopsis "Universal RNA-seq aligner")
5601     (description
5602      "The Spliced Transcripts Alignment to a Reference (STAR) software is
5603 based on a previously undescribed RNA-seq alignment algorithm that uses
5604 sequential maximum mappable seed search in uncompressed suffix arrays followed
5605 by seed clustering and stitching procedure.  In addition to unbiased de novo
5606 detection of canonical junctions, STAR can discover non-canonical splices and
5607 chimeric (fusion) transcripts, and is also capable of mapping full-length RNA
5608 sequences.")
5609     ;; Only 64-bit systems are supported according to the README.
5610     (supported-systems '("x86_64-linux" "mips64el-linux"))
5611     ;; STAR is licensed under GPLv3 or later; htslib is MIT-licensed.
5612     (license license:gpl3+)))
5613
5614 (define-public subread
5615   (package
5616     (name "subread")
5617     (version "1.6.0")
5618     (source (origin
5619               (method url-fetch)
5620               (uri (string-append "mirror://sourceforge/subread/subread-"
5621                                   version "/subread-" version "-source.tar.gz"))
5622               (sha256
5623                (base32
5624                 "0ah0n4jx6ksk2m2j7xk385x2qzmk1y4rfc6a4mfrdqrlq721w99i"))))
5625     (build-system gnu-build-system)
5626     (arguments
5627      `(#:tests? #f ;no "check" target
5628       ;; The CC and CCFLAGS variables are set to contain a lot of x86_64
5629       ;; optimizations by default, so we override these flags such that x86_64
5630       ;; flags are only added when the build target is an x86_64 system.
5631        #:make-flags
5632        (list (let ((system ,(or (%current-target-system)
5633                                 (%current-system)))
5634                    (flags '("-ggdb" "-fomit-frame-pointer"
5635                             "-ffast-math" "-funroll-loops"
5636                             "-fmessage-length=0"
5637                             "-O9" "-Wall" "-DMAKE_FOR_EXON"
5638                             "-DMAKE_STANDALONE"
5639                             "-DSUBREAD_VERSION=\\\"${SUBREAD_VERSION}\\\""))
5640                    (flags64 '("-mmmx" "-msse" "-msse2" "-msse3")))
5641                (if (string-prefix? "x86_64" system)
5642                    (string-append "CCFLAGS=" (string-join (append flags flags64)))
5643                    (string-append "CCFLAGS=" (string-join flags))))
5644              "-f" "Makefile.Linux"
5645              "CC=gcc ${CCFLAGS}")
5646        #:phases
5647        (modify-phases %standard-phases
5648          (add-after 'unpack 'enter-dir
5649            (lambda _ (chdir "src") #t))
5650          (replace 'install
5651            (lambda* (#:key outputs #:allow-other-keys)
5652              (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5653                (mkdir-p bin)
5654                (copy-recursively "../bin" bin))))
5655          ;; no "configure" script
5656          (delete 'configure))))
5657     (inputs `(("zlib" ,zlib)))
5658     (home-page "http://bioinf.wehi.edu.au/subread-package/")
5659     (synopsis "Tool kit for processing next-gen sequencing data")
5660     (description
5661      "The subread package contains the following tools: subread aligner, a
5662 general-purpose read aligner; subjunc aligner: detecting exon-exon junctions
5663 and mapping RNA-seq reads; featureCounts: counting mapped reads for genomic
5664 features; exactSNP: a SNP caller that discovers SNPs by testing signals
5665 against local background noises.")
5666     (license license:gpl3+)))
5667
5668 (define-public stringtie
5669   (package
5670     (name "stringtie")
5671     (version "1.2.1")
5672     (source (origin
5673               (method url-fetch)
5674               (uri (string-append "http://ccb.jhu.edu/software/stringtie/dl/"
5675                                   "stringtie-" version ".tar.gz"))
5676               (sha256
5677                (base32
5678                 "1cqllsc1maq4kh92isi8yadgzbmnf042hlnalpk3y59aph1z3bfz"))
5679               (modules '((guix build utils)))
5680               (snippet
5681                '(begin
5682                   (delete-file-recursively "samtools-0.1.18")
5683                   #t))))
5684     (build-system gnu-build-system)
5685     (arguments
5686      `(#:tests? #f ;no test suite
5687        #:phases
5688        (modify-phases %standard-phases
5689          ;; no configure script
5690          (delete 'configure)
5691          (add-before 'build 'use-system-samtools
5692            (lambda _
5693              (substitute* "Makefile"
5694                (("stringtie: \\$\\{BAM\\}/libbam\\.a")
5695                 "stringtie: "))
5696              (substitute* '("gclib/GBam.h"
5697                             "gclib/GBam.cpp")
5698                (("#include \"(bam|sam|kstring).h\"" _ header)
5699                 (string-append "#include <samtools/" header ".h>")))
5700              #t))
5701          (add-after 'unpack 'remove-duplicate-typedef
5702            (lambda _
5703              ;; This typedef conflicts with the typedef in
5704              ;; glibc-2.25/include/bits/types.h
5705              (substitute* "gclib/GThreads.h"
5706                (("typedef long long __intmax_t;") ""))
5707              #t))
5708          (replace 'install
5709           (lambda* (#:key outputs #:allow-other-keys)
5710             (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5711               (install-file "stringtie" bin)
5712               #t))))))
5713     (inputs
5714      `(("samtools" ,samtools-0.1)
5715        ("zlib" ,zlib)))
5716     (home-page "http://ccb.jhu.edu/software/stringtie/")
5717     (synopsis "Transcript assembly and quantification for RNA-Seq data")
5718     (description
5719      "StringTie is a fast and efficient assembler of RNA-Seq sequence
5720 alignments into potential transcripts.  It uses a novel network flow algorithm
5721 as well as an optional de novo assembly step to assemble and quantitate
5722 full-length transcripts representing multiple splice variants for each gene
5723 locus.  Its input can include not only the alignments of raw reads used by
5724 other transcript assemblers, but also alignments of longer sequences that have
5725 been assembled from those reads.  To identify differentially expressed genes
5726 between experiments, StringTie's output can be processed either by the
5727 Cuffdiff or Ballgown programs.")
5728     (license license:artistic2.0)))
5729
5730 (define-public taxtastic
5731   (package
5732     (name "taxtastic")
5733     (version "0.6.4")
5734     (source (origin
5735               (method url-fetch)
5736               (uri (pypi-uri "taxtastic" version))
5737               (sha256
5738                (base32
5739                 "0s79z8kfl853x7l4h8ms05k31q87aw62nrchlk20w9n227j35929"))))
5740     (build-system python-build-system)
5741     (arguments
5742      `(#:python ,python-2
5743        #:phases
5744        (modify-phases %standard-phases
5745          (replace 'check
5746            (lambda _
5747              (zero? (system* "python" "-m" "unittest" "discover" "-v")))))))
5748     (propagated-inputs
5749      `(("python-sqlalchemy" ,python2-sqlalchemy)
5750        ("python-decorator" ,python2-decorator)
5751        ("python-biopython" ,python2-biopython)
5752        ("python-pandas" ,python2-pandas)))
5753     (home-page "https://github.com/fhcrc/taxtastic")
5754     (synopsis "Tools for taxonomic naming and annotation")
5755     (description
5756      "Taxtastic is software written in python used to build and maintain
5757 reference packages i.e. collections of reference trees, reference alignments,
5758 profiles, and associated taxonomic information.")
5759     (license license:gpl3+)))
5760
5761 (define-public vcftools
5762   (package
5763     (name "vcftools")
5764     (version "0.1.15")
5765     (source (origin
5766               (method url-fetch)
5767               (uri (string-append
5768                     "https://github.com/vcftools/vcftools/releases/download/v"
5769                     version "/vcftools-" version ".tar.gz"))
5770               (sha256
5771                (base32
5772                 "1qw30c45wihgy632rbz4rh3njnwj4msj46l1rsgdhyg6bgypmr1i"))))
5773     (build-system gnu-build-system)
5774     (arguments
5775      `(#:tests? #f ; no "check" target
5776        #:make-flags (list
5777                      "CFLAGS=-O2" ; override "-m64" flag
5778                      (string-append "PREFIX=" (assoc-ref %outputs "out"))
5779                      (string-append "MANDIR=" (assoc-ref %outputs "out")
5780                                     "/share/man/man1"))))
5781     (native-inputs
5782      `(("pkg-config" ,pkg-config)))
5783     (inputs
5784      `(("perl" ,perl)
5785        ("zlib" ,zlib)))
5786     (home-page "https://vcftools.github.io/")
5787     (synopsis "Tools for working with VCF files")
5788     (description
5789      "VCFtools is a program package designed for working with VCF files, such
5790 as those generated by the 1000 Genomes Project.  The aim of VCFtools is to
5791 provide easily accessible methods for working with complex genetic variation
5792 data in the form of VCF files.")
5793     ;; The license is declared as LGPLv3 in the README and
5794     ;; at https://vcftools.github.io/license.html
5795     (license license:lgpl3)))
5796
5797 (define-public infernal
5798   (package
5799     (name "infernal")
5800     (version "1.1.2")
5801     (source (origin
5802               (method url-fetch)
5803               (uri (string-append "http://eddylab.org/software/infernal/"
5804                                   "infernal-" version ".tar.gz"))
5805               (sha256
5806                (base32
5807                 "0sr2hiz3qxfwqpz3whxr6n82p3x27336v3f34iqznp10hks2935c"))))
5808     (build-system gnu-build-system)
5809     (native-inputs
5810      `(("perl" ,perl))) ; for tests
5811     (home-page "http://eddylab.org/infernal/")
5812     (synopsis "Inference of RNA alignments")
5813     (description "Infernal (\"INFERence of RNA ALignment\") is a tool for
5814 searching DNA sequence databases for RNA structure and sequence similarities.
5815 It is an implementation of a special case of profile stochastic context-free
5816 grammars called @dfn{covariance models} (CMs).  A CM is like a sequence
5817 profile, but it scores a combination of sequence consensus and RNA secondary
5818 structure consensus, so in many cases, it is more capable of identifying RNA
5819 homologs that conserve their secondary structure more than their primary
5820 sequence.")
5821     ;; Infernal 1.1.2 requires VMX or SSE capability for parallel instructions.
5822     (supported-systems '("i686-linux" "x86_64-linux"))
5823     (license license:bsd-3)))
5824
5825 (define-public r-centipede
5826   (package
5827     (name "r-centipede")
5828     (version "1.2")
5829     (source (origin
5830               (method url-fetch)
5831               (uri (string-append "http://download.r-forge.r-project.org/"
5832                                   "src/contrib/CENTIPEDE_" version ".tar.gz"))
5833               (sha256
5834                (base32
5835                 "1hsx6qgwr0i67fhy9257zj7s0ppncph2hjgbia5nn6nfmj0ax6l9"))))
5836     (build-system r-build-system)
5837     (home-page "http://centipede.uchicago.edu/")
5838     (synopsis "Predict transcription factor binding sites")
5839     (description
5840      "CENTIPEDE applies a hierarchical Bayesian mixture model to infer regions
5841 of the genome that are bound by particular transcription factors.  It starts
5842 by identifying a set of candidate binding sites, and then aims to classify the
5843 sites according to whether each site is bound or not bound by a transcription
5844 factor.  CENTIPEDE is an unsupervised learning algorithm that discriminates
5845 between two different types of motif instances using as much relevant
5846 information as possible.")
5847     (license (list license:gpl2+ license:gpl3+))))
5848
5849 (define-public r-vegan
5850   (package
5851     (name "r-vegan")
5852     (version "2.4-6")
5853     (source
5854      (origin
5855        (method url-fetch)
5856        (uri (cran-uri "vegan" version))
5857        (sha256
5858         (base32
5859          "175mqr42mmxy98gpf3rky8alnkw3w1k90ggdw3cispl36841p76w"))))
5860     (build-system r-build-system)
5861     (native-inputs
5862      `(("gfortran" ,gfortran)))
5863     (propagated-inputs
5864      `(("r-cluster" ,r-cluster)
5865        ("r-lattice" ,r-lattice)
5866        ("r-mass" ,r-mass)
5867        ("r-mgcv" ,r-mgcv)
5868        ("r-permute" ,r-permute)))
5869     (home-page "https://cran.r-project.org/web/packages/vegan")
5870     (synopsis "Functions for community ecology")
5871     (description
5872      "The vegan package provides tools for descriptive community ecology.  It
5873 has most basic functions of diversity analysis, community ordination and
5874 dissimilarity analysis.  Most of its multivariate tools can be used for other
5875 data types as well.")
5876     (license license:gpl2+)))
5877
5878 (define-public r-annotate
5879   (package
5880     (name "r-annotate")
5881     (version "1.56.1")
5882     (source
5883      (origin
5884        (method url-fetch)
5885        (uri (bioconductor-uri "annotate" version))
5886        (sha256
5887         (base32
5888          "14c5xd9kasvcwg5gbjys2c1vizxhlqlzxakqc2kml0kw97hmx0rq"))))
5889     (build-system r-build-system)
5890     (propagated-inputs
5891      `(("r-annotationdbi" ,r-annotationdbi)
5892        ("r-biobase" ,r-biobase)
5893        ("r-biocgenerics" ,r-biocgenerics)
5894        ("r-dbi" ,r-dbi)
5895        ("r-rcurl" ,r-rcurl)
5896        ("r-xml" ,r-xml)
5897        ("r-xtable" ,r-xtable)))
5898     (home-page
5899      "https://bioconductor.org/packages/annotate")
5900     (synopsis "Annotation for microarrays")
5901     (description "This package provides R environments for the annotation of
5902 microarrays.")
5903     (license license:artistic2.0)))
5904
5905 (define-public r-copynumber
5906   (package
5907     (name "r-copynumber")
5908     (version "1.18.0")
5909     (source (origin
5910               (method url-fetch)
5911               (uri (bioconductor-uri "copynumber" version))
5912               (sha256
5913                (base32
5914                 "01kcwzl485yjrkgyg8117b1il957ss0v6rq4bbxf4ksd5fzcjmyx"))))
5915     (build-system r-build-system)
5916     (propagated-inputs
5917      `(("r-s4vectors" ,r-s4vectors)
5918        ("r-iranges" ,r-iranges)
5919        ("r-genomicranges" ,r-genomicranges)
5920        ("r-biocgenerics" ,r-biocgenerics)))
5921     (home-page "https://bioconductor.org/packages/copynumber")
5922     (synopsis "Segmentation of single- and multi-track copy number data")
5923     (description
5924      "This package segments single- and multi-track copy number data by a
5925 penalized least squares regression method.")
5926     (license license:artistic2.0)))
5927
5928 (define-public r-geneplotter
5929   (package
5930     (name "r-geneplotter")
5931     (version "1.56.0")
5932     (source
5933      (origin
5934        (method url-fetch)
5935        (uri (bioconductor-uri "geneplotter" version))
5936        (sha256
5937         (base32
5938          "1z3g7frc1iviwrsv2dlm4nqvkc0685h4va0388yfxn102ln8wwma"))))
5939     (build-system r-build-system)
5940     (propagated-inputs
5941      `(("r-annotate" ,r-annotate)
5942        ("r-annotationdbi" ,r-annotationdbi)
5943        ("r-biobase" ,r-biobase)
5944        ("r-biocgenerics" ,r-biocgenerics)
5945        ("r-lattice" ,r-lattice)
5946        ("r-rcolorbrewer" ,r-rcolorbrewer)))
5947     (home-page "https://bioconductor.org/packages/geneplotter")
5948     (synopsis "Graphics functions for genomic data")
5949     (description
5950      "This package provides functions for plotting genomic data.")
5951     (license license:artistic2.0)))
5952
5953 (define-public r-genefilter
5954   (package
5955     (name "r-genefilter")
5956     (version "1.60.0")
5957     (source
5958      (origin
5959        (method url-fetch)
5960        (uri (bioconductor-uri "genefilter" version))
5961        (sha256
5962         (base32
5963          "173swlg6gj4kdllbqvyiw5dggbcxiwlwpqmllsv4dxzn7h25i3g7"))))
5964     (build-system r-build-system)
5965     (native-inputs
5966      `(("gfortran" ,gfortran)))
5967     (propagated-inputs
5968      `(("r-annotate" ,r-annotate)
5969        ("r-annotationdbi" ,r-annotationdbi)
5970        ("r-biobase" ,r-biobase)
5971        ("r-s4vectors" ,r-s4vectors)
5972        ("r-survival" ,r-survival)))
5973     (home-page "https://bioconductor.org/packages/genefilter")
5974     (synopsis "Filter genes from high-throughput experiments")
5975     (description
5976      "This package provides basic functions for filtering genes from
5977 high-throughput sequencing experiments.")
5978     (license license:artistic2.0)))
5979
5980 (define-public r-deseq2
5981   (package
5982     (name "r-deseq2")
5983     (version "1.18.1")
5984     (source
5985      (origin
5986        (method url-fetch)
5987        (uri (bioconductor-uri "DESeq2" version))
5988        (sha256
5989         (base32
5990          "1iyimg1s0x5pdmvl8x08s8h0v019y0nhjzs50chagbpk2x91fsmv"))))
5991     (properties `((upstream-name . "DESeq2")))
5992     (build-system r-build-system)
5993     (propagated-inputs
5994      `(("r-biobase" ,r-biobase)
5995        ("r-biocgenerics" ,r-biocgenerics)
5996        ("r-biocparallel" ,r-biocparallel)
5997        ("r-genefilter" ,r-genefilter)
5998        ("r-geneplotter" ,r-geneplotter)
5999        ("r-genomicranges" ,r-genomicranges)
6000        ("r-ggplot2" ,r-ggplot2)
6001        ("r-hmisc" ,r-hmisc)
6002        ("r-iranges" ,r-iranges)
6003        ("r-locfit" ,r-locfit)
6004        ("r-rcpp" ,r-rcpp)
6005        ("r-rcpparmadillo" ,r-rcpparmadillo)
6006        ("r-s4vectors" ,r-s4vectors)
6007        ("r-summarizedexperiment" ,r-summarizedexperiment)))
6008     (home-page "https://bioconductor.org/packages/DESeq2")
6009     (synopsis "Differential gene expression analysis")
6010     (description
6011      "This package provides functions to estimate variance-mean dependence in
6012 count data from high-throughput nucleotide sequencing assays and test for
6013 differential expression based on a model using the negative binomial
6014 distribution.")
6015     (license license:lgpl3+)))
6016
6017 (define-public r-dexseq
6018   (package
6019     (name "r-dexseq")
6020     (version "1.24.2")
6021     (source
6022      (origin
6023        (method url-fetch)
6024        (uri (bioconductor-uri "DEXSeq" version))
6025        (sha256
6026         (base32
6027          "18nh8ynxirfwkmc4sawdxgl7w1sl9ny5zpv8zbhv9vi5vgb8pxmj"))))
6028     (properties `((upstream-name . "DEXSeq")))
6029     (build-system r-build-system)
6030     (propagated-inputs
6031      `(("r-annotationdbi" ,r-annotationdbi)
6032        ("r-biobase" ,r-biobase)
6033        ("r-biocgenerics" ,r-biocgenerics)
6034        ("r-biocparallel" ,r-biocparallel)
6035        ("r-biomart" ,r-biomart)
6036        ("r-deseq2" ,r-deseq2)
6037        ("r-genefilter" ,r-genefilter)
6038        ("r-geneplotter" ,r-geneplotter)
6039        ("r-genomicranges" ,r-genomicranges)
6040        ("r-hwriter" ,r-hwriter)
6041        ("r-iranges" ,r-iranges)
6042        ("r-rcolorbrewer" ,r-rcolorbrewer)
6043        ("r-rsamtools" ,r-rsamtools)
6044        ("r-s4vectors" ,r-s4vectors)
6045        ("r-statmod" ,r-statmod)
6046        ("r-stringr" ,r-stringr)
6047        ("r-summarizedexperiment" ,r-summarizedexperiment)))
6048     (home-page "https://bioconductor.org/packages/DEXSeq")
6049     (synopsis "Inference of differential exon usage in RNA-Seq")
6050     (description
6051      "This package is focused on finding differential exon usage using RNA-seq
6052 exon counts between samples with different experimental designs.  It provides
6053 functions that allows the user to make the necessary statistical tests based
6054 on a model that uses the negative binomial distribution to estimate the
6055 variance between biological replicates and generalized linear models for
6056 testing.  The package also provides functions for the visualization and
6057 exploration of the results.")
6058     (license license:gpl3+)))
6059
6060 (define-public r-annotationforge
6061   (package
6062     (name "r-annotationforge")
6063     (version "1.20.0")
6064     (source
6065      (origin
6066        (method url-fetch)
6067        (uri (bioconductor-uri "AnnotationForge" version))
6068        (sha256
6069         (base32
6070          "01vbrf76vqfvxh6vpfxkjwccxggnha3byqzj333glqz2b6kwx5q1"))))
6071     (properties
6072      `((upstream-name . "AnnotationForge")))
6073     (build-system r-build-system)
6074     (propagated-inputs
6075      `(("r-annotationdbi" ,r-annotationdbi)
6076        ("r-biobase" ,r-biobase)
6077        ("r-biocgenerics" ,r-biocgenerics)
6078        ("r-dbi" ,r-dbi)
6079        ("r-rcurl" ,r-rcurl)
6080        ("r-rsqlite" ,r-rsqlite)
6081        ("r-s4vectors" ,r-s4vectors)
6082        ("r-xml" ,r-xml)))
6083     (home-page "https://bioconductor.org/packages/AnnotationForge")
6084     (synopsis "Code for building annotation database packages")
6085     (description
6086      "This package provides code for generating Annotation packages and their
6087 databases.  Packages produced are intended to be used with AnnotationDbi.")
6088     (license license:artistic2.0)))
6089
6090 (define-public r-rbgl
6091   (package
6092     (name "r-rbgl")
6093     (version "1.54.0")
6094     (source
6095      (origin
6096        (method url-fetch)
6097        (uri (bioconductor-uri "RBGL" version))
6098        (sha256
6099         (base32
6100          "18jad23i3899ypv4bg3l47cvvs3qnj1pqis2p9x0135yv5y6wnv7"))))
6101     (properties `((upstream-name . "RBGL")))
6102     (build-system r-build-system)
6103     (propagated-inputs `(("r-graph" ,r-graph)))
6104     (home-page "https://www.bioconductor.org/packages/RBGL")
6105     (synopsis "Interface to the Boost graph library")
6106     (description
6107      "This package provides a fairly extensive and comprehensive interface to
6108 the graph algorithms contained in the Boost library.")
6109     (license license:artistic2.0)))
6110
6111 (define-public r-gseabase
6112   (package
6113     (name "r-gseabase")
6114     (version "1.40.1")
6115     (source
6116      (origin
6117        (method url-fetch)
6118        (uri (bioconductor-uri "GSEABase" version))
6119        (sha256
6120         (base32
6121          "10cmjxahg2plwacfan6g0k8cwyzya96ypc7m1r79gwqkyykxw5fz"))))
6122     (properties `((upstream-name . "GSEABase")))
6123     (build-system r-build-system)
6124     (propagated-inputs
6125      `(("r-annotate" ,r-annotate)
6126        ("r-annotationdbi" ,r-annotationdbi)
6127        ("r-biobase" ,r-biobase)
6128        ("r-biocgenerics" ,r-biocgenerics)
6129        ("r-graph" ,r-graph)
6130        ("r-xml" ,r-xml)))
6131     (home-page "https://bioconductor.org/packages/GSEABase")
6132     (synopsis "Gene set enrichment data structures and methods")
6133     (description
6134      "This package provides classes and methods to support @dfn{Gene Set
6135 Enrichment Analysis} (GSEA).")
6136     (license license:artistic2.0)))
6137
6138 (define-public r-category
6139   (package
6140     (name "r-category")
6141     (version "2.44.0")
6142     (source
6143      (origin
6144        (method url-fetch)
6145        (uri (bioconductor-uri "Category" version))
6146        (sha256
6147         (base32
6148          "0mkav04vbla0xfa0dssxdd0rjs589sxi83xklf5iq5hj3dm8y0i8"))))
6149     (properties `((upstream-name . "Category")))
6150     (build-system r-build-system)
6151     (propagated-inputs
6152      `(("r-annotate" ,r-annotate)
6153        ("r-annotationdbi" ,r-annotationdbi)
6154        ("r-biobase" ,r-biobase)
6155        ("r-biocgenerics" ,r-biocgenerics)
6156        ("r-genefilter" ,r-genefilter)
6157        ("r-graph" ,r-graph)
6158        ("r-gseabase" ,r-gseabase)
6159        ("r-matrix" ,r-matrix)
6160        ("r-rbgl" ,r-rbgl)
6161        ("r-dbi" ,r-dbi)))
6162     (home-page "https://bioconductor.org/packages/Category")
6163     (synopsis "Category analysis")
6164     (description
6165      "This package provides a collection of tools for performing category
6166 analysis.")
6167     (license license:artistic2.0)))
6168
6169 (define-public r-gostats
6170   (package
6171     (name "r-gostats")
6172     (version "2.44.0")
6173     (source
6174      (origin
6175        (method url-fetch)
6176        (uri (bioconductor-uri "GOstats" version))
6177        (sha256
6178         (base32
6179          "04gqfdlx9fxf97qf0l28x4aaqvl10n6v58qiz5fiaw05sbj1pf1i"))))
6180     (properties `((upstream-name . "GOstats")))
6181     (build-system r-build-system)
6182     (propagated-inputs
6183      `(("r-annotate" ,r-annotate)
6184        ("r-annotationdbi" ,r-annotationdbi)
6185        ("r-annotationforge" ,r-annotationforge)
6186        ("r-biobase" ,r-biobase)
6187        ("r-category" ,r-category)
6188        ("r-go-db" ,r-go-db)
6189        ("r-graph" ,r-graph)
6190        ("r-rgraphviz" ,r-rgraphviz)
6191        ("r-rbgl" ,r-rbgl)))
6192     (home-page "https://bioconductor.org/packages/GOstats")
6193     (synopsis "Tools for manipulating GO and microarrays")
6194     (description
6195      "This package provides a set of tools for interacting with GO and
6196 microarray data.  A variety of basic manipulation tools for graphs, hypothesis
6197 testing and other simple calculations.")
6198     (license license:artistic2.0)))
6199
6200 (define-public r-shortread
6201   (package
6202     (name "r-shortread")
6203     (version "1.36.0")
6204     (source
6205      (origin
6206        (method url-fetch)
6207        (uri (bioconductor-uri "ShortRead" version))
6208        (sha256
6209         (base32
6210          "06mknlsmd4hnaxzdjapgvp2kgdnf9w103y500dsac5jgsz4vwzcz"))))
6211     (properties `((upstream-name . "ShortRead")))
6212     (build-system r-build-system)
6213     (inputs
6214      `(("zlib" ,zlib)))
6215     (propagated-inputs
6216      `(("r-biobase" ,r-biobase)
6217        ("r-biocgenerics" ,r-biocgenerics)
6218        ("r-biocparallel" ,r-biocparallel)
6219        ("r-biostrings" ,r-biostrings)
6220        ("r-genomeinfodb" ,r-genomeinfodb)
6221        ("r-genomicalignments" ,r-genomicalignments)
6222        ("r-genomicranges" ,r-genomicranges)
6223        ("r-hwriter" ,r-hwriter)
6224        ("r-iranges" ,r-iranges)
6225        ("r-lattice" ,r-lattice)
6226        ("r-latticeextra" ,r-latticeextra)
6227        ("r-rsamtools" ,r-rsamtools)
6228        ("r-s4vectors" ,r-s4vectors)
6229        ("r-xvector" ,r-xvector)
6230        ("r-zlibbioc" ,r-zlibbioc)))
6231     (home-page "https://bioconductor.org/packages/ShortRead")
6232     (synopsis "FASTQ input and manipulation tools")
6233     (description
6234      "This package implements sampling, iteration, and input of FASTQ files.
6235 It includes functions for filtering and trimming reads, and for generating a
6236 quality assessment report.  Data are represented as
6237 @code{DNAStringSet}-derived objects, and easily manipulated for a diversity of
6238 purposes.  The package also contains legacy support for early single-end,
6239 ungapped alignment formats.")
6240     (license license:artistic2.0)))
6241
6242 (define-public r-systempiper
6243   (package
6244     (name "r-systempiper")
6245     (version "1.12.0")
6246     (source
6247      (origin
6248        (method url-fetch)
6249        (uri (bioconductor-uri "systemPipeR" version))
6250        (sha256
6251         (base32
6252          "11mj8pjq5vj25768vmagpzv74fvi3p3kdk5zdlznqyiaggri04cv"))))
6253     (properties `((upstream-name . "systemPipeR")))
6254     (build-system r-build-system)
6255     (propagated-inputs
6256      `(("r-annotate" ,r-annotate)
6257        ("r-batchjobs" ,r-batchjobs)
6258        ("r-biocgenerics" ,r-biocgenerics)
6259        ("r-biostrings" ,r-biostrings)
6260        ("r-deseq2" ,r-deseq2)
6261        ("r-edger" ,r-edger)
6262        ("r-genomicfeatures" ,r-genomicfeatures)
6263        ("r-genomicranges" ,r-genomicranges)
6264        ("r-ggplot2" ,r-ggplot2)
6265        ("r-go-db" ,r-go-db)
6266        ("r-gostats" ,r-gostats)
6267        ("r-limma" ,r-limma)
6268        ("r-pheatmap" ,r-pheatmap)
6269        ("r-rjson" ,r-rjson)
6270        ("r-rsamtools" ,r-rsamtools)
6271        ("r-shortread" ,r-shortread)
6272        ("r-summarizedexperiment" ,r-summarizedexperiment)
6273        ("r-variantannotation" ,r-variantannotation)))
6274     (home-page "https://github.com/tgirke/systemPipeR")
6275     (synopsis "Next generation sequencing workflow and reporting environment")
6276     (description
6277      "This R package provides tools for building and running automated
6278 end-to-end analysis workflows for a wide range of @dfn{next generation
6279 sequence} (NGS) applications such as RNA-Seq, ChIP-Seq, VAR-Seq and Ribo-Seq.
6280 Important features include a uniform workflow interface across different NGS
6281 applications, automated report generation, and support for running both R and
6282 command-line software, such as NGS aligners or peak/variant callers, on local
6283 computers or compute clusters.  Efficient handling of complex sample sets and
6284 experimental designs is facilitated by a consistently implemented sample
6285 annotation infrastructure.")
6286     (license license:artistic2.0)))
6287
6288 (define-public r-grohmm
6289   (package
6290     (name "r-grohmm")
6291     (version "1.12.0")
6292     (source
6293      (origin
6294        (method url-fetch)
6295        (uri (bioconductor-uri "groHMM" version))
6296        (sha256
6297         (base32
6298          "0cjkj0ypyc4dfi9s8dh88kh6q4xlpnc0wal7njg4b4gqj0l2hva7"))))
6299     (properties `((upstream-name . "groHMM")))
6300     (build-system r-build-system)
6301     (propagated-inputs
6302      `(("r-genomeinfodb" ,r-genomeinfodb)
6303        ("r-genomicalignments" ,r-genomicalignments)
6304        ("r-genomicranges" ,r-genomicranges)
6305        ("r-iranges" ,r-iranges)
6306        ("r-mass" ,r-mass)
6307        ("r-rtracklayer" ,r-rtracklayer)
6308        ("r-s4vectors" ,r-s4vectors)))
6309     (home-page "https://github.com/Kraus-Lab/groHMM")
6310     (synopsis "GRO-seq analysis pipeline")
6311     (description
6312      "This package provides a pipeline for the analysis of GRO-seq data.")
6313     (license license:gpl3+)))
6314
6315 (define-public r-txdb-hsapiens-ucsc-hg19-knowngene
6316   (package
6317     (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
6318     (version "3.2.2")
6319     (source (origin
6320               (method url-fetch)
6321               ;; We cannot use bioconductor-uri here because this tarball is
6322               ;; located under "data/annotation/" instead of "bioc/".
6323               (uri (string-append "https://bioconductor.org/packages/"
6324                                   "release/data/annotation/src/contrib"
6325                                   "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
6326                                   version ".tar.gz"))
6327               (sha256
6328                (base32
6329                 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
6330     (properties
6331      `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
6332     (build-system r-build-system)
6333     ;; As this package provides little more than a very large data file it
6334     ;; doesn't make sense to build substitutes.
6335     (arguments `(#:substitutable? #f))
6336     (propagated-inputs
6337      `(("r-genomicfeatures" ,r-genomicfeatures)))
6338     (home-page
6339      "https://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
6340     (synopsis "Annotation package for human genome in TxDb format")
6341     (description
6342      "This package provides an annotation database of Homo sapiens genome
6343 data.  It is derived from the UCSC hg19 genome and based on the \"knownGene\"
6344 track.  The database is exposed as a @code{TxDb} object.")
6345     (license license:artistic2.0)))
6346
6347 (define-public r-sparql
6348   (package
6349   (name "r-sparql")
6350   (version "1.16")
6351   (source (origin
6352            (method url-fetch)
6353            (uri (cran-uri "SPARQL" version))
6354            (sha256
6355             (base32
6356              "0gak1q06yyhdmcxb2n3v0h9gr1vqd0viqji52wpw211qp6r6dcrc"))))
6357   (properties `((upstream-name . "SPARQL")))
6358   (build-system r-build-system)
6359   (propagated-inputs
6360    `(("r-rcurl" ,r-rcurl)
6361      ("r-xml" ,r-xml)))
6362   (home-page "https://cran.r-project.org/web/packages/SPARQL")
6363   (synopsis "SPARQL client for R")
6364   (description "This package provides an interface to use SPARQL to pose
6365 SELECT or UPDATE queries to an end-point.")
6366   ;; The only license indication is found in the DESCRIPTION file,
6367   ;; which states GPL-3.  So we cannot assume GPLv3+.
6368   (license license:gpl3)))
6369
6370 (define-public vsearch
6371   (package
6372     (name "vsearch")
6373     (version "2.7.1")
6374     (source
6375      (origin
6376        (method url-fetch)
6377        (uri (string-append
6378              "https://github.com/torognes/vsearch/archive/v"
6379              version ".tar.gz"))
6380        (file-name (string-append name "-" version ".tar.gz"))
6381        (sha256
6382         (base32
6383          "0jlzfgh79fzmb4g7sngzdjjsjyc37icvs1k7vmc2ksxglj6x5i7f"))
6384        (patches (search-patches "vsearch-unbundle-cityhash.patch"))
6385        (snippet
6386         '(begin
6387            ;; Remove bundled cityhash sources.  The vsearch source is adjusted
6388            ;; for this in the patch.
6389            (delete-file "src/city.h")
6390            (delete-file "src/citycrc.h")
6391            (delete-file "src/city.cc")
6392            #t))))
6393     (build-system gnu-build-system)
6394     (arguments
6395      `(#:phases
6396        (modify-phases %standard-phases
6397          (add-after 'unpack 'autogen
6398            (lambda _ (zero? (system* "autoreconf" "-vif")))))))
6399     (inputs
6400      `(("zlib" ,zlib)
6401        ("bzip2" ,bzip2)
6402        ("cityhash" ,cityhash)))
6403     (native-inputs
6404      `(("autoconf" ,autoconf)
6405        ("automake" ,automake)))
6406     (synopsis "Sequence search tools for metagenomics")
6407     (description
6408      "VSEARCH supports DNA sequence searching, clustering, chimera detection,
6409 dereplication, pairwise alignment, shuffling, subsampling, sorting and
6410 masking.  The tool takes advantage of parallelism in the form of SIMD
6411 vectorization as well as multiple threads to perform accurate alignments at
6412 high speed.  VSEARCH uses an optimal global aligner (full dynamic programming
6413 Needleman-Wunsch).")
6414     (home-page "https://github.com/torognes/vsearch")
6415     ;; vsearch uses non-portable SSE intrinsics so building fails on other
6416     ;; platforms.
6417     (supported-systems '("x86_64-linux"))
6418     ;; Dual licensed; also includes public domain source.
6419     (license (list license:gpl3 license:bsd-2))))
6420
6421 (define-public pardre
6422   (package
6423     (name "pardre")
6424     ;; The source of 1.1.5 changed in place, so we append "-1" to the version.
6425     (version "1.1.5-1")
6426     (source
6427      (origin
6428        (method url-fetch)
6429        (uri (string-append "mirror://sourceforge/pardre/ParDRe-rel"
6430                            "1.1.5" ".tar.gz"))
6431        (sha256
6432         (base32
6433          "17j73nc0viq4f6qj50nrndsrif5d6b71q8fl87m54psiv0ilns2b"))))
6434     (build-system gnu-build-system)
6435     (arguments
6436      `(#:tests? #f ; no tests included
6437        #:phases
6438        (modify-phases %standard-phases
6439          (delete 'configure)
6440          (replace 'install
6441            (lambda* (#:key outputs #:allow-other-keys)
6442              (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
6443                (install-file "ParDRe" bin)
6444                #t))))))
6445     (inputs
6446      `(("openmpi" ,openmpi)
6447        ("zlib" ,zlib)))
6448     (synopsis "Parallel tool to remove duplicate DNA reads")
6449     (description
6450      "ParDRe is a parallel tool to remove duplicate genetic sequence reads.
6451 Duplicate reads can be seen as identical or nearly identical sequences with
6452 some mismatches.  This tool lets users avoid the analysis of unnecessary
6453 reads, reducing the time of subsequent procedures with the
6454 dataset (e.g. assemblies, mappings, etc.).  The tool is implemented with MPI
6455 in order to exploit the parallel capabilities of multicore clusters.  It is
6456 faster than multithreaded counterparts (end of 2015) for the same number of
6457 cores and, thanks to the message-passing technology, it can be executed on
6458 clusters.")
6459     (home-page "https://sourceforge.net/projects/pardre/")
6460     (license license:gpl3+)))
6461
6462 (define-public ruby-bio-kseq
6463   (package
6464     (name "ruby-bio-kseq")
6465     (version "0.0.2")
6466     (source
6467      (origin
6468        (method url-fetch)
6469        (uri (rubygems-uri "bio-kseq" version))
6470        (sha256
6471         (base32
6472          "1xyaha46khb5jc6wzkbf7040jagac49jbimn0vcrzid0j8jdikrz"))))
6473     (build-system ruby-build-system)
6474     (arguments
6475      `(#:test-target "spec"))
6476     (native-inputs
6477      `(("bundler" ,bundler)
6478        ("ruby-rspec" ,ruby-rspec)
6479        ("ruby-rake-compiler" ,ruby-rake-compiler)))
6480     (inputs
6481      `(("zlib" ,zlib)))
6482     (synopsis "Ruby bindings for the kseq.h FASTA/Q parser")
6483     (description
6484      "@code{Bio::Kseq} provides ruby bindings to the @code{kseq.h} FASTA and
6485 FASTQ parsing code.  It provides a fast iterator over sequences and their
6486 quality scores.")
6487     (home-page "https://github.com/gusevfe/bio-kseq")
6488     (license license:expat)))
6489
6490 (define-public bio-locus
6491   (package
6492     (name "bio-locus")
6493     (version "0.0.7")
6494     (source
6495      (origin
6496        (method url-fetch)
6497        (uri (rubygems-uri "bio-locus" version))
6498        (sha256
6499         (base32
6500          "02vmrxyimkj9sahsp4zhfhnmbvz6dbbqz1y01vglf8cbwvkajfl0"))))
6501     (build-system ruby-build-system)
6502     (native-inputs
6503      `(("ruby-rspec" ,ruby-rspec)))
6504     (synopsis "Tool for fast querying of genome locations")
6505     (description
6506      "Bio-locus is a tabix-like tool for fast querying of genome
6507 locations.  Many file formats in bioinformatics contain records that
6508 start with a chromosome name and a position for a SNP, or a start-end
6509 position for indels.  Bio-locus allows users to store this chr+pos or
6510 chr+pos+alt information in a database.")
6511     (home-page "https://github.com/pjotrp/bio-locus")
6512     (license license:expat)))
6513
6514 (define-public bio-blastxmlparser
6515   (package
6516     (name "bio-blastxmlparser")
6517     (version "2.0.4")
6518     (source (origin
6519               (method url-fetch)
6520               (uri (rubygems-uri "bio-blastxmlparser" version))
6521               (sha256
6522                (base32
6523                 "1wf4qygcmdjgcqm6flmvsagfr1gs9lf63mj32qv3z1f481zc5692"))))
6524     (build-system ruby-build-system)
6525     (propagated-inputs
6526      `(("ruby-bio-logger" ,ruby-bio-logger)
6527        ("ruby-nokogiri" ,ruby-nokogiri)))
6528     (inputs
6529      `(("ruby-rspec" ,ruby-rspec)))
6530     (synopsis "Fast big data BLAST XML parser and library")
6531     (description
6532      "Very fast parallel big-data BLAST XML file parser which can be used as
6533 command line utility.  Use blastxmlparser to: Parse BLAST XML; filter output;
6534 generate FASTA, JSON, YAML, RDF, JSON-LD, HTML, CSV, tabular output etc.")
6535     (home-page "https://github.com/pjotrp/blastxmlparser")
6536     (license license:expat)))
6537
6538 (define-public bioruby
6539   (package
6540     (name "bioruby")
6541     (version "1.5.1")
6542     (source
6543      (origin
6544        (method url-fetch)
6545        (uri (rubygems-uri "bio" version))
6546        (sha256
6547         (base32
6548          "0hdl0789c9n4mprnx5pgd46bfwl8d000rqpamj5h6kkjgspijv49"))))
6549     (build-system ruby-build-system)
6550     (propagated-inputs
6551      `(("ruby-libxml" ,ruby-libxml)))
6552     (native-inputs
6553      `(("which" ,which)))  ; required for test phase
6554     (arguments
6555      `(#:phases
6556        (modify-phases %standard-phases
6557          (add-before 'build 'patch-test-command
6558           (lambda _
6559             (substitute* '("test/functional/bio/test_command.rb")
6560               (("/bin/sh") (which "sh")))
6561             (substitute* '("test/functional/bio/test_command.rb")
6562               (("/bin/ls") (which "ls")))
6563             (substitute* '("test/functional/bio/test_command.rb")
6564               (("which") (which "which")))
6565             (substitute* '("test/functional/bio/test_command.rb",
6566                            "test/data/command/echoarg2.sh")
6567               (("/bin/echo") (which "echo")))
6568             #t)))))
6569     (synopsis "Ruby library, shell and utilities for bioinformatics")
6570     (description "BioRuby comes with a comprehensive set of Ruby development
6571 tools and libraries for bioinformatics and molecular biology.  BioRuby has
6572 components for sequence analysis, pathway analysis, protein modelling and
6573 phylogenetic analysis; it supports many widely used data formats and provides
6574 easy access to databases, external programs and public web services, including
6575 BLAST, KEGG, GenBank, MEDLINE and GO.")
6576     (home-page "http://bioruby.org/")
6577     ;; Code is released under Ruby license, except for setup
6578     ;; (LGPLv2.1+) and scripts in samples (which have GPL2 and GPL2+)
6579     (license (list license:ruby license:lgpl2.1+ license:gpl2+ ))))
6580
6581 (define-public r-acsnminer
6582   (package
6583     (name "r-acsnminer")
6584     (version "0.16.8.25")
6585     (source (origin
6586               (method url-fetch)
6587               (uri (cran-uri "ACSNMineR" version))
6588               (sha256
6589                (base32
6590                 "0gh604s8qall6zfjlwcg2ilxjvz08dplf9k5g47idhv43scm748l"))))
6591     (properties `((upstream-name . "ACSNMineR")))
6592     (build-system r-build-system)
6593     (propagated-inputs
6594       `(("r-ggplot2" ,r-ggplot2)
6595         ("r-gridextra" ,r-gridextra)))
6596     (home-page "https://cran.r-project.org/web/packages/ACSNMineR")
6597     (synopsis "Gene enrichment analysis")
6598     (description
6599      "This package provides tools to compute and represent gene set enrichment
6600 or depletion from your data based on pre-saved maps from the @dfn{Atlas of
6601 Cancer Signalling Networks} (ACSN) or user imported maps.  The gene set
6602 enrichment can be run with hypergeometric test or Fisher exact test, and can
6603 use multiple corrections.  Visualization of data can be done either by
6604 barplots or heatmaps.")
6605     (license license:gpl2+)))
6606
6607 (define-public r-biocgenerics
6608   (package
6609     (name "r-biocgenerics")
6610     (version "0.24.0")
6611     (source (origin
6612               (method url-fetch)
6613               (uri (bioconductor-uri "BiocGenerics" version))
6614               (sha256
6615                (base32
6616                 "03wxvhxyrhipbgcg83lqlfn7p9gbzzrnl48y0dq7303xgp232zai"))))
6617     (properties
6618      `((upstream-name . "BiocGenerics")))
6619     (build-system r-build-system)
6620     (home-page "https://bioconductor.org/packages/BiocGenerics")
6621     (synopsis "S4 generic functions for Bioconductor")
6622     (description
6623      "This package provides S4 generic functions needed by many Bioconductor
6624 packages.")
6625     (license license:artistic2.0)))
6626
6627 (define-public r-biocinstaller
6628   (package
6629     (name "r-biocinstaller")
6630     (version "1.28.0")
6631     (source (origin
6632               (method url-fetch)
6633               (uri (bioconductor-uri "BiocInstaller" version))
6634               (sha256
6635                (base32
6636                 "19fga27bv6q9v5mpil74y76lahmnwvpg2h33rdx1r79nvljkd19d"))))
6637     (properties
6638      `((upstream-name . "BiocInstaller")))
6639     (build-system r-build-system)
6640     (home-page "https://bioconductor.org/packages/BiocInstaller")
6641     (synopsis "Install Bioconductor packages")
6642     (description "This package is used to install and update R packages from
6643 Bioconductor, CRAN, and Github.")
6644     (license license:artistic2.0)))
6645
6646 (define-public r-biocviews
6647   (package
6648     (name "r-biocviews")
6649     (version "1.46.0")
6650     (source (origin
6651               (method url-fetch)
6652               (uri (bioconductor-uri "biocViews" version))
6653               (sha256
6654                (base32
6655                 "09zyqj1kqc089lmh9sliy0acanx9zimcasvp71dsrg2bqm08r1md"))))
6656     (properties
6657      `((upstream-name . "biocViews")))
6658     (build-system r-build-system)
6659     (propagated-inputs
6660      `(("r-biobase" ,r-biobase)
6661        ("r-graph" ,r-graph)
6662        ("r-rbgl" ,r-rbgl)
6663        ("r-rcurl" ,r-rcurl)
6664        ("r-xml" ,r-xml)
6665        ("r-runit" ,r-runit)))
6666     (home-page "https://bioconductor.org/packages/biocViews")
6667     (synopsis "Bioconductor package categorization helper")
6668     (description "The purpose of biocViews is to create HTML pages that
6669 categorize packages in a Bioconductor package repository according to keywords,
6670 also known as views, in a controlled vocabulary.")
6671     (license license:artistic2.0)))
6672
6673 (define-public r-bookdown
6674   (package
6675     (name "r-bookdown")
6676     (version "0.7")
6677     (source (origin
6678               (method url-fetch)
6679               (uri (cran-uri "bookdown" version))
6680               (sha256
6681                (base32
6682                 "1b3fw1f41zph5yw3kynb47aijq53vhaa6mnnvxly72zamyzdf95q"))))
6683     (build-system r-build-system)
6684     (propagated-inputs
6685      `(("r-htmltools" ,r-htmltools)
6686        ("r-knitr" ,r-knitr)
6687        ("r-rmarkdown" ,r-rmarkdown)
6688        ("r-tinytex" ,r-tinytex)
6689        ("r-yaml" ,r-yaml)
6690        ("r-xfun" ,r-xfun)
6691        ("ghc-pandoc" ,ghc-pandoc)))
6692     (home-page "https://github.com/rstudio/bookdown")
6693     (synopsis "Authoring books and technical documents with R markdown")
6694     (description "This package provides output formats and utilities for
6695 authoring books and technical documents with R Markdown.")
6696     (license license:gpl3)))
6697
6698 (define-public r-biocstyle
6699   (package
6700    (name "r-biocstyle")
6701    (version "2.6.1")
6702     (source (origin
6703               (method url-fetch)
6704               (uri (bioconductor-uri "BiocStyle" version))
6705               (sha256
6706                (base32
6707                 "03pp04pkcq99kdv2spzr995h2cxsza7l6w3d4gp4112m06prcybm"))))
6708     (properties
6709      `((upstream-name . "BiocStyle")))
6710     (build-system r-build-system)
6711     (propagated-inputs
6712      `(("r-bookdown" ,r-bookdown)
6713        ("r-knitr" ,r-knitr)
6714        ("r-rmarkdown" ,r-rmarkdown)
6715        ("r-yaml" ,r-yaml)))
6716     (home-page "https://bioconductor.org/packages/BiocStyle")
6717     (synopsis "Bioconductor formatting styles")
6718     (description "This package provides standard formatting styles for
6719 Bioconductor PDF and HTML documents.  Package vignettes illustrate use and
6720 functionality.")
6721     (license license:artistic2.0)))
6722
6723 (define-public r-bioccheck
6724   (package
6725     (name "r-bioccheck")
6726     (version "1.14.0")
6727     (source (origin
6728               (method url-fetch)
6729               (uri (bioconductor-uri "BiocCheck" version))
6730               (sha256
6731                (base32
6732                 "1nzp8kgw13z9pgf885rplj6k37jcldfhbz0adqclxr2gq0yalmyx"))))
6733     (properties
6734      `((upstream-name . "BiocCheck")))
6735     (build-system r-build-system)
6736     (arguments
6737      '(#:phases
6738        (modify-phases %standard-phases
6739          ;; This package can be used by calling BiocCheck(<package>) from
6740          ;; within R, or by running R CMD BiocCheck <package>.  This phase
6741          ;; makes sure the latter works.  For this to work, the BiocCheck
6742          ;; script must be somewhere on the PATH (not the R bin directory).
6743          (add-after 'install 'install-bioccheck-subcommand
6744            (lambda* (#:key outputs #:allow-other-keys)
6745              (let* ((out (assoc-ref outputs "out"))
6746                     (dest-dir (string-append out "/bin"))
6747                     (script-dir
6748                      (string-append out "/site-library/BiocCheck/script/")))
6749                (mkdir-p dest-dir)
6750                (symlink (string-append script-dir "/checkBadDeps.R")
6751                         (string-append dest-dir "/checkBadDeps.R"))
6752                (symlink (string-append script-dir "/BiocCheck")
6753                         (string-append dest-dir "/BiocCheck")))
6754              #t)))))
6755     (propagated-inputs
6756      `(("r-codetools" ,r-codetools)
6757        ("r-graph" ,r-graph)
6758        ("r-httr" ,r-httr)
6759        ("r-optparse" ,r-optparse)
6760        ("r-biocinstaller" ,r-biocinstaller)
6761        ("r-biocviews" ,r-biocviews)
6762        ("r-stringdist" ,r-stringdist)))
6763     (home-page "https://bioconductor.org/packages/BiocCheck")
6764     (synopsis "Executes Bioconductor-specific package checks")
6765     (description "This package contains tools to perform additional quality
6766 checks on R packages that are to be submitted to the Bioconductor repository.")
6767     (license license:artistic2.0)))
6768
6769 (define-public r-getopt
6770   (package
6771     (name "r-getopt")
6772     (version "1.20.2")
6773     (source
6774      (origin
6775        (method url-fetch)
6776        (uri (cran-uri "getopt" version))
6777        (sha256
6778         (base32
6779          "13p35lbpy7i578752fa71sbfvcsqw5qfa9p6kf8b5m3c5p9i4v1x"))))
6780     (build-system r-build-system)
6781     (home-page "https://github.com/trevorld/getopt")
6782     (synopsis "Command-line option processor for R")
6783     (description
6784      "This package is designed to be used with Rscript to write shebang
6785 scripts that accept short and long options.  Many users will prefer to
6786 use the packages @code{optparse} or @code{argparse} which add extra
6787 features like automatically generated help options and usage texts,
6788 support for default values, positional argument support, etc.")
6789     (license license:gpl2+)))
6790
6791 (define-public r-optparse
6792   (package
6793     (name "r-optparse")
6794     (version "1.4.4")
6795     (source
6796      (origin
6797        (method url-fetch)
6798        (uri (cran-uri "optparse" version))
6799        (sha256
6800         (base32
6801          "1ff4wmsszrb3spwfp7ynfs8w11qpy1sdzfxm1wk8dqqvdwris7qb"))))
6802     (build-system r-build-system)
6803     (propagated-inputs
6804      `(("r-getopt" ,r-getopt)))
6805     (home-page
6806      "https://github.com/trevorld/optparse")
6807     (synopsis "Command line option parser")
6808     (description
6809      "This package provides a command line parser inspired by Python's
6810 @code{optparse} library to be used with Rscript to write shebang scripts
6811 that accept short and long options.")
6812     (license license:gpl2+)))
6813
6814 (define-public r-dnacopy
6815   (package
6816     (name "r-dnacopy")
6817     (version "1.52.0")
6818     (source (origin
6819               (method url-fetch)
6820               (uri (bioconductor-uri "DNAcopy" version))
6821               (sha256
6822                (base32
6823                 "127il5rlg1hzjlhwhs64x3nm18p00q1pd9ckb2b9ifl0rax95wai"))))
6824     (properties
6825      `((upstream-name . "DNAcopy")))
6826     (build-system r-build-system)
6827     (inputs
6828      `(("gfortran" ,gfortran)))
6829     (home-page "https://bioconductor.org/packages/DNAcopy")
6830     (synopsis "Implementation of a circular binary segmentation algorithm")
6831     (description "This package implements the circular binary segmentation (CBS)
6832 algorithm to segment DNA copy number data and identify genomic regions with
6833 abnormal copy number.")
6834     (license license:gpl2+)))
6835
6836 (define-public r-s4vectors
6837   (package
6838     (name "r-s4vectors")
6839     (version "0.16.0")
6840     (source (origin
6841               (method url-fetch)
6842               (uri (bioconductor-uri "S4Vectors" version))
6843               (sha256
6844                (base32
6845                 "03s8vz33nl6mivjb7dbvj702dkypi340lji1sjban03fyyls0hw0"))))
6846     (properties
6847      `((upstream-name . "S4Vectors")))
6848     (build-system r-build-system)
6849     (propagated-inputs
6850      `(("r-biocgenerics" ,r-biocgenerics)))
6851     (home-page "https://bioconductor.org/packages/S4Vectors")
6852     (synopsis "S4 implementation of vectors and lists")
6853     (description
6854      "The S4Vectors package defines the @code{Vector} and @code{List} virtual
6855 classes and a set of generic functions that extend the semantic of ordinary
6856 vectors and lists in R.  Package developers can easily implement vector-like
6857 or list-like objects as concrete subclasses of @code{Vector} or @code{List}.
6858 In addition, a few low-level concrete subclasses of general interest (e.g.
6859 @code{DataFrame}, @code{Rle}, and @code{Hits}) are implemented in the
6860 S4Vectors package itself.")
6861     (license license:artistic2.0)))
6862
6863 (define-public r-seqinr
6864   (package
6865     (name "r-seqinr")
6866     (version "3.4-5")
6867     (source
6868       (origin
6869         (method url-fetch)
6870         (uri (cran-uri "seqinr" version))
6871         (sha256
6872           (base32
6873             "17zv0n5cji17izwmwg0jcbxbjl3w5rls91w15svcnlpxjms38ahn"))))
6874     (build-system r-build-system)
6875     (propagated-inputs
6876      `(("r-ade4" ,r-ade4)
6877        ("r-segmented" ,r-segmented)))
6878     (inputs
6879      `(("zlib" ,zlib)))
6880     (home-page "http://seqinr.r-forge.r-project.org/")
6881     (synopsis "Biological sequences retrieval and analysis")
6882     (description
6883      "This package provides tools for exploratory data analysis and data
6884 visualization of biological sequence (DNA and protein) data.  It also includes
6885 utilities for sequence data management under the ACNUC system.")
6886     (license license:gpl2+)))
6887
6888 (define-public r-iranges
6889   (package
6890     (name "r-iranges")
6891     (version "2.12.0")
6892     (source (origin
6893               (method url-fetch)
6894               (uri (bioconductor-uri "IRanges" version))
6895               (sha256
6896                (base32
6897                 "1vqczb9wlxsmpwpqig6j1dmiblcfpq6mgnq8qwzcrvddm4cp47m5"))))
6898     (properties
6899      `((upstream-name . "IRanges")))
6900     (build-system r-build-system)
6901     (propagated-inputs
6902      `(("r-biocgenerics" ,r-biocgenerics)
6903        ("r-s4vectors" ,r-s4vectors)))
6904     (home-page "https://bioconductor.org/packages/IRanges")
6905     (synopsis "Infrastructure for manipulating intervals on sequences")
6906     (description
6907      "This package provides efficient low-level and highly reusable S4 classes
6908 for storing ranges of integers, RLE vectors (Run-Length Encoding), and, more
6909 generally, data that can be organized sequentially (formally defined as
6910 @code{Vector} objects), as well as views on these @code{Vector} objects.
6911 Efficient list-like classes are also provided for storing big collections of
6912 instances of the basic classes.  All classes in the package use consistent
6913 naming and share the same rich and consistent \"Vector API\" as much as
6914 possible.")
6915     (license license:artistic2.0)))
6916
6917 (define-public r-genomeinfodbdata
6918   (package
6919     (name "r-genomeinfodbdata")
6920     (version "0.99.1")
6921     (source (origin
6922               (method url-fetch)
6923               ;; We cannot use bioconductor-uri here because this tarball is
6924               ;; located under "data/annotation/" instead of "bioc/".
6925               (uri (string-append "https://bioconductor.org/packages/release/"
6926                                   "data/annotation/src/contrib/GenomeInfoDbData_"
6927                                   version ".tar.gz"))
6928               (sha256
6929                (base32
6930                 "0hipipvyvrh75n68hsjg35sxbcfzrghzxv547vnkk2f8ya99g01r"))))
6931     (properties
6932      `((upstream-name . "GenomeInfoDbData")))
6933     (build-system r-build-system)
6934     (home-page "https://bioconductor.org/packages/GenomeInfoDbData")
6935     (synopsis "Species and taxonomy ID look up tables for GenomeInfoDb")
6936     (description "This package contains data for mapping between NCBI taxonomy
6937 ID and species.  It is used by functions in the GenomeInfoDb package.")
6938     (license license:artistic2.0)))
6939
6940 (define-public r-genomeinfodb
6941   (package
6942     (name "r-genomeinfodb")
6943     (version "1.14.0")
6944     (source (origin
6945               (method url-fetch)
6946               (uri (bioconductor-uri "GenomeInfoDb" version))
6947               (sha256
6948                (base32
6949                 "1jhm0imkac4gvchbjxj408aakk39xdv2fyh818d3lk295bz6bnyp"))))
6950     (properties
6951      `((upstream-name . "GenomeInfoDb")))
6952     (build-system r-build-system)
6953     (propagated-inputs
6954      `(("r-biocgenerics" ,r-biocgenerics)
6955        ("r-genomeinfodbdata" ,r-genomeinfodbdata)
6956        ("r-iranges" ,r-iranges)
6957        ("r-rcurl" ,r-rcurl)
6958        ("r-s4vectors" ,r-s4vectors)))
6959     (home-page "https://bioconductor.org/packages/GenomeInfoDb")
6960     (synopsis "Utilities for manipulating chromosome identifiers")
6961     (description
6962      "This package contains data and functions that define and allow
6963 translation between different chromosome sequence naming conventions (e.g.,
6964 \"chr1\" versus \"1\"), including a function that attempts to place sequence
6965 names in their natural, rather than lexicographic, order.")
6966     (license license:artistic2.0)))
6967
6968 (define-public r-edger
6969   (package
6970     (name "r-edger")
6971     (version "3.20.8")
6972     (source (origin
6973               (method url-fetch)
6974               (uri (bioconductor-uri "edgeR" version))
6975               (sha256
6976                (base32
6977                 "0g7bj6w61blw3m22hw9rc01n554k9qkjizh7njr3j4shmhads58d"))))
6978     (properties `((upstream-name . "edgeR")))
6979     (build-system r-build-system)
6980     (propagated-inputs
6981      `(("r-limma" ,r-limma)
6982        ("r-locfit" ,r-locfit)
6983        ("r-rcpp" ,r-rcpp)
6984        ("r-statmod" ,r-statmod))) ;for estimateDisp
6985     (home-page "http://bioinf.wehi.edu.au/edgeR")
6986     (synopsis "EdgeR does empirical analysis of digital gene expression data")
6987     (description "This package can do differential expression analysis of
6988 RNA-seq expression profiles with biological replication.  It implements a range
6989 of statistical methodology based on the negative binomial distributions,
6990 including empirical Bayes estimation, exact tests, generalized linear models
6991 and quasi-likelihood tests.  It be applied to differential signal analysis of
6992 other types of genomic data that produce counts, including ChIP-seq, SAGE and
6993 CAGE.")
6994     (license license:gpl2+)))
6995
6996 (define-public r-variantannotation
6997   (package
6998     (name "r-variantannotation")
6999     (version "1.24.5")
7000     (source (origin
7001               (method url-fetch)
7002               (uri (bioconductor-uri "VariantAnnotation" version))
7003               (sha256
7004                (base32
7005                 "07ywn3c4w83l3sr76d0z3b1nv9icgdh3phsjlc6cfx7i6nfmvxw2"))))
7006     (properties
7007      `((upstream-name . "VariantAnnotation")))
7008     (inputs
7009      `(("zlib" ,zlib)))
7010     (propagated-inputs
7011      `(("r-annotationdbi" ,r-annotationdbi)
7012        ("r-biobase" ,r-biobase)
7013        ("r-biocgenerics" ,r-biocgenerics)
7014        ("r-biostrings" ,r-biostrings)
7015        ("r-bsgenome" ,r-bsgenome)
7016        ("r-dbi" ,r-dbi)
7017        ("r-genomeinfodb" ,r-genomeinfodb)
7018        ("r-genomicfeatures" ,r-genomicfeatures)
7019        ("r-genomicranges" ,r-genomicranges)
7020        ("r-iranges" ,r-iranges)
7021        ("r-summarizedexperiment" ,r-summarizedexperiment)
7022        ("r-rsamtools" ,r-rsamtools)
7023        ("r-rtracklayer" ,r-rtracklayer)
7024        ("r-s4vectors" ,r-s4vectors)
7025        ("r-xvector" ,r-xvector)
7026        ("r-zlibbioc" ,r-zlibbioc)))
7027     (build-system r-build-system)
7028     (home-page "https://bioconductor.org/packages/VariantAnnotation")
7029     (synopsis "Package for annotation of genetic variants")
7030     (description "This R package can annotate variants, compute amino acid
7031 coding changes and predict coding outcomes.")
7032     (license license:artistic2.0)))
7033
7034 (define-public r-limma
7035   (package
7036     (name "r-limma")
7037     (version "3.34.8")
7038     (source (origin
7039               (method url-fetch)
7040               (uri (bioconductor-uri "limma" version))
7041               (sha256
7042                (base32
7043                 "0bmxsgz2yll83sd3wbxsrsfd35468igb0d8lldym0d0lqfz906bw"))))
7044     (build-system r-build-system)
7045     (home-page "http://bioinf.wehi.edu.au/limma")
7046     (synopsis "Package for linear models for microarray and RNA-seq data")
7047     (description "This package can be used for the analysis of gene expression
7048 studies, especially the use of linear models for analysing designed experiments
7049 and the assessment of differential expression.  The analysis methods apply to
7050 different technologies, including microarrays, RNA-seq, and quantitative PCR.")
7051     (license license:gpl2+)))
7052
7053 (define-public r-xvector
7054   (package
7055     (name "r-xvector")
7056     (version "0.18.0")
7057     (source (origin
7058               (method url-fetch)
7059               (uri (bioconductor-uri "XVector" version))
7060               (sha256
7061                (base32
7062                 "1i4i3kdxr78lr1kcxq657p11ybi7kq10c8kyaqyh6gfc8i9rhvmk"))))
7063     (properties
7064      `((upstream-name . "XVector")))
7065     (build-system r-build-system)
7066     (arguments
7067      `(#:phases
7068        (modify-phases %standard-phases
7069          (add-after 'unpack 'use-system-zlib
7070            (lambda _
7071              (substitute* "DESCRIPTION"
7072                (("zlibbioc, ") ""))
7073              (substitute* "NAMESPACE"
7074                (("import\\(zlibbioc\\)") ""))
7075              #t)))))
7076     (inputs
7077      `(("zlib" ,zlib)))
7078     (propagated-inputs
7079      `(("r-biocgenerics" ,r-biocgenerics)
7080        ("r-iranges" ,r-iranges)
7081        ("r-s4vectors" ,r-s4vectors)))
7082     (home-page "https://bioconductor.org/packages/XVector")
7083     (synopsis "Representation and manpulation of external sequences")
7084     (description
7085      "This package provides memory efficient S4 classes for storing sequences
7086 \"externally\" (behind an R external pointer, or on disk).")
7087     (license license:artistic2.0)))
7088
7089 (define-public r-genomicranges
7090   (package
7091     (name "r-genomicranges")
7092     (version "1.30.2")
7093     (source (origin
7094               (method url-fetch)
7095               (uri (bioconductor-uri "GenomicRanges" version))
7096               (sha256
7097                (base32
7098                 "0c3r155603vb4zjs3adqa72770bh8karc11y8gl62l1m24jrkbnj"))))
7099     (properties
7100      `((upstream-name . "GenomicRanges")))
7101     (build-system r-build-system)
7102     (propagated-inputs
7103      `(("r-biocgenerics" ,r-biocgenerics)
7104        ("r-genomeinfodb" ,r-genomeinfodb)
7105        ("r-iranges" ,r-iranges)
7106        ("r-s4vectors" ,r-s4vectors)
7107        ("r-xvector" ,r-xvector)))
7108     (home-page "https://bioconductor.org/packages/GenomicRanges")
7109     (synopsis "Representation and manipulation of genomic intervals")
7110     (description
7111      "This package provides tools to efficiently represent and manipulate
7112 genomic annotations and alignments is playing a central role when it comes to
7113 analyzing high-throughput sequencing data (a.k.a. NGS data).  The
7114 GenomicRanges package defines general purpose containers for storing and
7115 manipulating genomic intervals and variables defined along a genome.")
7116     (license license:artistic2.0)))
7117
7118 (define-public r-biobase
7119   (package
7120     (name "r-biobase")
7121     (version "2.38.0")
7122     (source (origin
7123               (method url-fetch)
7124               (uri (bioconductor-uri "Biobase" version))
7125               (sha256
7126                (base32
7127                 "1cgm1ja1kp56zdlzyy9ggbkfn8r2vbsd4hncmz8g4hjd47fg18kg"))))
7128     (properties
7129      `((upstream-name . "Biobase")))
7130     (build-system r-build-system)
7131     (propagated-inputs
7132      `(("r-biocgenerics" ,r-biocgenerics)))
7133     (home-page "https://bioconductor.org/packages/Biobase")
7134     (synopsis "Base functions for Bioconductor")
7135     (description
7136      "This package provides functions that are needed by many other packages
7137 on Bioconductor or which replace R functions.")
7138     (license license:artistic2.0)))
7139
7140 (define-public r-annotationdbi
7141   (package
7142     (name "r-annotationdbi")
7143     (version "1.40.0")
7144     (source (origin
7145               (method url-fetch)
7146               (uri (bioconductor-uri "AnnotationDbi" version))
7147               (sha256
7148                (base32
7149                 "1dh4qs1a757n640gs34lf6z2glc96nan86x0sqaw5csadl2rhnlc"))))
7150     (properties
7151      `((upstream-name . "AnnotationDbi")))
7152     (build-system r-build-system)
7153     (propagated-inputs
7154      `(("r-biobase" ,r-biobase)
7155        ("r-biocgenerics" ,r-biocgenerics)
7156        ("r-dbi" ,r-dbi)
7157        ("r-iranges" ,r-iranges)
7158        ("r-rsqlite" ,r-rsqlite)
7159        ("r-s4vectors" ,r-s4vectors)))
7160     (home-page "https://bioconductor.org/packages/AnnotationDbi")
7161     (synopsis "Annotation database interface")
7162     (description
7163      "This package provides user interface and database connection code for
7164 annotation data packages using SQLite data storage.")
7165     (license license:artistic2.0)))
7166
7167 (define-public r-biomart
7168   (package
7169     (name "r-biomart")
7170     (version "2.34.2")
7171     (source (origin
7172               (method url-fetch)
7173               (uri (bioconductor-uri "biomaRt" version))
7174               (sha256
7175                (base32
7176                 "1zlgs2zg0lmnk572p55n7m34nkxka8w10x8f2ndssjkffl2csy79"))))
7177     (properties
7178      `((upstream-name . "biomaRt")))
7179     (build-system r-build-system)
7180     (propagated-inputs
7181      `(("r-annotationdbi" ,r-annotationdbi)
7182        ("r-httr" ,r-httr)
7183        ("r-progress" ,r-progress)
7184        ("r-rcurl" ,r-rcurl)
7185        ("r-stringr" ,r-stringr)
7186        ("r-xml" ,r-xml)))
7187     (home-page "https://bioconductor.org/packages/biomaRt")
7188     (synopsis "Interface to BioMart databases")
7189     (description
7190      "biomaRt provides an interface to a growing collection of databases
7191 implementing the @url{BioMart software suite, http://www.biomart.org}.  The
7192 package enables retrieval of large amounts of data in a uniform way without
7193 the need to know the underlying database schemas or write complex SQL queries.
7194 Examples of BioMart databases are Ensembl, COSMIC, Uniprot, HGNC, Gramene,
7195 Wormbase and dbSNP mapped to Ensembl.  These major databases give biomaRt
7196 users direct access to a diverse set of data and enable a wide range of
7197 powerful online queries from gene annotation to database mining.")
7198     (license license:artistic2.0)))
7199
7200 (define-public r-biocparallel
7201   (package
7202     (name "r-biocparallel")
7203     (version "1.12.0")
7204     (source (origin
7205               (method url-fetch)
7206               (uri (bioconductor-uri "BiocParallel" version))
7207               (sha256
7208                (base32
7209                 "13ng3n2wsgl3fh0v6jnz3vg51k5c1sh44pqdvblcrcd1qyjmmqhd"))))
7210     (properties
7211      `((upstream-name . "BiocParallel")))
7212     (build-system r-build-system)
7213     (propagated-inputs
7214      `(("r-futile-logger" ,r-futile-logger)
7215        ("r-snow" ,r-snow)
7216        ("r-bh" ,r-bh)))
7217     (home-page "https://bioconductor.org/packages/BiocParallel")
7218     (synopsis "Bioconductor facilities for parallel evaluation")
7219     (description
7220      "This package provides modified versions and novel implementation of
7221 functions for parallel evaluation, tailored to use with Bioconductor
7222 objects.")
7223     (license (list license:gpl2+ license:gpl3+))))
7224
7225 (define-public r-biostrings
7226   (package
7227     (name "r-biostrings")
7228     (version "2.46.0")
7229     (source (origin
7230               (method url-fetch)
7231               (uri (bioconductor-uri "Biostrings" version))
7232               (sha256
7233                (base32
7234                 "0vg50qdlxqcm2d6axjnzg8wh8pr4c5gz03l8bdl0llmwzp0zclzk"))))
7235     (properties
7236      `((upstream-name . "Biostrings")))
7237     (build-system r-build-system)
7238     (propagated-inputs
7239      `(("r-biocgenerics" ,r-biocgenerics)
7240        ("r-iranges" ,r-iranges)
7241        ("r-s4vectors" ,r-s4vectors)
7242        ("r-xvector" ,r-xvector)))
7243     (home-page "https://bioconductor.org/packages/Biostrings")
7244     (synopsis "String objects and algorithms for biological sequences")
7245     (description
7246      "This package provides memory efficient string containers, string
7247 matching algorithms, and other utilities, for fast manipulation of large
7248 biological sequences or sets of sequences.")
7249     (license license:artistic2.0)))
7250
7251 (define-public r-rsamtools
7252   (package
7253     (name "r-rsamtools")
7254     (version "1.30.0")
7255     (source (origin
7256               (method url-fetch)
7257               (uri (bioconductor-uri "Rsamtools" version))
7258               (sha256
7259                (base32
7260                 "0pjny5fjvbnfdyhl3bwxin678sha2drvs00sivxh3l772cn6yams"))))
7261     (properties
7262      `((upstream-name . "Rsamtools")))
7263     (build-system r-build-system)
7264     (arguments
7265      `(#:phases
7266        (modify-phases %standard-phases
7267          (add-after 'unpack 'use-system-zlib
7268            (lambda _
7269              (substitute* "DESCRIPTION"
7270                (("zlibbioc, ") ""))
7271              (substitute* "NAMESPACE"
7272                (("import\\(zlibbioc\\)") ""))
7273              #t)))))
7274     (inputs
7275      `(("zlib" ,zlib)))
7276     (propagated-inputs
7277      `(("r-biocgenerics" ,r-biocgenerics)
7278        ("r-biocparallel" ,r-biocparallel)
7279        ("r-biostrings" ,r-biostrings)
7280        ("r-bitops" ,r-bitops)
7281        ("r-genomeinfodb" ,r-genomeinfodb)
7282        ("r-genomicranges" ,r-genomicranges)
7283        ("r-iranges" ,r-iranges)
7284        ("r-s4vectors" ,r-s4vectors)
7285        ("r-xvector" ,r-xvector)))
7286     (home-page "https://bioconductor.org/packages/release/bioc/html/Rsamtools.html")
7287     (synopsis "Interface to samtools, bcftools, and tabix")
7288     (description
7289      "This package provides an interface to the 'samtools', 'bcftools', and
7290 'tabix' utilities for manipulating SAM (Sequence Alignment / Map), FASTA,
7291 binary variant call (BCF) and compressed indexed tab-delimited (tabix)
7292 files.")
7293     (license license:expat)))
7294
7295 (define-public r-delayedarray
7296   (package
7297     (name "r-delayedarray")
7298     (version "0.4.1")
7299     (source (origin
7300               (method url-fetch)
7301               (uri (bioconductor-uri "DelayedArray" version))
7302               (sha256
7303                (base32
7304                 "0s7h2giyvz04cg6248kbbzpwhxdrpnsvl2s8k5c8ricisd9aaz4b"))))
7305     (properties
7306      `((upstream-name . "DelayedArray")))
7307     (build-system r-build-system)
7308     (propagated-inputs
7309      `(("r-biocgenerics" ,r-biocgenerics)
7310        ("r-s4vectors" ,r-s4vectors)
7311        ("r-iranges" ,r-iranges)
7312        ("r-matrixstats" ,r-matrixstats)))
7313     (home-page "https://bioconductor.org/packages/DelayedArray")
7314     (synopsis "Delayed operations on array-like objects")
7315     (description
7316      "Wrapping an array-like object (typically an on-disk object) in a
7317 @code{DelayedArray} object allows one to perform common array operations on it
7318 without loading the object in memory.  In order to reduce memory usage and
7319 optimize performance, operations on the object are either delayed or executed
7320 using a block processing mechanism.  Note that this also works on in-memory
7321 array-like objects like @code{DataFrame} objects (typically with Rle columns),
7322 @code{Matrix} objects, and ordinary arrays and data frames.")
7323     (license license:artistic2.0)))
7324
7325 (define-public r-summarizedexperiment
7326   (package
7327     (name "r-summarizedexperiment")
7328     (version "1.8.1")
7329     (source (origin
7330               (method url-fetch)
7331               (uri (bioconductor-uri "SummarizedExperiment" version))
7332               (sha256
7333                (base32
7334                 "19vlwnby83fqjrilsxvnvgz0gvby7mrxvlmx18nb3p1w591ddfjh"))))
7335     (properties
7336      `((upstream-name . "SummarizedExperiment")))
7337     (build-system r-build-system)
7338     (propagated-inputs
7339      `(("r-biobase" ,r-biobase)
7340        ("r-biocgenerics" ,r-biocgenerics)
7341        ("r-delayedarray" ,r-delayedarray)
7342        ("r-genomeinfodb" ,r-genomeinfodb)
7343        ("r-genomicranges" ,r-genomicranges)
7344        ("r-iranges" ,r-iranges)
7345        ("r-matrix" ,r-matrix)
7346        ("r-s4vectors" ,r-s4vectors)))
7347     (home-page "https://bioconductor.org/packages/SummarizedExperiment")
7348     (synopsis "Container for representing genomic ranges by sample")
7349     (description
7350      "The SummarizedExperiment container contains one or more assays, each
7351 represented by a matrix-like object of numeric or other mode.  The rows
7352 typically represent genomic ranges of interest and the columns represent
7353 samples.")
7354     (license license:artistic2.0)))
7355
7356 (define-public r-genomicalignments
7357   (package
7358     (name "r-genomicalignments")
7359     (version "1.14.1")
7360     (source (origin
7361               (method url-fetch)
7362               (uri (bioconductor-uri "GenomicAlignments" version))
7363               (sha256
7364                (base32
7365                 "033p6fw46sn7w2yyn14nb9qcnkf30cl0nv6zh014ixflm3iifz39"))))
7366     (properties
7367      `((upstream-name . "GenomicAlignments")))
7368     (build-system r-build-system)
7369     (propagated-inputs
7370      `(("r-biocgenerics" ,r-biocgenerics)
7371        ("r-biocparallel" ,r-biocparallel)
7372        ("r-biostrings" ,r-biostrings)
7373        ("r-genomeinfodb" ,r-genomeinfodb)
7374        ("r-genomicranges" ,r-genomicranges)
7375        ("r-iranges" ,r-iranges)
7376        ("r-rsamtools" ,r-rsamtools)
7377        ("r-s4vectors" ,r-s4vectors)
7378        ("r-summarizedexperiment" ,r-summarizedexperiment)))
7379     (home-page "https://bioconductor.org/packages/GenomicAlignments")
7380     (synopsis "Representation and manipulation of short genomic alignments")
7381     (description
7382      "This package provides efficient containers for storing and manipulating
7383 short genomic alignments (typically obtained by aligning short reads to a
7384 reference genome).  This includes read counting, computing the coverage,
7385 junction detection, and working with the nucleotide content of the
7386 alignments.")
7387     (license license:artistic2.0)))
7388
7389 (define-public r-rtracklayer
7390   (package
7391     (name "r-rtracklayer")
7392     (version "1.38.3")
7393     (source (origin
7394               (method url-fetch)
7395               (uri (bioconductor-uri "rtracklayer" version))
7396               (sha256
7397                (base32
7398                 "1khzfczm35k5lq9h0jlqrq01192spzjyh8s6is89spj006flwn4k"))))
7399     (build-system r-build-system)
7400     (arguments
7401      `(#:phases
7402        (modify-phases %standard-phases
7403          (add-after 'unpack 'use-system-zlib
7404            (lambda _
7405              (substitute* "DESCRIPTION"
7406                ((" zlibbioc,") ""))
7407              (substitute* "NAMESPACE"
7408                (("import\\(zlibbioc\\)") ""))
7409              #t)))))
7410     (inputs
7411      `(("zlib" ,zlib)))
7412     (propagated-inputs
7413      `(("r-biocgenerics" ,r-biocgenerics)
7414        ("r-biostrings" ,r-biostrings)
7415        ("r-genomeinfodb" ,r-genomeinfodb)
7416        ("r-genomicalignments" ,r-genomicalignments)
7417        ("r-genomicranges" ,r-genomicranges)
7418        ("r-iranges" ,r-iranges)
7419        ("r-rcurl" ,r-rcurl)
7420        ("r-rsamtools" ,r-rsamtools)
7421        ("r-s4vectors" ,r-s4vectors)
7422        ("r-xml" ,r-xml)
7423        ("r-xvector" ,r-xvector)))
7424     (home-page "https://bioconductor.org/packages/rtracklayer")
7425     (synopsis "R interface to genome browsers and their annotation tracks")
7426     (description
7427      "rtracklayer is an extensible framework for interacting with multiple
7428 genome browsers (currently UCSC built-in) and manipulating annotation tracks
7429 in various formats (currently GFF, BED, bedGraph, BED15, WIG, BigWig and 2bit
7430 built-in).  The user may export/import tracks to/from the supported browsers,
7431 as well as query and modify the browser state, such as the current viewport.")
7432     (license license:artistic2.0)))
7433
7434 (define-public r-genomicfeatures
7435   (package
7436     (name "r-genomicfeatures")
7437     (version "1.30.3")
7438     (source (origin
7439               (method url-fetch)
7440               (uri (bioconductor-uri "GenomicFeatures" version))
7441               (sha256
7442                (base32
7443                 "010vn8hlwbnw12pd1d8pv6m12yp3xwx557gba5rbjq9p4qypnn3z"))))
7444     (properties
7445      `((upstream-name . "GenomicFeatures")))
7446     (build-system r-build-system)
7447     (propagated-inputs
7448      `(("r-annotationdbi" ,r-annotationdbi)
7449        ("r-biobase" ,r-biobase)
7450        ("r-biocgenerics" ,r-biocgenerics)
7451        ("r-biomart" ,r-biomart)
7452        ("r-biostrings" ,r-biostrings)
7453        ("r-dbi" ,r-dbi)
7454        ("r-genomeinfodb" ,r-genomeinfodb)
7455        ("r-genomicranges" ,r-genomicranges)
7456        ("r-iranges" ,r-iranges)
7457        ("r-rcurl" ,r-rcurl)
7458        ("r-rsqlite" ,r-rsqlite)
7459        ("r-rmysql" ,r-rmysql)
7460        ("r-rtracklayer" ,r-rtracklayer)
7461        ("r-s4vectors" ,r-s4vectors)
7462        ("r-xvector" ,r-xvector)))
7463     (home-page "https://bioconductor.org/packages/GenomicFeatures")
7464     (synopsis "Tools for working with transcript centric annotations")
7465     (description
7466      "This package provides a set of tools and methods for making and
7467 manipulating transcript centric annotations.  With these tools the user can
7468 easily download the genomic locations of the transcripts, exons and cds of a
7469 given organism, from either the UCSC Genome Browser or a BioMart
7470 database (more sources will be supported in the future).  This information is
7471 then stored in a local database that keeps track of the relationship between
7472 transcripts, exons, cds and genes.  Flexible methods are provided for
7473 extracting the desired features in a convenient format.")
7474     (license license:artistic2.0)))
7475
7476 (define-public r-go-db
7477   (package
7478     (name "r-go-db")
7479     (version "3.5.0")
7480     (source (origin
7481               (method url-fetch)
7482               (uri (string-append "https://www.bioconductor.org/packages/"
7483                                   "release/data/annotation/src/contrib/GO.db_"
7484                                   version ".tar.gz"))
7485               (sha256
7486                (base32
7487                 "02d1mn1al3q7qvhx1ylrr3ar4w4iw0qyi5d89v2336rzwk9maq35"))))
7488     (properties
7489      `((upstream-name . "GO.db")))
7490     (build-system r-build-system)
7491     (propagated-inputs
7492      `(("r-annotationdbi" ,r-annotationdbi)))
7493     (home-page "https://bioconductor.org/packages/GO.db")
7494     (synopsis "Annotation maps describing the entire Gene Ontology")
7495     (description
7496      "The purpose of this GO.db annotation package is to provide detailed
7497 information about the latest version of the Gene Ontologies.")
7498     (license license:artistic2.0)))
7499
7500 (define-public r-graph
7501   (package
7502     (name "r-graph")
7503     (version "1.56.0")
7504     (source (origin
7505               (method url-fetch)
7506               (uri (bioconductor-uri "graph" version))
7507               (sha256
7508                (base32
7509                 "15aajjp8h2z14p80c8hyd4rrmr9vqsm7bvwb989jxjl4k6g52an1"))))
7510     (build-system r-build-system)
7511     (propagated-inputs
7512      `(("r-biocgenerics" ,r-biocgenerics)))
7513     (home-page "https://bioconductor.org/packages/graph")
7514     (synopsis "Handle graph data structures in R")
7515     (description
7516      "This package implements some simple graph handling capabilities for R.")
7517     (license license:artistic2.0)))
7518
7519 (define-public r-topgo
7520   (package
7521     (name "r-topgo")
7522     (version "2.30.0")
7523     (source (origin
7524               (method url-fetch)
7525               (uri (bioconductor-uri "topGO" version))
7526               (sha256
7527                (base32
7528                 "1hqffz5qp7glxdvjp37005g8qk5nam3f9wpf6d1wjnzpar04f3dz"))))
7529     (properties
7530      `((upstream-name . "topGO")))
7531     (build-system r-build-system)
7532     (propagated-inputs
7533      `(("r-annotationdbi" ,r-annotationdbi)
7534        ("r-dbi" ,r-dbi)
7535        ("r-biobase" ,r-biobase)
7536        ("r-biocgenerics" ,r-biocgenerics)
7537        ("r-go-db" ,r-go-db)
7538        ("r-graph" ,r-graph)
7539        ("r-lattice" ,r-lattice)
7540        ("r-matrixstats" ,r-matrixstats)
7541        ("r-sparsem" ,r-sparsem)))
7542     (home-page "https://bioconductor.org/packages/topGO")
7543     (synopsis "Enrichment analysis for gene ontology")
7544     (description
7545      "The topGO package provides tools for testing @dfn{gene ontology} (GO)
7546 terms while accounting for the topology of the GO graph.  Different test
7547 statistics and different methods for eliminating local similarities and
7548 dependencies between GO terms can be implemented and applied.")
7549     ;; Any version of the LGPL applies.
7550     (license license:lgpl2.1+)))
7551
7552 (define-public r-bsgenome
7553   (package
7554     (name "r-bsgenome")
7555     (version "1.46.0")
7556     (source (origin
7557               (method url-fetch)
7558               (uri (bioconductor-uri "BSgenome" version))
7559               (sha256
7560                (base32
7561                 "1jbzq7lm2iajajn2bifxnkss0k9fdvgqr30mral17cbhp5f6w4lq"))))
7562     (properties
7563      `((upstream-name . "BSgenome")))
7564     (build-system r-build-system)
7565     (propagated-inputs
7566      `(("r-biocgenerics" ,r-biocgenerics)
7567        ("r-biostrings" ,r-biostrings)
7568        ("r-genomeinfodb" ,r-genomeinfodb)
7569        ("r-genomicranges" ,r-genomicranges)
7570        ("r-iranges" ,r-iranges)
7571        ("r-rsamtools" ,r-rsamtools)
7572        ("r-rtracklayer" ,r-rtracklayer)
7573        ("r-s4vectors" ,r-s4vectors)
7574        ("r-xvector" ,r-xvector)))
7575     (home-page "https://bioconductor.org/packages/BSgenome")
7576     (synopsis "Infrastructure for Biostrings-based genome data packages")
7577     (description
7578      "This package provides infrastructure shared by all Biostrings-based
7579 genome data packages and support for efficient SNP representation.")
7580     (license license:artistic2.0)))
7581
7582 (define-public r-bsgenome-hsapiens-1000genomes-hs37d5
7583   (package
7584     (name "r-bsgenome-hsapiens-1000genomes-hs37d5")
7585     (version "0.99.1")
7586     (source (origin
7587               (method url-fetch)
7588               ;; We cannot use bioconductor-uri here because this tarball is
7589               ;; located under "data/annotation/" instead of "bioc/".
7590               (uri (string-append "https://www.bioconductor.org/packages/"
7591                                   "release/data/annotation/src/contrib/"
7592                                   "BSgenome.Hsapiens.1000genomes.hs37d5_"
7593                                   version ".tar.gz"))
7594               (sha256
7595                (base32
7596                 "1cg0g5fqmsvwyw2p9hp2yy4ilk21jkbbrnpgqvb5c36ihjwvc7sr"))))
7597     (properties
7598      `((upstream-name . "BSgenome.Hsapiens.1000genomes.hs37d5")))
7599     (build-system r-build-system)
7600     ;; As this package provides little more than a very large data file it
7601     ;; doesn't make sense to build substitutes.
7602     (arguments `(#:substitutable? #f))
7603     (propagated-inputs
7604      `(("r-bsgenome" ,r-bsgenome)))
7605     (home-page
7606      "https://www.bioconductor.org/packages/BSgenome.Hsapiens.1000genomes.hs37d5/")
7607     (synopsis "Full genome sequences for Homo sapiens")
7608     (description
7609      "This package provides full genome sequences for Homo sapiens from
7610 1000genomes phase2 reference genome sequence (hs37d5), based on NCBI GRCh37.")
7611     (license license:artistic2.0)))
7612
7613 (define-public r-impute
7614   (package
7615     (name "r-impute")
7616     (version "1.52.0")
7617     (source (origin
7618               (method url-fetch)
7619               (uri (bioconductor-uri "impute" version))
7620               (sha256
7621                (base32
7622                 "0b8r4swvyx3cjcc2ky8yn0ncpzlbi1pgfsn3wpbjmhh7sqrffm2n"))))
7623     (inputs
7624      `(("gfortran" ,gfortran)))
7625     (build-system r-build-system)
7626     (home-page "https://bioconductor.org/packages/impute")
7627     (synopsis "Imputation for microarray data")
7628     (description
7629      "This package provides a function to impute missing gene expression
7630 microarray data, using nearest neighbor averaging.")
7631     (license license:gpl2+)))
7632
7633 (define-public r-seqpattern
7634   (package
7635     (name "r-seqpattern")
7636     (version "1.10.0")
7637     (source (origin
7638               (method url-fetch)
7639               (uri (bioconductor-uri "seqPattern" version))
7640               (sha256
7641                (base32
7642                 "1kcm5w83q7w0v0vs7nyp4gq5z86c6n6pqy9zmyyhxcrns7f597pm"))))
7643     (properties
7644      `((upstream-name . "seqPattern")))
7645     (build-system r-build-system)
7646     (propagated-inputs
7647      `(("r-biostrings" ,r-biostrings)
7648        ("r-genomicranges" ,r-genomicranges)
7649        ("r-iranges" ,r-iranges)
7650        ("r-kernsmooth" ,r-kernsmooth)
7651        ("r-plotrix" ,r-plotrix)))
7652     (home-page "https://bioconductor.org/packages/seqPattern")
7653     (synopsis "Visualising oligonucleotide patterns and motif occurrences")
7654     (description
7655      "This package provides tools to visualize oligonucleotide patterns and
7656 sequence motif occurrences across a large set of sequences centred at a common
7657 reference point and sorted by a user defined feature.")
7658     (license license:gpl3+)))
7659
7660 (define-public r-genomation
7661   (package
7662     (name "r-genomation")
7663     (version "1.11.3")
7664     (source (origin
7665               (method url-fetch)
7666               (uri (bioconductor-uri "genomation" version))
7667               (sha256
7668                (base32
7669                 "1d2g1v6xhrf3gm86pv8ln22df5g6v6k6i4i39v4j82zn4apany6v"))))
7670     (build-system r-build-system)
7671     (propagated-inputs
7672      `(("r-biostrings" ,r-biostrings)
7673        ("r-bsgenome" ,r-bsgenome)
7674        ("r-data-table" ,r-data-table)
7675        ("r-genomeinfodb" ,r-genomeinfodb)
7676        ("r-genomicalignments" ,r-genomicalignments)
7677        ("r-genomicranges" ,r-genomicranges)
7678        ("r-ggplot2" ,r-ggplot2)
7679        ("r-gridbase" ,r-gridbase)
7680        ("r-impute" ,r-impute)
7681        ("r-iranges" ,r-iranges)
7682        ("r-matrixstats" ,r-matrixstats)
7683        ("r-plotrix" ,r-plotrix)
7684        ("r-plyr" ,r-plyr)
7685        ("r-rcpp" ,r-rcpp)
7686        ("r-readr" ,r-readr)
7687        ("r-reshape2" ,r-reshape2)
7688        ("r-rsamtools" ,r-rsamtools)
7689        ("r-rtracklayer" ,r-rtracklayer)
7690        ("r-runit" ,r-runit)
7691        ("r-s4vectors" ,r-s4vectors)
7692        ("r-seqpattern" ,r-seqpattern)))
7693     (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7694     (synopsis "Summary, annotation and visualization of genomic data")
7695     (description
7696      "This package provides a package for summary and annotation of genomic
7697 intervals.  Users can visualize and quantify genomic intervals over
7698 pre-defined functional regions, such as promoters, exons, introns, etc.  The
7699 genomic intervals represent regions with a defined chromosome position, which
7700 may be associated with a score, such as aligned reads from HT-seq experiments,
7701 TF binding sites, methylation scores, etc.  The package can use any tabular
7702 genomic feature data as long as it has minimal information on the locations of
7703 genomic intervals.  In addition, it can use BAM or BigWig files as input.")
7704     (license license:artistic2.0)))
7705
7706 (define-public r-genomationdata
7707   (package
7708     (name "r-genomationdata")
7709     (version "1.10.0")
7710     (source (origin
7711               (method url-fetch)
7712               ;; We cannot use bioconductor-uri here because this tarball is
7713               ;; located under "data/annotation/" instead of "bioc/".
7714               (uri (string-append "https://bioconductor.org/packages/"
7715                                   "release/data/experiment/src/contrib/"
7716                                   "genomationData_" version ".tar.gz"))
7717               (sha256
7718                (base32
7719                 "0h7g5x3kyb50qlblz5hc85lfm6n6f5nb68i146way3ggs04sqvla"))))
7720     (build-system r-build-system)
7721     ;; As this package provides little more than large data files, it doesn't
7722     ;; make sense to build substitutes.
7723     (arguments `(#:substitutable? #f))
7724     (native-inputs
7725      `(("r-knitr" ,r-knitr)))
7726     (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7727     (synopsis "Experimental data for use with the genomation package")
7728     (description
7729      "This package contains experimental genetic data for use with the
7730 genomation package.  Included are Chip Seq, Methylation and Cage data,
7731 downloaded from Encode.")
7732     (license license:gpl3+)))
7733
7734 (define-public r-org-hs-eg-db
7735   (package
7736     (name "r-org-hs-eg-db")
7737     (version "3.5.0")
7738     (source (origin
7739               (method url-fetch)
7740               ;; We cannot use bioconductor-uri here because this tarball is
7741               ;; located under "data/annotation/" instead of "bioc/".
7742               (uri (string-append "https://www.bioconductor.org/packages/"
7743                                   "release/data/annotation/src/contrib/"
7744                                   "org.Hs.eg.db_" version ".tar.gz"))
7745               (sha256
7746                (base32
7747                 "1v6wa5613cjq59xd7x1qz8lr9nb2abm9abl2cci1khrnrlpla927"))))
7748     (properties
7749      `((upstream-name . "org.Hs.eg.db")))
7750     (build-system r-build-system)
7751     (propagated-inputs
7752      `(("r-annotationdbi" ,r-annotationdbi)))
7753     (home-page "https://www.bioconductor.org/packages/org.Hs.eg.db/")
7754     (synopsis "Genome wide annotation for Human")
7755     (description
7756      "This package contains genome-wide annotations for Human, primarily based
7757 on mapping using Entrez Gene identifiers.")
7758     (license license:artistic2.0)))
7759
7760 (define-public r-org-ce-eg-db
7761   (package
7762     (name "r-org-ce-eg-db")
7763     (version "3.5.0")
7764     (source (origin
7765               (method url-fetch)
7766               ;; We cannot use bioconductor-uri here because this tarball is
7767               ;; located under "data/annotation/" instead of "bioc/".
7768               (uri (string-append "https://www.bioconductor.org/packages/"
7769                                   "release/data/annotation/src/contrib/"
7770                                   "org.Ce.eg.db_" version ".tar.gz"))
7771               (sha256
7772                (base32
7773                 "02ggchixlmzywhsbr0h2ms4dravv7m5964cjxqcjxqs16vjwlbk9"))))
7774     (properties
7775      `((upstream-name . "org.Ce.eg.db")))
7776     (build-system r-build-system)
7777     (propagated-inputs
7778      `(("r-annotationdbi" ,r-annotationdbi)))
7779     (home-page "https://www.bioconductor.org/packages/org.Ce.eg.db/")
7780     (synopsis "Genome wide annotation for Worm")
7781     (description
7782      "This package provides mappings from Entrez gene identifiers to various
7783 annotations for the genome of the model worm Caenorhabditis elegans.")
7784     (license license:artistic2.0)))
7785
7786 (define-public r-org-dm-eg-db
7787   (package
7788     (name "r-org-dm-eg-db")
7789     (version "3.5.0")
7790     (source (origin
7791               (method url-fetch)
7792               ;; We cannot use bioconductor-uri here because this tarball is
7793               ;; located under "data/annotation/" instead of "bioc/".
7794               (uri (string-append "https://www.bioconductor.org/packages/"
7795                                   "release/data/annotation/src/contrib/"
7796                                   "org.Dm.eg.db_" version ".tar.gz"))
7797               (sha256
7798                (base32
7799                 "033qak1d3wwz17va0bh8z8p8arx0aw2va6gm1qfwsvdkj9cd9d7d"))))
7800     (properties
7801      `((upstream-name . "org.Dm.eg.db")))
7802     (build-system r-build-system)
7803     (propagated-inputs
7804      `(("r-annotationdbi" ,r-annotationdbi)))
7805     (home-page "https://www.bioconductor.org/packages/org.Dm.eg.db/")
7806     (synopsis "Genome wide annotation for Fly")
7807     (description
7808      "This package provides mappings from Entrez gene identifiers to various
7809 annotations for the genome of the model fruit fly Drosophila melanogaster.")
7810     (license license:artistic2.0)))
7811
7812 (define-public r-org-mm-eg-db
7813   (package
7814     (name "r-org-mm-eg-db")
7815     (version "3.5.0")
7816     (source (origin
7817               (method url-fetch)
7818               ;; We cannot use bioconductor-uri here because this tarball is
7819               ;; located under "data/annotation/" instead of "bioc/".
7820               (uri (string-append "https://www.bioconductor.org/packages/"
7821                                   "release/data/annotation/src/contrib/"
7822                                   "org.Mm.eg.db_" version ".tar.gz"))
7823               (sha256
7824                (base32
7825                 "11q21p3ki4bn4hb3aix0g775l45l66jmas6m94nfhqqnpjhv4d6g"))))
7826     (properties
7827      `((upstream-name . "org.Mm.eg.db")))
7828     (build-system r-build-system)
7829     (propagated-inputs
7830      `(("r-annotationdbi" ,r-annotationdbi)))
7831     (home-page "https://www.bioconductor.org/packages/org.Mm.eg.db/")
7832     (synopsis "Genome wide annotation for Mouse")
7833     (description
7834      "This package provides mappings from Entrez gene identifiers to various
7835 annotations for the genome of the model mouse Mus musculus.")
7836     (license license:artistic2.0)))
7837
7838 (define-public r-seqlogo
7839   (package
7840     (name "r-seqlogo")
7841     (version "1.44.0")
7842     (source
7843      (origin
7844        (method url-fetch)
7845        (uri (bioconductor-uri "seqLogo" version))
7846        (sha256
7847         (base32
7848          "1ql4q4vx0j61a893dqc3c8zxmgs8sqhy3j1qhyfdvbd01vw9w1kq"))))
7849     (properties `((upstream-name . "seqLogo")))
7850     (build-system r-build-system)
7851     (home-page "https://bioconductor.org/packages/seqLogo")
7852     (synopsis "Sequence logos for DNA sequence alignments")
7853     (description
7854      "seqLogo takes the position weight matrix of a DNA sequence motif and
7855 plots the corresponding sequence logo as introduced by Schneider and
7856 Stephens (1990).")
7857     (license license:lgpl2.0+)))
7858
7859 (define-public r-bsgenome-hsapiens-ucsc-hg19
7860   (package
7861     (name "r-bsgenome-hsapiens-ucsc-hg19")
7862     (version "1.4.0")
7863     (source (origin
7864               (method url-fetch)
7865               ;; We cannot use bioconductor-uri here because this tarball is
7866               ;; located under "data/annotation/" instead of "bioc/".
7867               (uri (string-append "https://www.bioconductor.org/packages/"
7868                                   "release/data/annotation/src/contrib/"
7869                                   "BSgenome.Hsapiens.UCSC.hg19_"
7870                                   version ".tar.gz"))
7871               (sha256
7872                (base32
7873                 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
7874     (properties
7875      `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
7876     (build-system r-build-system)
7877     ;; As this package provides little more than a very large data file it
7878     ;; doesn't make sense to build substitutes.
7879     (arguments `(#:substitutable? #f))
7880     (propagated-inputs
7881      `(("r-bsgenome" ,r-bsgenome)))
7882     (home-page
7883      "https://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
7884     (synopsis "Full genome sequences for Homo sapiens")
7885     (description
7886      "This package provides full genome sequences for Homo sapiens as provided
7887 by UCSC (hg19, February 2009) and stored in Biostrings objects.")
7888     (license license:artistic2.0)))
7889
7890 (define-public r-bsgenome-mmusculus-ucsc-mm9
7891   (package
7892     (name "r-bsgenome-mmusculus-ucsc-mm9")
7893     (version "1.4.0")
7894     (source (origin
7895               (method url-fetch)
7896               ;; We cannot use bioconductor-uri here because this tarball is
7897               ;; located under "data/annotation/" instead of "bioc/".
7898               (uri (string-append "https://www.bioconductor.org/packages/"
7899                                   "release/data/annotation/src/contrib/"
7900                                   "BSgenome.Mmusculus.UCSC.mm9_"
7901                                   version ".tar.gz"))
7902               (sha256
7903                (base32
7904                 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
7905     (properties
7906      `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
7907     (build-system r-build-system)
7908     ;; As this package provides little more than a very large data file it
7909     ;; doesn't make sense to build substitutes.
7910     (arguments `(#:substitutable? #f))
7911     (propagated-inputs
7912      `(("r-bsgenome" ,r-bsgenome)))
7913     (home-page
7914      "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
7915     (synopsis "Full genome sequences for Mouse")
7916     (description
7917      "This package provides full genome sequences for Mus musculus (Mouse) as
7918 provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
7919     (license license:artistic2.0)))
7920
7921 (define-public r-bsgenome-mmusculus-ucsc-mm10
7922   (package
7923     (name "r-bsgenome-mmusculus-ucsc-mm10")
7924     (version "1.4.0")
7925     (source (origin
7926               (method url-fetch)
7927               ;; We cannot use bioconductor-uri here because this tarball is
7928               ;; located under "data/annotation/" instead of "bioc/".
7929               (uri (string-append "https://www.bioconductor.org/packages/"
7930                                   "release/data/annotation/src/contrib/"
7931                                   "BSgenome.Mmusculus.UCSC.mm10_"
7932                                   version ".tar.gz"))
7933               (sha256
7934                (base32
7935                 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
7936     (properties
7937      `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
7938     (build-system r-build-system)
7939     ;; As this package provides little more than a very large data file it
7940     ;; doesn't make sense to build substitutes.
7941     (arguments `(#:substitutable? #f))
7942     (propagated-inputs
7943      `(("r-bsgenome" ,r-bsgenome)))
7944     (home-page
7945      "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
7946     (synopsis "Full genome sequences for Mouse")
7947     (description
7948      "This package provides full genome sequences for Mus
7949 musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
7950 in Biostrings objects.")
7951     (license license:artistic2.0)))
7952
7953 (define-public r-txdb-mmusculus-ucsc-mm10-knowngene
7954   (package
7955     (name "r-txdb-mmusculus-ucsc-mm10-knowngene")
7956     (version "3.4.0")
7957     (source (origin
7958               (method url-fetch)
7959               ;; We cannot use bioconductor-uri here because this tarball is
7960               ;; located under "data/annotation/" instead of "bioc/".
7961               (uri (string-append "https://www.bioconductor.org/packages/"
7962                                   "release/data/annotation/src/contrib/"
7963                                   "TxDb.Mmusculus.UCSC.mm10.knownGene_"
7964                                   version ".tar.gz"))
7965               (sha256
7966                (base32
7967                 "08gava9wsvpcqz51k2sni3pj03n5155v32d9riqbf305nbirqbkb"))))
7968     (properties
7969      `((upstream-name . "TxDb.Mmusculus.UCSC.mm10.knownGene")))
7970     (build-system r-build-system)
7971     ;; As this package provides little more than a very large data file it
7972     ;; doesn't make sense to build substitutes.
7973     (arguments `(#:substitutable? #f))
7974     (propagated-inputs
7975      `(("r-bsgenome" ,r-bsgenome)
7976        ("r-genomicfeatures" ,r-genomicfeatures)
7977        ("r-annotationdbi" ,r-annotationdbi)))
7978     (home-page
7979      "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm10.knownGene/")
7980     (synopsis "Annotation package for TxDb knownGene object(s) for Mouse")
7981     (description
7982      "This package loads a TxDb object, which is an R interface to
7983 prefabricated databases contained in this package.  This package provides
7984 the TxDb object of Mouse data as provided by UCSC (mm10, December 2011)
7985 based on the knownGene track.")
7986     (license license:artistic2.0)))
7987
7988 (define-public r-bsgenome-celegans-ucsc-ce6
7989   (package
7990     (name "r-bsgenome-celegans-ucsc-ce6")
7991     (version "1.4.0")
7992     (source (origin
7993               (method url-fetch)
7994               ;; We cannot use bioconductor-uri here because this tarball is
7995               ;; located under "data/annotation/" instead of "bioc/".
7996               (uri (string-append "https://www.bioconductor.org/packages/"
7997                                   "release/data/annotation/src/contrib/"
7998                                   "BSgenome.Celegans.UCSC.ce6_"
7999                                   version ".tar.gz"))
8000               (sha256
8001                (base32
8002                 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
8003     (properties
8004      `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
8005     (build-system r-build-system)
8006     ;; As this package provides little more than a very large data file it
8007     ;; doesn't make sense to build substitutes.
8008     (arguments `(#:substitutable? #f))
8009     (propagated-inputs
8010      `(("r-bsgenome" ,r-bsgenome)))
8011     (home-page
8012      "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
8013     (synopsis "Full genome sequences for Worm")
8014     (description
8015      "This package provides full genome sequences for Caenorhabditis
8016 elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
8017 objects.")
8018     (license license:artistic2.0)))
8019
8020 (define-public r-bsgenome-celegans-ucsc-ce10
8021   (package
8022     (name "r-bsgenome-celegans-ucsc-ce10")
8023     (version "1.4.0")
8024     (source (origin
8025               (method url-fetch)
8026               ;; We cannot use bioconductor-uri here because this tarball is
8027               ;; located under "data/annotation/" instead of "bioc/".
8028               (uri (string-append "https://www.bioconductor.org/packages/"
8029                                   "release/data/annotation/src/contrib/"
8030                                   "BSgenome.Celegans.UCSC.ce10_"
8031                                   version ".tar.gz"))
8032               (sha256
8033                (base32
8034                 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
8035     (properties
8036      `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
8037     (build-system r-build-system)
8038     ;; As this package provides little more than a very large data file it
8039     ;; doesn't make sense to build substitutes.
8040     (arguments `(#:substitutable? #f))
8041     (propagated-inputs
8042      `(("r-bsgenome" ,r-bsgenome)))
8043     (home-page
8044      "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
8045     (synopsis "Full genome sequences for Worm")
8046     (description
8047      "This package provides full genome sequences for Caenorhabditis
8048 elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
8049 objects.")
8050     (license license:artistic2.0)))
8051
8052 (define-public r-bsgenome-dmelanogaster-ucsc-dm3
8053   (package
8054     (name "r-bsgenome-dmelanogaster-ucsc-dm3")
8055     (version "1.4.0")
8056     (source (origin
8057               (method url-fetch)
8058               ;; We cannot use bioconductor-uri here because this tarball is
8059               ;; located under "data/annotation/" instead of "bioc/".
8060               (uri (string-append "https://www.bioconductor.org/packages/"
8061                                   "release/data/annotation/src/contrib/"
8062                                   "BSgenome.Dmelanogaster.UCSC.dm3_"
8063                                   version ".tar.gz"))
8064               (sha256
8065                (base32
8066                 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
8067     (properties
8068      `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
8069     (build-system r-build-system)
8070     ;; As this package provides little more than a very large data file it
8071     ;; doesn't make sense to build substitutes.
8072     (arguments `(#:substitutable? #f))
8073     (propagated-inputs
8074      `(("r-bsgenome" ,r-bsgenome)))
8075     (home-page
8076      "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
8077     (synopsis "Full genome sequences for Fly")
8078     (description
8079      "This package provides full genome sequences for Drosophila
8080 melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
8081 Biostrings objects.")
8082     (license license:artistic2.0)))
8083
8084 (define-public r-motifrg
8085   (package
8086     (name "r-motifrg")
8087     (version "1.22.0")
8088     (source
8089      (origin
8090        (method url-fetch)
8091        (uri (bioconductor-uri "motifRG" version))
8092        (sha256
8093         (base32
8094          "193zl2rlzwxv9p9q5i7rilj3w05ndqfyp9bdpvagp5s5cin4hf44"))))
8095     (properties `((upstream-name . "motifRG")))
8096     (build-system r-build-system)
8097     (propagated-inputs
8098      `(("r-biostrings" ,r-biostrings)
8099        ("r-bsgenome" ,r-bsgenome)
8100        ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8101        ("r-iranges" ,r-iranges)
8102        ("r-seqlogo" ,r-seqlogo)
8103        ("r-xvector" ,r-xvector)))
8104     (home-page "https://bioconductor.org/packages/motifRG")
8105     (synopsis "Discover motifs in high throughput sequencing data")
8106     (description
8107      "This package provides tools for discriminative motif discovery in high
8108 throughput genetic sequencing data sets using regression methods.")
8109     (license license:artistic2.0)))
8110
8111 (define-public r-qtl
8112  (package
8113   (name "r-qtl")
8114   (version "1.42-8")
8115   (source
8116    (origin
8117     (method url-fetch)
8118     (uri (string-append "mirror://cran/src/contrib/qtl_"
8119                         version ".tar.gz"))
8120     (sha256
8121      (base32
8122       "1l528dwvfpdlr05imrrm4rq32axp6hld9nqm6mm43kn5n7z2f5k6"))))
8123   (build-system r-build-system)
8124   (home-page "http://rqtl.org/")
8125   (synopsis "R package for analyzing QTL experiments in genetics")
8126   (description "R/qtl is an extension library for the R statistics
8127 system.  It is used to analyze experimental crosses for identifying
8128 genes contributing to variation in quantitative traits (so-called
8129 quantitative trait loci, QTLs).
8130
8131 Using a hidden Markov model, R/qtl allows to estimate genetic maps, to
8132 identify genotyping errors, and to perform single-QTL and two-QTL,
8133 two-dimensional genome scans.")
8134   (license license:gpl3)))
8135
8136 (define-public r-zlibbioc
8137   (package
8138     (name "r-zlibbioc")
8139     (version "1.24.0")
8140     (source (origin
8141               (method url-fetch)
8142               (uri (bioconductor-uri "zlibbioc" version))
8143               (sha256
8144                (base32
8145                 "1zr9hbh55hglfpy15cpxwmddxblhyb0an15953l3rbhmlh2vpy92"))))
8146     (properties
8147      `((upstream-name . "zlibbioc")))
8148     (build-system r-build-system)
8149     (home-page "https://bioconductor.org/packages/zlibbioc")
8150     (synopsis "Provider for zlib-1.2.5 to R packages")
8151     (description "This package uses the source code of zlib-1.2.5 to create
8152 libraries for systems that do not have these available via other means.")
8153     (license license:artistic2.0)))
8154
8155 (define-public r-r4rna
8156   (package
8157     (name "r-r4rna")
8158     (version "0.1.4")
8159     (source
8160      (origin
8161        (method url-fetch)
8162        (uri (string-append "http://www.e-rna.org/r-chie/files/R4RNA_"
8163                            version ".tar.gz"))
8164        (sha256
8165         (base32
8166          "1p0i78wh76jfgmn9jphbwwaz6yy6pipzfg08xs54cxavxg2j81p5"))))
8167     (build-system r-build-system)
8168     (propagated-inputs
8169      `(("r-optparse" ,r-optparse)
8170        ("r-rcolorbrewer" ,r-rcolorbrewer)))
8171     (home-page "http://www.e-rna.org/r-chie/index.cgi")
8172     (synopsis "Analysis framework for RNA secondary structure")
8173     (description
8174      "The R4RNA package aims to be a general framework for the analysis of RNA
8175 secondary structure and comparative analysis in R.")
8176     (license license:gpl3+)))
8177
8178 (define-public r-rhtslib
8179   (package
8180     (name "r-rhtslib")
8181     (version "1.10.0")
8182     (source
8183      (origin
8184        (method url-fetch)
8185        (uri (bioconductor-uri "Rhtslib" version))
8186        (sha256
8187         (base32
8188          "1dw3p44bfr0m7w39ckc2k37sjcp1zz0b9g12mr8am15jaj6v0q2j"))))
8189     (properties `((upstream-name . "Rhtslib")))
8190     (build-system r-build-system)
8191     (propagated-inputs
8192      `(("r-zlibbioc" ,r-zlibbioc)))
8193     (inputs
8194      `(("zlib" ,zlib)))
8195     (native-inputs
8196      `(("autoconf" ,autoconf)))
8197     (home-page "https://github.com/nhayden/Rhtslib")
8198     (synopsis "High-throughput sequencing library as an R package")
8199     (description
8200      "This package provides the HTSlib C library for high-throughput
8201 nucleotide sequence analysis.  The package is primarily useful to developers
8202 of other R packages who wish to make use of HTSlib.")
8203     (license license:lgpl2.0+)))
8204
8205 (define-public r-bamsignals
8206   (package
8207     (name "r-bamsignals")
8208     (version "1.10.0")
8209     (source
8210      (origin
8211        (method url-fetch)
8212        (uri (bioconductor-uri "bamsignals" version))
8213        (sha256
8214         (base32
8215          "15id6mkj95skb4kfafvfs2j7ylydal60c3pspcl7llhwpq6vcqvl"))))
8216     (build-system r-build-system)
8217     (propagated-inputs
8218      `(("r-biocgenerics" ,r-biocgenerics)
8219        ("r-genomicranges" ,r-genomicranges)
8220        ("r-iranges" ,r-iranges)
8221        ("r-rcpp" ,r-rcpp)
8222        ("r-rhtslib" ,r-rhtslib)
8223        ("r-zlibbioc" ,r-zlibbioc)))
8224     (inputs
8225      `(("zlib" ,zlib)))
8226     (home-page "https://bioconductor.org/packages/bamsignals")
8227     (synopsis "Extract read count signals from bam files")
8228     (description
8229      "This package allows to efficiently obtain count vectors from indexed bam
8230 files.  It counts the number of nucleotide sequence reads in given genomic
8231 ranges and it computes reads profiles and coverage profiles.  It also handles
8232 paired-end data.")
8233     (license license:gpl2+)))
8234
8235 (define-public r-rcas
8236   (package
8237     (name "r-rcas")
8238     (version "1.3.4")
8239     (source (origin
8240               (method url-fetch)
8241               (uri (string-append "https://github.com/BIMSBbioinfo/RCAS/archive/v"
8242                                   version ".tar.gz"))
8243               (file-name (string-append name "-" version ".tar.gz"))
8244               (sha256
8245                (base32
8246                 "1qgc7vi6fpzl440yg7jhiycg5q336kd4pxqzx10yx2zcq3bq3msg"))))
8247     (build-system r-build-system)
8248     (native-inputs
8249      `(("r-knitr" ,r-knitr)
8250        ("r-testthat" ,r-testthat)
8251        ;; During vignette building knitr checks that "pandoc-citeproc"
8252        ;; is in the PATH.
8253        ("ghc-pandoc-citeproc" ,ghc-pandoc-citeproc)))
8254     (propagated-inputs
8255      `(("r-data-table" ,r-data-table)
8256        ("r-biomart" ,r-biomart)
8257        ("r-org-hs-eg-db" ,r-org-hs-eg-db)
8258        ("r-org-ce-eg-db" ,r-org-ce-eg-db)
8259        ("r-org-dm-eg-db" ,r-org-dm-eg-db)
8260        ("r-org-mm-eg-db" ,r-org-mm-eg-db)
8261        ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8262        ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
8263        ("r-bsgenome-celegans-ucsc-ce10" ,r-bsgenome-celegans-ucsc-ce10)
8264        ("r-bsgenome-dmelanogaster-ucsc-dm3" ,r-bsgenome-dmelanogaster-ucsc-dm3)
8265        ("r-topgo" ,r-topgo)
8266        ("r-dt" ,r-dt)
8267        ("r-pbapply" ,r-pbapply)
8268        ("r-plotly" ,r-plotly)
8269        ("r-plotrix" ,r-plotrix)
8270        ("r-motifrg" ,r-motifrg)
8271        ("r-genomation" ,r-genomation)
8272        ("r-genomicfeatures" ,r-genomicfeatures)
8273        ("r-rtracklayer" ,r-rtracklayer)
8274        ("r-rmarkdown" ,r-rmarkdown)))
8275     (synopsis "RNA-centric annotation system")
8276     (description
8277      "RCAS aims to be a standalone RNA-centric annotation system that provides
8278 intuitive reports and publication-ready graphics.  This package provides the R
8279 library implementing most of the pipeline's features.")
8280     (home-page "https://github.com/BIMSBbioinfo/RCAS")
8281     (license license:artistic2.0)))
8282
8283 (define-public rcas-web
8284   (package
8285     (name "rcas-web")
8286     (version "0.0.4")
8287     (source
8288      (origin
8289        (method url-fetch)
8290        (uri (string-append "https://github.com/BIMSBbioinfo/rcas-web/"
8291                            "releases/download/v" version
8292                            "/rcas-web-" version ".tar.gz"))
8293        (sha256
8294         (base32
8295          "1p16frfys41a8yaa4gkm457nzkqhqs2pc3lkac0ds457w9w5j1gm"))))
8296     (build-system gnu-build-system)
8297     (arguments
8298      `(#:phases
8299        (modify-phases %standard-phases
8300          (add-after 'install 'wrap-executable
8301            (lambda* (#:key inputs outputs #:allow-other-keys)
8302              (let* ((out (assoc-ref outputs "out"))
8303                     (json   (assoc-ref inputs "guile-json"))
8304                     (redis  (assoc-ref inputs "guile-redis"))
8305                     (path   (string-append
8306                              json  "/share/guile/site/2.2:"
8307                              redis "/share/guile/site/2.2")))
8308                (wrap-program (string-append out "/bin/rcas-web")
8309                  `("GUILE_LOAD_PATH" ":" = (,path))
8310                  `("GUILE_LOAD_COMPILED_PATH" ":" = (,path))
8311                  `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
8312              #t)))))
8313     (inputs
8314      `(("r-minimal" ,r-minimal)
8315        ("r-rcas" ,r-rcas)
8316        ("guile-next" ,guile-2.2)
8317        ("guile-json" ,guile-json)
8318        ("guile-redis" ,guile2.2-redis)))
8319     (native-inputs
8320      `(("pkg-config" ,pkg-config)))
8321     (home-page "https://github.com/BIMSBbioinfo/rcas-web")
8322     (synopsis "Web interface for RNA-centric annotation system (RCAS)")
8323     (description "This package provides a simple web interface for the
8324 @dfn{RNA-centric annotation system} (RCAS).")
8325     (license license:agpl3+)))
8326
8327 (define-public r-mutationalpatterns
8328   (package
8329     (name "r-mutationalpatterns")
8330     (version "1.4.2")
8331     (source
8332      (origin
8333        (method url-fetch)
8334        (uri (bioconductor-uri "MutationalPatterns" version))
8335        (sha256
8336         (base32
8337          "08ay9h5cqsi8ypb6r0g4rfa5l1g06jgfzl64wmhgz134yqbl7vfv"))))
8338     (build-system r-build-system)
8339     (propagated-inputs
8340      `(("r-biocgenerics" ,r-biocgenerics)
8341        ("r-biostrings" ,r-biostrings)
8342        ("r-bsgenome-hsapiens-1000g" ,r-bsgenome-hsapiens-1000genomes-hs37d5)
8343        ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8344        ("r-genomicranges" ,r-genomicranges)
8345        ("r-genomeinfodb" ,r-genomeinfodb)
8346        ("r-ggplot2" ,r-ggplot2)
8347        ("r-gridextra" ,r-gridextra)
8348        ("r-iranges" ,r-iranges)
8349        ("r-nmf" ,r-nmf)
8350        ("r-plyr" ,r-plyr)
8351        ("r-pracma" ,r-pracma)
8352        ("r-reshape2" ,r-reshape2)
8353        ("r-cowplot" ,r-cowplot)
8354        ("r-ggdendro" ,r-ggdendro)
8355        ("r-s4vectors" ,r-s4vectors)
8356        ("r-summarizedexperiment" ,r-summarizedexperiment)
8357        ("r-variantannotation" ,r-variantannotation)))
8358     (home-page "https://bioconductor.org/packages/MutationalPatterns/")
8359     (synopsis "Extract and visualize mutational patterns in genomic data")
8360     (description "This package provides an extensive toolset for the
8361 characterization and visualization of a wide range of mutational patterns
8362 in SNV base substitution data.")
8363     (license license:expat)))
8364
8365 (define-public r-wgcna
8366   (package
8367     (name "r-wgcna")
8368     (version "1.62")
8369     (source
8370      (origin
8371        (method url-fetch)
8372        (uri (cran-uri "WGCNA" version))
8373        (sha256
8374         (base32
8375          "0c52rp09gqphz6g5x9jzkdcsyvfrknq0dkq9saslgy8q8ap974vx"))))
8376     (properties `((upstream-name . "WGCNA")))
8377     (build-system r-build-system)
8378     (propagated-inputs
8379      `(("r-annotationdbi" ,r-annotationdbi)
8380        ("r-doparallel" ,r-doparallel)
8381        ("r-dynamictreecut" ,r-dynamictreecut)
8382        ("r-fastcluster" ,r-fastcluster)
8383        ("r-foreach" ,r-foreach)
8384        ("r-go-db" ,r-go-db)
8385        ("r-hmisc" ,r-hmisc)
8386        ("r-impute" ,r-impute)
8387        ("r-rcpp" ,r-rcpp)
8388        ("r-robust" ,r-robust)
8389        ("r-survival" ,r-survival)
8390        ("r-matrixstats" ,r-matrixstats)
8391        ("r-preprocesscore" ,r-preprocesscore)))
8392     (home-page
8393      "http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/Rpackages/WGCNA/")
8394     (synopsis "Weighted correlation network analysis")
8395     (description
8396      "This package provides functions necessary to perform Weighted
8397 Correlation Network Analysis on high-dimensional data.  It includes functions
8398 for rudimentary data cleaning, construction and summarization of correlation
8399 networks, module identification and functions for relating both variables and
8400 modules to sample traits.  It also includes a number of utility functions for
8401 data manipulation and visualization.")
8402     (license license:gpl2+)))
8403
8404 (define-public r-chipkernels
8405   (let ((commit "c9cfcacb626b1221094fb3490ea7bac0fd625372")
8406         (revision "1"))
8407     (package
8408       (name "r-chipkernels")
8409       (version (string-append "1.1-" revision "." (string-take commit 9)))
8410       (source
8411        (origin
8412          (method git-fetch)
8413          (uri (git-reference
8414                (url "https://github.com/ManuSetty/ChIPKernels.git")
8415                (commit commit)))
8416          (file-name (string-append name "-" version))
8417          (sha256
8418           (base32
8419            "14bj5qhjm1hsm9ay561nfbqi9wxsa7y487df2idsaaf6z10nw4v0"))))
8420       (build-system r-build-system)
8421       (propagated-inputs
8422        `(("r-iranges" ,r-iranges)
8423          ("r-xvector" ,r-xvector)
8424          ("r-biostrings" ,r-biostrings)
8425          ("r-bsgenome" ,r-bsgenome)
8426          ("r-gtools" ,r-gtools)
8427          ("r-genomicranges" ,r-genomicranges)
8428          ("r-sfsmisc" ,r-sfsmisc)
8429          ("r-kernlab" ,r-kernlab)
8430          ("r-s4vectors" ,r-s4vectors)
8431          ("r-biocgenerics" ,r-biocgenerics)))
8432       (home-page "https://github.com/ManuSetty/ChIPKernels")
8433       (synopsis "Build string kernels for DNA Sequence analysis")
8434       (description "ChIPKernels is an R package for building different string
8435 kernels used for DNA Sequence analysis.  A dictionary of the desired kernel
8436 must be built and this dictionary can be used for determining kernels for DNA
8437 Sequences.")
8438       (license license:gpl2+))))
8439
8440 (define-public r-seqgl
8441   (package
8442     (name "r-seqgl")
8443     (version "1.1.4")
8444     (source
8445      (origin
8446        (method url-fetch)
8447        (uri (string-append "https://github.com/ManuSetty/SeqGL/"
8448                            "archive/" version ".tar.gz"))
8449        (file-name (string-append name "-" version ".tar.gz"))
8450        (sha256
8451         (base32
8452          "0pnk1p3sci5yipyc8xnb6jbmydpl80fld927xgnbcv104hy8h8yh"))))
8453     (build-system r-build-system)
8454     (propagated-inputs
8455      `(("r-biostrings" ,r-biostrings)
8456        ("r-chipkernels" ,r-chipkernels)
8457        ("r-genomicranges" ,r-genomicranges)
8458        ("r-spams" ,r-spams)
8459        ("r-wgcna" ,r-wgcna)
8460        ("r-fastcluster" ,r-fastcluster)))
8461     (home-page "https://github.com/ManuSetty/SeqGL")
8462     (synopsis "Group lasso for Dnase/ChIP-seq data")
8463     (description "SeqGL is a group lasso based algorithm to extract
8464 transcription factor sequence signals from ChIP, DNase and ATAC-seq profiles.
8465 This package presents a method which uses group lasso to discriminate between
8466 bound and non bound genomic regions to accurately identify transcription
8467 factors bound at the specific regions.")
8468     (license license:gpl2+)))
8469
8470 (define-public r-gkmsvm
8471   (package
8472     (name "r-gkmsvm")
8473     (version "0.79.0")
8474     (source
8475      (origin
8476        (method url-fetch)
8477        (uri (cran-uri "gkmSVM" version))
8478        (sha256
8479         (base32
8480          "04dakbgfvfalz4rm4fvvybp506dn5fbj5g86ybfhrc6wywjllsz3"))))
8481     (properties `((upstream-name . "gkmSVM")))
8482     (build-system r-build-system)
8483     (propagated-inputs
8484      `(("r-biocgenerics" ,r-biocgenerics)
8485        ("r-biostrings" ,r-biostrings)
8486        ("r-genomeinfodb" ,r-genomeinfodb)
8487        ("r-genomicranges" ,r-genomicranges)
8488        ("r-iranges" ,r-iranges)
8489        ("r-kernlab" ,r-kernlab)
8490        ("r-rcpp" ,r-rcpp)
8491        ("r-rocr" ,r-rocr)
8492        ("r-rtracklayer" ,r-rtracklayer)
8493        ("r-s4vectors" ,r-s4vectors)
8494        ("r-seqinr" ,r-seqinr)))
8495     (home-page "https://cran.r-project.org/web/packages/gkmSVM")
8496     (synopsis "Gapped-kmer support vector machine")
8497     (description
8498      "This R package provides tools for training gapped-kmer SVM classifiers
8499 for DNA and protein sequences.  This package supports several sequence
8500 kernels, including: gkmSVM, kmer-SVM, mismatch kernel and wildcard kernel.")
8501     (license license:gpl2+)))
8502
8503 (define-public r-tximport
8504   (package
8505     (name "r-tximport")
8506     (version "1.6.0")
8507     (source (origin
8508               (method url-fetch)
8509               (uri (bioconductor-uri "tximport" version))
8510               (sha256
8511                (base32
8512                 "1gyqcm91hxg1kgjqcz2qw1n56yp9pymjzs50rwcpb2893dr8sp2h"))))
8513     (build-system r-build-system)
8514     (home-page "https://bioconductor.org/packages/tximport")
8515     (synopsis "Import and summarize transcript-level estimates for gene-level analysis")
8516     (description
8517      "This package provides tools to import transcript-level abundance,
8518 estimated counts and transcript lengths, and to summarize them into matrices
8519 for use with downstream gene-level analysis packages.  Average transcript
8520 length, weighted by sample-specific transcript abundance estimates, is
8521 provided as a matrix which can be used as an offset for different expression
8522 of gene-level counts.")
8523     (license license:gpl2+)))
8524
8525 (define-public r-rhdf5
8526   (package
8527     (name "r-rhdf5")
8528     (version "2.22.0")
8529     (source (origin
8530               (method url-fetch)
8531               (uri (bioconductor-uri "rhdf5" version))
8532               (sha256
8533                (base32
8534                 "145858qg1xan6imxcbprzq3yn3mdf532aahdr6cibvdjg47hs4c1"))))
8535     (build-system r-build-system)
8536     (arguments
8537      `(#:phases
8538        (modify-phases %standard-phases
8539          (add-after 'unpack 'unpack-smallhdf5
8540            (lambda* (#:key outputs #:allow-other-keys)
8541              (system* "tar" "-xzvf"
8542                       "src/hdf5source/hdf5small.tgz" "-C" "src/" )
8543              (substitute* "src/hdf5/configure"
8544                (("/bin/mv") "mv"))
8545              #t)))))
8546     (propagated-inputs
8547      `(("r-zlibbioc" ,r-zlibbioc)))
8548     (inputs
8549      `(("perl" ,perl)
8550        ("zlib" ,zlib)))
8551     (home-page "https://bioconductor.org/packages/rhdf5")
8552     (synopsis "HDF5 interface to R")
8553     (description
8554      "This R/Bioconductor package provides an interface between HDF5 and R.
8555 HDF5's main features are the ability to store and access very large and/or
8556 complex datasets and a wide variety of metadata on mass storage (disk) through
8557 a completely portable file format.  The rhdf5 package is thus suited for the
8558 exchange of large and/or complex datasets between R and other software
8559 package, and for letting R applications work on datasets that are larger than
8560 the available RAM.")
8561     (license license:artistic2.0)))
8562
8563 (define-public r-annotationfilter
8564   (package
8565     (name "r-annotationfilter")
8566     (version "1.2.0")
8567     (source (origin
8568               (method url-fetch)
8569               (uri (bioconductor-uri "AnnotationFilter" version))
8570               (sha256
8571                (base32
8572                 "04zf864c1fvdlaay2r5cn30fc1n5i3czh31fs62qlrvs61wjiscs"))))
8573     (properties
8574      `((upstream-name . "AnnotationFilter")))
8575     (build-system r-build-system)
8576     (propagated-inputs
8577      `(("r-genomicranges" ,r-genomicranges)
8578        ("r-lazyeval" ,r-lazyeval)))
8579     (home-page "https://github.com/Bioconductor/AnnotationFilter")
8580     (synopsis "Facilities for filtering Bioconductor annotation resources")
8581     (description
8582      "This package provides classes and other infrastructure to implement
8583 filters for manipulating Bioconductor annotation resources.  The filters are
8584 used by @code{ensembldb}, @code{Organism.dplyr}, and other packages.")
8585     (license license:artistic2.0)))
8586
8587 (define-public emboss
8588   (package
8589     (name "emboss")
8590     (version "6.5.7")
8591     (source (origin
8592               (method url-fetch)
8593               (uri (string-append "ftp://emboss.open-bio.org/pub/EMBOSS/old/"
8594                                   (version-major+minor version) ".0/"
8595                                   "EMBOSS-" version ".tar.gz"))
8596               (sha256
8597                (base32
8598                 "0vsmz96gc411yj2iyzdrsmg4l2n1nhgmp7vrgzlxx3xixv9xbf0q"))))
8599     (build-system gnu-build-system)
8600     (arguments
8601      `(#:configure-flags
8602        (list (string-append "--with-hpdf="
8603                             (assoc-ref %build-inputs "libharu")))
8604        #:phases
8605        (modify-phases %standard-phases
8606          (add-after 'unpack 'fix-checks
8607            (lambda _
8608              ;; The PNGDRIVER tests check for the presence of libgd, libpng
8609              ;; and zlib, but assume that they are all found at the same
8610              ;; prefix.
8611              (substitute* "configure.in"
8612                (("CHECK_PNGDRIVER")
8613                 "LIBS=\"$LIBS -lgd -lpng -lz -lm\"
8614 AC_DEFINE([PLD_png], [1], [Define to 1 if PNG support is available])
8615 AM_CONDITIONAL(AMPNG, true)"))
8616              #t))
8617          (add-after 'fix-checks 'disable-update-check
8618            (lambda _
8619              ;; At build time there is no connection to the Internet, so
8620              ;; looking for updates will not work.
8621              (substitute* "Makefile.am"
8622                (("\\$\\(bindir\\)/embossupdate") ""))
8623              #t))
8624          (add-after 'disable-update-check 'autogen
8625            (lambda _ (zero? (system* "autoreconf" "-vif")))))))
8626     (inputs
8627      `(("perl" ,perl)
8628        ("libpng" ,libpng)
8629        ("gd" ,gd)
8630        ("libx11" ,libx11)
8631        ("libharu" ,libharu)
8632        ("zlib" ,zlib)))
8633     (native-inputs
8634      `(("autoconf" ,autoconf)
8635        ("automake" ,automake)
8636        ("libtool" ,libtool)
8637        ("pkg-config" ,pkg-config)))
8638     (home-page "http://emboss.sourceforge.net")
8639     (synopsis "Molecular biology analysis suite")
8640     (description "EMBOSS is the \"European Molecular Biology Open Software
8641 Suite\".  EMBOSS is an analysis package specially developed for the needs of
8642 the molecular biology (e.g. EMBnet) user community.  The software
8643 automatically copes with data in a variety of formats and even allows
8644 transparent retrieval of sequence data from the web.  It also provides a
8645 number of libraries for the development of software in the field of molecular
8646 biology.  EMBOSS also integrates a range of currently available packages and
8647 tools for sequence analysis into a seamless whole.")
8648     (license license:gpl2+)))
8649
8650 (define-public bits
8651   (let ((revision "1")
8652         (commit "3cc4567896d9d6442923da944beb704750a08d2d"))
8653     (package
8654       (name "bits")
8655       ;; The version is 2.13.0 even though no release archives have been
8656       ;; published as yet.
8657       (version (string-append "2.13.0-" revision "." (string-take commit 9)))
8658       (source (origin
8659                 (method git-fetch)
8660                 (uri (git-reference
8661                       (url "https://github.com/arq5x/bits.git")
8662                       (commit commit)))
8663                 (file-name (string-append name "-" version "-checkout"))
8664                 (sha256
8665                  (base32
8666                   "17n2kffk4kmhivd8c98g2vr6y1s23vbg4sxlxs689wni66797hbs"))))
8667       (build-system gnu-build-system)
8668       (arguments
8669        `(#:tests? #f ;no tests included
8670          #:phases
8671          (modify-phases %standard-phases
8672            (delete 'configure)
8673            (add-after 'unpack 'remove-cuda
8674              (lambda _
8675                (substitute* "Makefile"
8676                  ((".*_cuda") "")
8677                  (("(bits_test_intersections) \\\\" _ match) match))
8678                #t))
8679            (replace 'install
8680              (lambda* (#:key outputs #:allow-other-keys)
8681                (copy-recursively
8682                 "bin" (string-append (assoc-ref outputs "out") "/bin"))
8683                #t)))))
8684       (inputs
8685        `(("gsl" ,gsl)
8686          ("zlib" ,zlib)))
8687       (home-page "https://github.com/arq5x/bits")
8688       (synopsis "Implementation of binary interval search algorithm")
8689       (description "This package provides an implementation of the
8690 BITS (Binary Interval Search) algorithm, an approach to interval set
8691 intersection.  It is especially suited for the comparison of diverse genomic
8692 datasets and the exploration of large datasets of genome
8693 intervals (e.g. genes, sequence alignments).")
8694       (license license:gpl2))))
8695
8696 (define-public piranha
8697   ;; There is no release tarball for the latest version.  The latest commit is
8698   ;; older than one year at the time of this writing.
8699   (let ((revision "1")
8700         (commit   "0466d364b71117d01e4471b74c514436cc281233"))
8701     (package
8702       (name "piranha")
8703       (version (string-append "1.2.1-" revision "." (string-take commit 9)))
8704       (source (origin
8705                 (method git-fetch)
8706                 (uri (git-reference
8707                       (url "https://github.com/smithlabcode/piranha.git")
8708                       (commit commit)))
8709                 (sha256
8710                  (base32
8711                   "117dc0zf20c61jam69sk4abl57ah6yi6i7qra7d7y5zrbgk12q5n"))))
8712       (build-system gnu-build-system)
8713       (arguments
8714        `(#:test-target "test"
8715          #:phases
8716          (modify-phases %standard-phases
8717            (add-after 'unpack 'copy-smithlab-cpp
8718              (lambda* (#:key inputs #:allow-other-keys)
8719                (for-each (lambda (file)
8720                            (install-file file "./src/smithlab_cpp/"))
8721                          (find-files (assoc-ref inputs "smithlab-cpp")))
8722                #t))
8723            (add-after 'install 'install-to-store
8724              (lambda* (#:key outputs #:allow-other-keys)
8725                (let* ((out (assoc-ref outputs "out"))
8726                       (bin (string-append out "/bin")))
8727                  (for-each (lambda (file)
8728                              (install-file file bin))
8729                            (find-files "bin" ".*")))
8730                #t)))
8731          #:configure-flags
8732          (list (string-append "--with-bam_tools_headers="
8733                               (assoc-ref %build-inputs "bamtools") "/include/bamtools")
8734                (string-append "--with-bam_tools_library="
8735                               (assoc-ref %build-inputs "bamtools") "/lib/bamtools"))))
8736       (inputs
8737        `(("bamtools" ,bamtools)
8738          ("samtools" ,samtools-0.1)
8739          ("gsl" ,gsl)
8740          ("smithlab-cpp"
8741           ,(let ((commit "3723e2db438c51501d0423429ff396c3035ba46a"))
8742              (origin
8743                (method git-fetch)
8744                (uri (git-reference
8745                      (url "https://github.com/smithlabcode/smithlab_cpp.git")
8746                      (commit commit)))
8747                (file-name (string-append "smithlab_cpp-" commit "-checkout"))
8748                (sha256
8749                 (base32
8750                  "0l4gvbwslw5ngziskja41c00x1r06l3yidv7y0xw9djibhykzy0g")))))))
8751       (native-inputs
8752        `(("python" ,python-2)))
8753       (home-page "https://github.com/smithlabcode/piranha")
8754       (synopsis "Peak-caller for CLIP-seq and RIP-seq data")
8755       (description
8756        "Piranha is a peak-caller for genomic data produced by CLIP-seq and
8757 RIP-seq experiments.  It takes input in BED or BAM format and identifies
8758 regions of statistically significant read enrichment.  Additional covariates
8759 may optionally be provided to further inform the peak-calling process.")
8760       (license license:gpl3+))))
8761
8762 (define-public pepr
8763   (package
8764     (name "pepr")
8765     (version "1.0.9")
8766     (source (origin
8767               (method url-fetch)
8768               (uri (string-append "https://pypi.python.org/packages/source/P"
8769                                   "/PePr/PePr-" version ".tar.gz"))
8770               (sha256
8771                (base32
8772                 "0qxjfdpl1b1y53nccws2d85f6k74zwmx8y8sd9rszcqhfayx6gdx"))))
8773     (build-system python-build-system)
8774     (arguments
8775      `(#:python ,python-2 ; python2 only
8776        #:tests? #f)) ; no tests included
8777     (propagated-inputs
8778      `(("python2-numpy" ,python2-numpy)
8779        ("python2-scipy" ,python2-scipy)
8780        ("python2-pysam" ,python2-pysam)))
8781     (home-page "https://github.com/shawnzhangyx/PePr")
8782     (synopsis "Peak-calling and prioritization pipeline for ChIP-Seq data")
8783     (description
8784      "PePr is a ChIP-Seq peak calling or differential binding analysis tool
8785 that is primarily designed for data with biological replicates.  It uses a
8786 negative binomial distribution to model the read counts among the samples in
8787 the same group, and look for consistent differences between ChIP and control
8788 group or two ChIP groups run under different conditions.")
8789     (license license:gpl3+)))
8790
8791 (define-public filevercmp
8792   (let ((commit "1a9b779b93d0b244040274794d402106907b71b7"))
8793     (package
8794       (name "filevercmp")
8795       (version (string-append "0-1." (string-take commit 7)))
8796       (source (origin
8797         (method url-fetch)
8798         (uri (string-append "https://github.com/ekg/filevercmp/archive/"
8799                             commit ".tar.gz"))
8800         (file-name (string-append name "-" version ".tar.gz"))
8801         (sha256
8802          (base32 "0yp5jswf5j2pqc6517x277s4s6h1ss99v57kxw9gy0jkfl3yh450"))))
8803       (build-system gnu-build-system)
8804       (arguments
8805        `(#:tests? #f ; There are no tests to run.
8806          #:phases
8807          (modify-phases %standard-phases
8808            (delete 'configure) ; There is no configure phase.
8809            (replace 'install
8810              (lambda* (#:key outputs #:allow-other-keys)
8811                (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
8812                  (install-file "filevercmp" bin)))))))
8813       (home-page "https://github.com/ekg/filevercmp")
8814       (synopsis "This program compares version strings")
8815       (description "This program compares version strings.  It intends to be a
8816 replacement for strverscmp.")
8817       (license license:gpl3+))))
8818
8819 (define-public multiqc
8820   (package
8821     (name "multiqc")
8822     (version "1.4")
8823     (source
8824      (origin
8825        (method url-fetch)
8826        (uri (pypi-uri "multiqc" version))
8827        (sha256
8828         (base32
8829          "0ihx4rzmsfphv4byn05qv6f1y95g2dxs6viwziipl4wjk96acgm8"))))
8830     (build-system python-build-system)
8831     (propagated-inputs
8832      `(("python-jinja2" ,python-jinja2)
8833        ("python-simplejson" ,python-simplejson)
8834        ("python-pyyaml" ,python-pyyaml)
8835        ("python-click" ,python-click)
8836        ("python-spectra" ,python-spectra)
8837        ("python-requests" ,python-requests)
8838        ("python-markdown" ,python-markdown)
8839        ("python-lzstring" ,python-lzstring)
8840        ("python-matplotlib" ,python-matplotlib)
8841        ("python-numpy" ,python-numpy)
8842        ;; MultQC checks for the presence of nose at runtime.
8843        ("python-nose" ,python-nose)))
8844     (home-page "http://multiqc.info")
8845     (synopsis "Aggregate bioinformatics analysis reports")
8846     (description
8847      "MultiQC is a tool to aggregate bioinformatics results across many
8848 samples into a single report.  It contains modules for a large number of
8849 common bioinformatics tools.")
8850     (license license:gpl3+)))
8851
8852 (define-public r-chipseq
8853   (package
8854     (name "r-chipseq")
8855     (version "1.28.0")
8856     (source
8857      (origin
8858        (method url-fetch)
8859        (uri (bioconductor-uri "chipseq" version))
8860        (sha256
8861         (base32
8862          "1ymcq77krwjzrkzzcw7i9909cmkqa7c0675z9wzvrrk81hgdssfq"))))
8863     (build-system r-build-system)
8864     (propagated-inputs
8865      `(("r-biocgenerics" ,r-biocgenerics)
8866        ("r-genomicranges" ,r-genomicranges)
8867        ("r-iranges" ,r-iranges)
8868        ("r-lattice" ,r-lattice)
8869        ("r-s4vectors" ,r-s4vectors)
8870        ("r-shortread" ,r-shortread)))
8871     (home-page "https://bioconductor.org/packages/chipseq")
8872     (synopsis "Package for analyzing ChIPseq data")
8873     (description
8874      "This package provides tools for processing short read data from ChIPseq
8875 experiments.")
8876     (license license:artistic2.0)))
8877
8878 (define-public r-copyhelper
8879   (package
8880     (name "r-copyhelper")
8881     (version "1.6.0")
8882     (source
8883      (origin
8884        (method url-fetch)
8885        (uri (string-append "https://bioconductor.org/packages/release/"
8886                            "data/experiment/src/contrib/CopyhelpeR_"
8887                            version ".tar.gz"))
8888        (sha256
8889         (base32
8890          "0x7cyynjmxls9as2gg0iyp9x5fpalxmdjq914ss7i84i9zyk5bhq"))))
8891     (properties `((upstream-name . "CopyhelpeR")))
8892     (build-system r-build-system)
8893     (home-page "https://bioconductor.org/packages/CopyhelpeR/")
8894     (synopsis "Helper files for CopywriteR")
8895     (description
8896      "This package contains the helper files that are required to run the
8897 Bioconductor package CopywriteR.  It contains pre-assembled 1kb bin GC-content
8898 and mappability files for the reference genomes hg18, hg19, hg38, mm9 and
8899 mm10.  In addition, it contains a blacklist filter to remove regions that
8900 display copy number variation.  Files are stored as GRanges objects from the
8901 GenomicRanges Bioconductor package.")
8902     (license license:gpl2)))
8903
8904 (define-public r-copywriter
8905   (package
8906     (name "r-copywriter")
8907     (version "2.10.0")
8908     (source
8909      (origin
8910        (method url-fetch)
8911        (uri (bioconductor-uri "CopywriteR" version))
8912        (sha256
8913         (base32
8914          "17fy2lc5yf3nh6v077kv87h53n263hqz2540lzrl0vjiqrl2plca"))))
8915     (properties `((upstream-name . "CopywriteR")))
8916     (build-system r-build-system)
8917     (propagated-inputs
8918      `(("r-biocparallel" ,r-biocparallel)
8919        ("r-chipseq" ,r-chipseq)
8920        ("r-copyhelper" ,r-copyhelper)
8921        ("r-data-table" ,r-data-table)
8922        ("r-dnacopy" ,r-dnacopy)
8923        ("r-futile-logger" ,r-futile-logger)
8924        ("r-genomeinfodb" ,r-genomeinfodb)
8925        ("r-genomicalignments" ,r-genomicalignments)
8926        ("r-genomicranges" ,r-genomicranges)
8927        ("r-gtools" ,r-gtools)
8928        ("r-iranges" ,r-iranges)
8929        ("r-matrixstats" ,r-matrixstats)
8930        ("r-rsamtools" ,r-rsamtools)
8931        ("r-s4vectors" ,r-s4vectors)))
8932     (home-page "https://github.com/PeeperLab/CopywriteR")
8933     (synopsis "Copy number information from targeted sequencing")
8934     (description
8935      "CopywriteR extracts DNA copy number information from targeted sequencing
8936 by utilizing off-target reads.  It allows for extracting uniformly distributed
8937 copy number information, can be used without reference, and can be applied to
8938 sequencing data obtained from various techniques including chromatin
8939 immunoprecipitation and target enrichment on small gene panels.  Thereby,
8940 CopywriteR constitutes a widely applicable alternative to available copy
8941 number detection tools.")
8942     (license license:gpl2)))
8943
8944 (define-public r-methylkit
8945   (package
8946     (name "r-methylkit")
8947     (version "1.4.1")
8948     (source (origin
8949               (method url-fetch)
8950               (uri (bioconductor-uri "methylKit" version))
8951               (sha256
8952                (base32
8953                 "1k0nfn9318sgwm4z963bhnbp4c3zv85v3f9886vc5hgaisr0yvai"))))
8954     (properties `((upstream-name . "methylKit")))
8955     (build-system r-build-system)
8956     (propagated-inputs
8957      `(("r-data-table" ,r-data-table)
8958        ("r-emdbook" ,r-emdbook)
8959        ("r-fastseg" ,r-fastseg)
8960        ("r-genomeinfodb" ,r-genomeinfodb)
8961        ("r-genomicranges" ,r-genomicranges)
8962        ("r-gtools" ,r-gtools)
8963        ("r-iranges" ,r-iranges)
8964        ("r-kernsmooth" ,r-kernsmooth)
8965        ("r-limma" ,r-limma)
8966        ("r-mclust" ,r-mclust)
8967        ("r-qvalue" ,r-qvalue)
8968        ("r-r-utils" ,r-r-utils)
8969        ("r-rcpp" ,r-rcpp)
8970        ("r-rhtslib" ,r-rhtslib)
8971        ("r-rsamtools" ,r-rsamtools)
8972        ("r-rtracklayer" ,r-rtracklayer)
8973        ("r-s4vectors" ,r-s4vectors)
8974        ("r-zlibbioc" ,r-zlibbioc)))
8975     (inputs
8976      `(("zlib" ,zlib)))
8977     (home-page "https://github.com/al2na/methylKit")
8978     (synopsis
8979      "DNA methylation analysis from high-throughput bisulfite sequencing results")
8980     (description
8981      "MethylKit is an R package for DNA methylation analysis and annotation
8982 from high-throughput bisulfite sequencing.  The package is designed to deal
8983 with sequencing data from @dfn{Reduced representation bisulfite
8984 sequencing} (RRBS) and its variants, but also target-capture methods and whole
8985 genome bisulfite sequencing.  It also has functions to analyze base-pair
8986 resolution 5hmC data from experimental protocols such as oxBS-Seq and
8987 TAB-Seq.")
8988     (license license:artistic2.0)))
8989
8990 (define-public r-sva
8991   (package
8992     (name "r-sva")
8993     (version "3.26.0")
8994     (source
8995      (origin
8996        (method url-fetch)
8997        (uri (bioconductor-uri "sva" version))
8998        (sha256
8999         (base32
9000          "0q5xb68wfcnchy8rkv5ma67pmz1i91lsnvmwmj8f1c3w4xan3pgw"))))
9001     (build-system r-build-system)
9002     (propagated-inputs
9003      `(("r-genefilter" ,r-genefilter)
9004        ("r-mgcv" ,r-mgcv)
9005        ("r-biocparallel" ,r-biocparallel)
9006        ("r-matrixstats" ,r-matrixstats)
9007        ("r-limma" ,r-limma)))
9008     (home-page "https://bioconductor.org/packages/sva")
9009     (synopsis "Surrogate variable analysis")
9010     (description
9011      "This package contains functions for removing batch effects and other
9012 unwanted variation in high-throughput experiment.  It also contains functions
9013 for identifying and building surrogate variables for high-dimensional data
9014 sets.  Surrogate variables are covariates constructed directly from
9015 high-dimensional data like gene expression/RNA sequencing/methylation/brain
9016 imaging data that can be used in subsequent analyses to adjust for unknown,
9017 unmodeled, or latent sources of noise.")
9018     (license license:artistic2.0)))
9019
9020 (define-public r-seqminer
9021   (package
9022     (name "r-seqminer")
9023     (version "6.0")
9024     (source
9025      (origin
9026        (method url-fetch)
9027        (uri (cran-uri "seqminer" version))
9028        (sha256
9029         (base32
9030          "057j1l6dip35l1aivilapl2zv9db677b3di2pb3sfgq2sxg0ps3l"))))
9031     (build-system r-build-system)
9032     (inputs
9033      `(("zlib" ,zlib)))
9034     (home-page "http://seqminer.genomic.codes")
9035     (synopsis "Read nucleotide sequence data (VCF, BCF, and METAL formats)")
9036     (description
9037      "This package provides tools to integrate nucleotide sequencing
9038 data (variant call format, e.g. VCF or BCF) or meta-analysis results in R.")
9039     ;; Any version of the GPL is acceptable
9040     (license (list license:gpl2+ license:gpl3+))))
9041
9042 (define-public r-raremetals2
9043   (package
9044     (name "r-raremetals2")
9045     (version "0.1")
9046     (source
9047      (origin
9048        (method url-fetch)
9049        (uri (string-append "http://genome.sph.umich.edu/w/images/"
9050                            "b/b7/RareMETALS2_" version ".tar.gz"))
9051        (sha256
9052         (base32
9053          "0z5ljcgvnm06ja9lm85a3cniq7slxcy37aqqkxrdidr79an5fs4s"))))
9054     (properties `((upstream-name . "RareMETALS2")))
9055     (build-system r-build-system)
9056     (propagated-inputs
9057      `(("r-seqminer" ,r-seqminer)
9058        ("r-mvtnorm" ,r-mvtnorm)
9059        ("r-mass" ,r-mass)
9060        ("r-compquadform" ,r-compquadform)
9061        ("r-getopt" ,r-getopt)))
9062     (home-page "http://genome.sph.umich.edu/wiki/RareMETALS2")
9063     (synopsis "Analyze gene-level association tests for binary trait")
9064     (description
9065      "The R package rareMETALS2 is an extension of the R package rareMETALS.
9066 It was designed to meta-analyze gene-level association tests for binary trait.
9067 While rareMETALS offers a near-complete solution for meta-analysis of
9068 gene-level tests for quantitative trait, it does not offer the optimal
9069 solution for binary trait.  The package rareMETALS2 offers improved features
9070 for analyzing gene-level association tests in meta-analyses for binary
9071 trait.")
9072     (license license:gpl3)))
9073
9074 (define-public r-maldiquant
9075   (package
9076     (name "r-maldiquant")
9077     (version "1.17")
9078     (source
9079      (origin
9080        (method url-fetch)
9081        (uri (cran-uri "MALDIquant" version))
9082        (sha256
9083         (base32
9084          "047s6007ydc38x8wm027mlb4mngz15n0d4238fr8h43wyll5zy0z"))))
9085     (properties `((upstream-name . "MALDIquant")))
9086     (build-system r-build-system)
9087     (home-page "https://cran.r-project.org/web/packages/MALDIquant")
9088     (synopsis "Quantitative analysis of mass spectrometry data")
9089     (description
9090      "This package provides a complete analysis pipeline for matrix-assisted
9091 laser desorption/ionization-time-of-flight (MALDI-TOF) and other
9092 two-dimensional mass spectrometry data.  In addition to commonly used plotting
9093 and processing methods it includes distinctive features, namely baseline
9094 subtraction methods such as morphological filters (TopHat) or the
9095 statistics-sensitive non-linear iterative peak-clipping algorithm (SNIP), peak
9096 alignment using warping functions, handling of replicated measurements as well
9097 as allowing spectra with different resolutions.")
9098     (license license:gpl3+)))
9099
9100 (define-public r-protgenerics
9101   (package
9102     (name "r-protgenerics")
9103     (version "1.10.0")
9104     (source
9105      (origin
9106        (method url-fetch)
9107        (uri (bioconductor-uri "ProtGenerics" version))
9108        (sha256
9109         (base32
9110          "16ijp50448wnabp43klx943rhdvh7x45hvy7cnpq1s4dckxhhyni"))))
9111     (properties `((upstream-name . "ProtGenerics")))
9112     (build-system r-build-system)
9113     (home-page "https://github.com/lgatto/ProtGenerics")
9114     (synopsis "S4 generic functions for proteomics infrastructure")
9115     (description
9116      "This package provides S4 generic functions needed by Bioconductor
9117 proteomics packages.")
9118     (license license:artistic2.0)))
9119
9120 (define-public r-mzr
9121   (package
9122     (name "r-mzr")
9123     (version "2.12.0")
9124     (source
9125      (origin
9126        (method url-fetch)
9127        (uri (bioconductor-uri "mzR" version))
9128        (sha256
9129         (base32
9130          "1x3gp30sfxz2v3k3swih9kff9b2rvk7hzhnlkp6ywlnn2wgb0q8c"))
9131        (modules '((guix build utils)))
9132        (snippet
9133         '(begin
9134            (delete-file-recursively "src/boost")
9135            #t))))
9136     (properties `((upstream-name . "mzR")))
9137     (build-system r-build-system)
9138     (arguments
9139      `(#:phases
9140        (modify-phases %standard-phases
9141          (add-after 'unpack 'use-system-boost
9142            (lambda _
9143              (substitute* "src/Makevars"
9144                (("\\./boost/libs.*") "")
9145                (("ARCH_OBJS=" line)
9146                 (string-append line
9147                                "\nARCH_LIBS=-lboost_system -lboost_regex \
9148 -lboost_iostreams -lboost_thread -lboost_filesystem -lboost_chrono\n")))
9149              #t)))))
9150     (inputs
9151      `(("boost" ,boost) ; use this instead of the bundled boost sources
9152        ("netcdf" ,netcdf)))
9153     (propagated-inputs
9154      `(("r-biobase" ,r-biobase)
9155        ("r-biocgenerics" ,r-biocgenerics)
9156        ("r-protgenerics" ,r-protgenerics)
9157        ("r-rcpp" ,r-rcpp)
9158        ("r-zlibbioc" ,r-zlibbioc)))
9159     (home-page "https://github.com/sneumann/mzR/")
9160     (synopsis "Parser for mass spectrometry data files")
9161     (description
9162      "The mzR package provides a unified API to the common file formats and
9163 parsers available for mass spectrometry data.  It comes with a wrapper for the
9164 ISB random access parser for mass spectrometry mzXML, mzData and mzML files.
9165 The package contains the original code written by the ISB, and a subset of the
9166 proteowizard library for mzML and mzIdentML.  The netCDF reading code has
9167 previously been used in XCMS.")
9168     (license license:artistic2.0)))
9169
9170 (define-public r-affyio
9171   (package
9172     (name "r-affyio")
9173     (version "1.48.0")
9174     (source
9175      (origin
9176        (method url-fetch)
9177        (uri (bioconductor-uri "affyio" version))
9178        (sha256
9179         (base32
9180          "1pzzp3d3dbmyf34gvivfiprkpscn36rgvhrq853a1d3avcwr5ak9"))))
9181     (build-system r-build-system)
9182     (propagated-inputs
9183      `(("r-zlibbioc" ,r-zlibbioc)))
9184     (inputs
9185      `(("zlib" ,zlib)))
9186     (home-page "https://github.com/bmbolstad/affyio")
9187     (synopsis "Tools for parsing Affymetrix data files")
9188     (description
9189      "This package provides routines for parsing Affymetrix data files based
9190 upon file format information.  The primary focus is on accessing the CEL and
9191 CDF file formats.")
9192     (license license:lgpl2.0+)))
9193
9194 (define-public r-affy
9195   (package
9196     (name "r-affy")
9197     (version "1.56.0")
9198     (source
9199      (origin
9200        (method url-fetch)
9201        (uri (bioconductor-uri "affy" version))
9202        (sha256
9203         (base32
9204          "0jmbkimma5ffsdkk3xp03g4lpz84gd95nkqakif2nqq6wmx0syrj"))))
9205     (build-system r-build-system)
9206     (propagated-inputs
9207      `(("r-affyio" ,r-affyio)
9208        ("r-biobase" ,r-biobase)
9209        ("r-biocgenerics" ,r-biocgenerics)
9210        ("r-biocinstaller" ,r-biocinstaller)
9211        ("r-preprocesscore" ,r-preprocesscore)
9212        ("r-zlibbioc" ,r-zlibbioc)))
9213     (home-page "https://bioconductor.org/packages/affy")
9214     (synopsis "Methods for affymetrix oligonucleotide arrays")
9215     (description
9216      "This package contains functions for exploratory oligonucleotide array
9217 analysis.")
9218     (license license:lgpl2.0+)))
9219
9220 (define-public r-vsn
9221   (package
9222     (name "r-vsn")
9223     (version "3.46.0")
9224     (source
9225      (origin
9226        (method url-fetch)
9227        (uri (bioconductor-uri "vsn" version))
9228        (sha256
9229         (base32
9230          "18y62phzirj75gg6v5l41jwybmk23ia6w7qhch0kxc4bl2rysw6j"))))
9231     (build-system r-build-system)
9232     (propagated-inputs
9233      `(("r-affy" ,r-affy)
9234        ("r-biobase" ,r-biobase)
9235        ("r-ggplot2" ,r-ggplot2)
9236        ("r-lattice" ,r-lattice)
9237        ("r-limma" ,r-limma)))
9238     (home-page "https://bioconductor.org/packages/release/bioc/html/vsn.html")
9239     (synopsis "Variance stabilization and calibration for microarray data")
9240     (description
9241      "The package implements a method for normalising microarray intensities,
9242 and works for single- and multiple-color arrays.  It can also be used for data
9243 from other technologies, as long as they have similar format.  The method uses
9244 a robust variant of the maximum-likelihood estimator for an
9245 additive-multiplicative error model and affine calibration.  The model
9246 incorporates data calibration step (a.k.a.  normalization), a model for the
9247 dependence of the variance on the mean intensity and a variance stabilizing
9248 data transformation.  Differences between transformed intensities are
9249 analogous to \"normalized log-ratios\".  However, in contrast to the latter,
9250 their variance is independent of the mean, and they are usually more sensitive
9251 and specific in detecting differential transcription.")
9252     (license license:artistic2.0)))
9253
9254 (define-public r-mzid
9255   (package
9256     (name "r-mzid")
9257     (version "1.16.0")
9258     (source
9259      (origin
9260        (method url-fetch)
9261        (uri (bioconductor-uri "mzID" version))
9262        (sha256
9263         (base32
9264          "0yk70dka56zd8w62f03ggx3mandj91gfa767h9ajj0sd3mjmfqb9"))))
9265     (properties `((upstream-name . "mzID")))
9266     (build-system r-build-system)
9267     (propagated-inputs
9268      `(("r-doparallel" ,r-doparallel)
9269        ("r-foreach" ,r-foreach)
9270        ("r-iterators" ,r-iterators)
9271        ("r-plyr" ,r-plyr)
9272        ("r-protgenerics" ,r-protgenerics)
9273        ("r-rcpp" ,r-rcpp)
9274        ("r-xml" ,r-xml)))
9275     (home-page "https://bioconductor.org/packages/mzID")
9276     (synopsis "Parser for mzIdentML files")
9277     (description
9278      "This package provides a parser for mzIdentML files implemented using the
9279 XML package.  The parser tries to be general and able to handle all types of
9280 mzIdentML files with the drawback of having less pretty output than a vendor
9281 specific parser.")
9282     (license license:gpl2+)))
9283
9284 (define-public r-pcamethods
9285   (package
9286     (name "r-pcamethods")
9287     (version "1.70.0")
9288     (source
9289      (origin
9290        (method url-fetch)
9291        (uri (bioconductor-uri "pcaMethods" version))
9292        (sha256
9293         (base32
9294          "0ii235g0x0492kh8cfrf28ni0b6vd6fh7kizkqmczzqggd6b1bk8"))))
9295     (properties `((upstream-name . "pcaMethods")))
9296     (build-system r-build-system)
9297     (propagated-inputs
9298      `(("r-biobase" ,r-biobase)
9299        ("r-biocgenerics" ,r-biocgenerics)
9300        ("r-mass" ,r-mass)
9301        ("r-rcpp" ,r-rcpp)))
9302     (home-page "https://github.com/hredestig/pcamethods")
9303     (synopsis "Collection of PCA methods")
9304     (description
9305      "This package provides Bayesian PCA, Probabilistic PCA, Nipals PCA,
9306 Inverse Non-Linear PCA and the conventional SVD PCA.  A cluster based method
9307 for missing value estimation is included for comparison.  BPCA, PPCA and
9308 NipalsPCA may be used to perform PCA on incomplete data as well as for
9309 accurate missing value estimation.  A set of methods for printing and plotting
9310 the results is also provided.  All PCA methods make use of the same data
9311 structure (pcaRes) to provide a common interface to the PCA results.")
9312     (license license:gpl3+)))
9313
9314 (define-public r-msnbase
9315   (package
9316     (name "r-msnbase")
9317     (version "2.4.2")
9318     (source
9319      (origin
9320        (method url-fetch)
9321        (uri (bioconductor-uri "MSnbase" version))
9322        (sha256
9323         (base32
9324          "1ig64bf881p118dwqfr0ry41m7yhnyv165smv8fdwfv7sb6sagif"))))
9325     (properties `((upstream-name . "MSnbase")))
9326     (build-system r-build-system)
9327     (propagated-inputs
9328      `(("r-affy" ,r-affy)
9329        ("r-biobase" ,r-biobase)
9330        ("r-biocgenerics" ,r-biocgenerics)
9331        ("r-biocparallel" ,r-biocparallel)
9332        ("r-digest" ,r-digest)
9333        ("r-ggplot2" ,r-ggplot2)
9334        ("r-impute" ,r-impute)
9335        ("r-iranges" ,r-iranges)
9336        ("r-lattice" ,r-lattice)
9337        ("r-maldiquant" ,r-maldiquant)
9338        ("r-mzid" ,r-mzid)
9339        ("r-mzr" ,r-mzr)
9340        ("r-pcamethods" ,r-pcamethods)
9341        ("r-plyr" ,r-plyr)
9342        ("r-preprocesscore" ,r-preprocesscore)
9343        ("r-protgenerics" ,r-protgenerics)
9344        ("r-rcpp" ,r-rcpp)
9345        ("r-s4vectors" ,r-s4vectors)
9346        ("r-vsn" ,r-vsn)
9347        ("r-xml" ,r-xml)))
9348     (home-page "https://github.com/lgatto/MSnbase")
9349     (synopsis "Base functions and classes for MS-based proteomics")
9350     (description
9351      "This package provides basic plotting, data manipulation and processing
9352 of mass spectrometry based proteomics data.")
9353     (license license:artistic2.0)))
9354
9355 (define-public r-msnid
9356   (package
9357     (name "r-msnid")
9358     (version "1.12.1")
9359     (source
9360      (origin
9361        (method url-fetch)
9362        (uri (bioconductor-uri "MSnID" version))
9363        (sha256
9364         (base32
9365          "1zw508kk4f8brg69674wp18gqkpx2kpya5f6x9cl3qng7v4h5pxx"))))
9366     (properties `((upstream-name . "MSnID")))
9367     (build-system r-build-system)
9368     (propagated-inputs
9369      `(("r-biobase" ,r-biobase)
9370        ("r-data-table" ,r-data-table)
9371        ("r-doparallel" ,r-doparallel)
9372        ("r-dplyr" ,r-dplyr)
9373        ("r-foreach" ,r-foreach)
9374        ("r-iterators" ,r-iterators)
9375        ("r-msnbase" ,r-msnbase)
9376        ("r-mzid" ,r-mzid)
9377        ("r-mzr" ,r-mzr)
9378        ("r-protgenerics" ,r-protgenerics)
9379        ("r-r-cache" ,r-r-cache)
9380        ("r-rcpp" ,r-rcpp)
9381        ("r-reshape2" ,r-reshape2)))
9382     (home-page "https://bioconductor.org/packages/MSnID")
9383     (synopsis "Utilities for LC-MSn proteomics identifications")
9384     (description
9385      "This package extracts @dfn{tandem mass spectrometry} (MS/MS) ID data
9386 from mzIdentML (leveraging the mzID package) or text files.  After collating
9387 the search results from multiple datasets it assesses their identification
9388 quality and optimize filtering criteria to achieve the maximum number of
9389 identifications while not exceeding a specified false discovery rate.  It also
9390 contains a number of utilities to explore the MS/MS results and assess missed
9391 and irregular enzymatic cleavages, mass measurement accuracy, etc.")
9392     (license license:artistic2.0)))
9393
9394 (define-public r-seurat
9395   (package
9396     (name "r-seurat")
9397     (version "2.2.1")
9398     (source (origin
9399               (method url-fetch)
9400               (uri (cran-uri "Seurat" version))
9401               (sha256
9402                (base32
9403                 "1sr82nf38hq07avrfn8vlrzjq7dfm4pcr8l1nh6mnglcql2bk9z2"))
9404               ;; Delete pre-built jar.
9405               (snippet
9406                '(begin (delete-file "inst/java/ModularityOptimizer.jar")
9407                        #t))))
9408     (properties `((upstream-name . "Seurat")))
9409     (build-system r-build-system)
9410     (arguments
9411      `(#:phases
9412        (modify-phases %standard-phases
9413          (add-after 'unpack 'build-jar
9414            (lambda* (#:key inputs #:allow-other-keys)
9415              (let ((classesdir "tmp-classes"))
9416                (setenv "JAVA_HOME" (assoc-ref inputs "jdk"))
9417                (mkdir classesdir)
9418                (with-output-to-file "manifest"
9419                  (lambda _
9420                    (display "Manifest-Version: 1.0
9421 Main-Class: ModularityOptimizer\n")))
9422                (and (zero? (apply system* `("javac" "-d" ,classesdir
9423                                             ,@(find-files "java" "\\.java$"))))
9424                     (zero? (system* "jar"
9425                                     "-cmf" "manifest"
9426                                     "inst/java/ModularityOptimizer.jar"
9427                                     "-C" classesdir  ".")))))))))
9428     (native-inputs
9429      `(("jdk" ,icedtea "jdk")))
9430     (propagated-inputs
9431      `(("r-ape" ,r-ape)
9432        ("r-caret" ,r-caret)
9433        ("r-cowplot" ,r-cowplot)
9434        ("r-diffusionmap" ,r-diffusionmap)
9435        ("r-dplyr" ,r-dplyr)
9436        ("r-dtw" ,r-dtw)
9437        ("r-fnn" ,r-fnn)
9438        ("r-fpc" ,r-fpc)
9439        ("r-gdata" ,r-gdata)
9440        ("r-ggplot2" ,r-ggplot2)
9441        ("r-ggridges" ,r-ggridges)
9442        ("r-gplots" ,r-gplots)
9443        ("r-gridextra" ,r-gridextra)
9444        ("r-hmisc" ,r-hmisc)
9445        ("r-ica" ,r-ica)
9446        ("r-igraph" ,r-igraph)
9447        ("r-irlba" ,r-irlba)
9448        ("r-lars" ,r-lars)
9449        ("r-mass" ,r-mass)
9450        ("r-matrix" ,r-matrix)
9451        ("r-metap" ,r-metap)
9452        ("r-mixtools" ,r-mixtools)
9453        ("r-pbapply" ,r-pbapply)
9454        ("r-plotly" ,r-plotly)
9455        ("r-ranger" ,r-ranger)
9456        ("r-rcolorbrewer" ,r-rcolorbrewer)
9457        ("r-rcpp" ,r-rcpp)
9458        ("r-rcppeigen" ,r-rcppeigen)
9459        ("r-rcppprogress" ,r-rcppprogress)
9460        ("r-reshape2" ,r-reshape2)
9461        ("r-rocr" ,r-rocr)
9462        ("r-rtsne" ,r-rtsne)
9463        ("r-sdmtools" ,r-sdmtools)
9464        ("r-stringr" ,r-stringr)
9465        ("r-tclust" ,r-tclust)
9466        ("r-tidyr" ,r-tidyr)
9467        ("r-tsne" ,r-tsne)
9468        ("r-vgam" ,r-vgam)))
9469     (home-page "http://www.satijalab.org/seurat")
9470     (synopsis "Seurat is an R toolkit for single cell genomics")
9471     (description
9472      "This package is an R package designed for QC, analysis, and
9473 exploration of single cell RNA-seq data.  It easily enables widely-used
9474 analytical techniques, including the identification of highly variable genes,
9475 dimensionality reduction; PCA, ICA, t-SNE, standard unsupervised clustering
9476 algorithms; density clustering, hierarchical clustering, k-means, and the
9477 discovery of differentially expressed genes and markers.")
9478     (license license:gpl3)))
9479
9480 (define-public r-aroma-light
9481   (package
9482     (name "r-aroma-light")
9483     (version "3.8.0")
9484     (source
9485      (origin
9486        (method url-fetch)
9487        (uri (bioconductor-uri "aroma.light" version))
9488        (sha256
9489         (base32
9490          "0crnk6851jwypqr5l5jcbbay0vi5vvdjyisaf6z2d69c39wmr6sc"))))
9491     (properties `((upstream-name . "aroma.light")))
9492     (build-system r-build-system)
9493     (propagated-inputs
9494      `(("r-matrixstats" ,r-matrixstats)
9495        ("r-r-methodss3" ,r-r-methodss3)
9496        ("r-r-oo" ,r-r-oo)
9497        ("r-r-utils" ,r-r-utils)))
9498     (home-page "https://github.com/HenrikBengtsson/aroma.light")
9499     (synopsis "Methods for normalization and visualization of microarray data")
9500     (description
9501      "This package provides methods for microarray analysis that take basic
9502 data types such as matrices and lists of vectors.  These methods can be used
9503 standalone, be utilized in other packages, or be wrapped up in higher-level
9504 classes.")
9505     (license license:gpl2+)))
9506
9507 (define-public r-deseq
9508   (package
9509     (name "r-deseq")
9510     (version "1.30.0")
9511     (source
9512      (origin
9513        (method url-fetch)
9514        (uri (bioconductor-uri "DESeq" version))
9515        (sha256
9516         (base32
9517          "0mn5w3cy16iwwk8zxs7za6aa6cnrca75z0g45zd5zh1py5d7nfv9"))))
9518     (properties `((upstream-name . "DESeq")))
9519     (build-system r-build-system)
9520     (propagated-inputs
9521      `(("r-biobase" ,r-biobase)
9522        ("r-biocgenerics" ,r-biocgenerics)
9523        ("r-genefilter" ,r-genefilter)
9524        ("r-geneplotter" ,r-geneplotter)
9525        ("r-lattice" ,r-lattice)
9526        ("r-locfit" ,r-locfit)
9527        ("r-mass" ,r-mass)
9528        ("r-rcolorbrewer" ,r-rcolorbrewer)))
9529     (home-page "http://www-huber.embl.de/users/anders/DESeq")
9530     (synopsis "Differential gene expression analysis")
9531     (description
9532      "This package provides tools for estimating variance-mean dependence in
9533 count data from high-throughput genetic sequencing assays and for testing for
9534 differential expression based on a model using the negative binomial
9535 distribution.")
9536     (license license:gpl3+)))
9537
9538 (define-public r-edaseq
9539   (package
9540     (name "r-edaseq")
9541     (version "2.12.0")
9542     (source
9543      (origin
9544        (method url-fetch)
9545        (uri (bioconductor-uri "EDASeq" version))
9546        (sha256
9547         (base32
9548          "07zm89zcivyn2261aq9grqmly8ji482kr9h9dyfknfdfrpv7jpwv"))))
9549     (properties `((upstream-name . "EDASeq")))
9550     (build-system r-build-system)
9551     (propagated-inputs
9552      `(("r-annotationdbi" ,r-annotationdbi)
9553        ("r-aroma-light" ,r-aroma-light)
9554        ("r-biobase" ,r-biobase)
9555        ("r-biocgenerics" ,r-biocgenerics)
9556        ("r-biomart" ,r-biomart)
9557        ("r-biostrings" ,r-biostrings)
9558        ("r-deseq" ,r-deseq)
9559        ("r-genomicfeatures" ,r-genomicfeatures)
9560        ("r-genomicranges" ,r-genomicranges)
9561        ("r-iranges" ,r-iranges)
9562        ("r-rsamtools" ,r-rsamtools)
9563        ("r-shortread" ,r-shortread)))
9564     (home-page "https://github.com/drisso/EDASeq")
9565     (synopsis "Exploratory data analysis and normalization for RNA-Seq")
9566     (description
9567      "This package provides support for numerical and graphical summaries of
9568 RNA-Seq genomic read data.  Provided within-lane normalization procedures to
9569 adjust for GC-content effect (or other gene-level effects) on read counts:
9570 loess robust local regression, global-scaling, and full-quantile
9571 normalization.  Between-lane normalization procedures to adjust for
9572 distributional differences between lanes (e.g., sequencing depth):
9573 global-scaling and full-quantile normalization.")
9574     (license license:artistic2.0)))
9575
9576 (define-public r-interactivedisplaybase
9577   (package
9578     (name "r-interactivedisplaybase")
9579     (version "1.16.0")
9580     (source
9581      (origin
9582        (method url-fetch)
9583        (uri (bioconductor-uri "interactiveDisplayBase" version))
9584        (sha256
9585         (base32
9586          "01yb945jqqimwjgriza6yy4dnp303cdirxrhl4hjyprfdlmnz5p5"))))
9587     (properties
9588      `((upstream-name . "interactiveDisplayBase")))
9589     (build-system r-build-system)
9590     (propagated-inputs
9591      `(("r-biocgenerics" ,r-biocgenerics)
9592        ("r-shiny" ,r-shiny)))
9593     (home-page "https://bioconductor.org/packages/interactiveDisplayBase")
9594     (synopsis "Base package for web displays of Bioconductor objects")
9595     (description
9596      "This package contains the basic methods needed to generate interactive
9597 Shiny-based display methods for Bioconductor objects.")
9598     (license license:artistic2.0)))
9599
9600 (define-public r-annotationhub
9601   (package
9602     (name "r-annotationhub")
9603     (version "2.10.1")
9604     (source
9605      (origin
9606        (method url-fetch)
9607        (uri (bioconductor-uri "AnnotationHub" version))
9608        (sha256
9609         (base32
9610          "14v8g44a6zg9j2rwn9x9y8509k0wr2cw8yccliz24glplb40wva4"))))
9611     (properties `((upstream-name . "AnnotationHub")))
9612     (build-system r-build-system)
9613     (propagated-inputs
9614      `(("r-annotationdbi" ,r-annotationdbi)
9615        ("r-biocgenerics" ,r-biocgenerics)
9616        ("r-biocinstaller" ,r-biocinstaller)
9617        ("r-curl" ,r-curl)
9618        ("r-httr" ,r-httr)
9619        ("r-interactivedisplaybase" ,r-interactivedisplaybase)
9620        ("r-rsqlite" ,r-rsqlite)
9621        ("r-s4vectors" ,r-s4vectors)
9622        ("r-yaml" ,r-yaml)))
9623     (home-page "https://bioconductor.org/packages/AnnotationHub")
9624     (synopsis "Client to access AnnotationHub resources")
9625     (description
9626      "This package provides a client for the Bioconductor AnnotationHub web
9627 resource.  The AnnotationHub web resource provides a central location where
9628 genomic files (e.g. VCF, bed, wig) and other resources from standard
9629 locations (e.g. UCSC, Ensembl) can be discovered.  The resource includes
9630 metadata about each resource, e.g., a textual description, tags, and date of
9631 modification.  The client creates and manages a local cache of files retrieved
9632 by the user, helping with quick and reproducible access.")
9633     (license license:artistic2.0)))
9634
9635 (define-public r-fastseg
9636   (package
9637     (name "r-fastseg")
9638     (version "1.24.0")
9639     (source
9640      (origin
9641        (method url-fetch)
9642        (uri (bioconductor-uri "fastseg" version))
9643        (sha256
9644         (base32
9645          "0dd7nr3klwz9ailwshnbynhd62lwb8zbbpj6jf3igpb94yi6x2jp"))))
9646     (build-system r-build-system)
9647     (propagated-inputs
9648      `(("r-biobase" ,r-biobase)
9649        ("r-biocgenerics" ,r-biocgenerics)
9650        ("r-genomicranges" ,r-genomicranges)
9651        ("r-iranges" ,r-iranges)
9652        ("r-s4vectors" ,r-s4vectors)))
9653     (home-page "http://www.bioinf.jku.at/software/fastseg/index.html")
9654     (synopsis "Fast segmentation algorithm for genetic sequencing data")
9655     (description
9656      "Fastseg implements a very fast and efficient segmentation algorithm.
9657 It can segment data from DNA microarrays and data from next generation
9658 sequencing for example to detect copy number segments.  Further it can segment
9659 data from RNA microarrays like tiling arrays to identify transcripts.  Most
9660 generally, it can segment data given as a matrix or as a vector.  Various data
9661 formats can be used as input to fastseg like expression set objects for
9662 microarrays or GRanges for sequencing data.")
9663     (license license:lgpl2.0+)))
9664
9665 (define-public r-keggrest
9666   (package
9667     (name "r-keggrest")
9668     (version "1.18.1")
9669     (source
9670      (origin
9671        (method url-fetch)
9672        (uri (bioconductor-uri "KEGGREST" version))
9673        (sha256
9674         (base32
9675          "02gwmm79djj55a90dzc80hlgwc6bafl7xd7fnx2q59pk945k3z9c"))))
9676     (properties `((upstream-name . "KEGGREST")))
9677     (build-system r-build-system)
9678     (propagated-inputs
9679      `(("r-biostrings" ,r-biostrings)
9680        ("r-httr" ,r-httr)
9681        ("r-png" ,r-png)))
9682     (home-page "https://bioconductor.org/packages/KEGGREST")
9683     (synopsis "Client-side REST access to KEGG")
9684     (description
9685      "This package provides a package that provides a client interface to the
9686 @dfn{Kyoto Encyclopedia of Genes and Genomes} (KEGG) REST server.")
9687     (license license:artistic2.0)))
9688
9689 (define-public r-gage
9690   (package
9691     (name "r-gage")
9692     (version "2.28.2")
9693     (source
9694      (origin
9695        (method url-fetch)
9696        (uri (bioconductor-uri "gage" version))
9697        (sha256
9698         (base32
9699          "0h0mlhns9j7cpfksvdlvx9jb7szm3r1dwqb3s4s8p8hmkb9byyii"))))
9700     (build-system r-build-system)
9701     (propagated-inputs
9702      `(("r-annotationdbi" ,r-annotationdbi)
9703        ("r-graph" ,r-graph)
9704        ("r-keggrest" ,r-keggrest)))
9705     (home-page "http://www.biomedcentral.com/1471-2105/10/161")
9706     (synopsis "Generally applicable gene-set enrichment for pathway analysis")
9707     (description
9708      "GAGE is a published method for gene set (enrichment or GSEA) or pathway
9709 analysis.  GAGE is generally applicable independent of microarray or RNA-Seq
9710 data attributes including sample sizes, experimental designs, assay platforms,
9711 and other types of heterogeneity.  The gage package provides functions for
9712 basic GAGE analysis, result processing and presentation.  In addition, it
9713 provides demo microarray data and commonly used gene set data based on KEGG
9714 pathways and GO terms.  These funtions and data are also useful for gene set
9715 analysis using other methods.")
9716     (license license:gpl2+)))
9717
9718 (define-public r-genomicfiles
9719   (package
9720     (name "r-genomicfiles")
9721     (version "1.14.0")
9722     (source
9723      (origin
9724        (method url-fetch)
9725        (uri (bioconductor-uri "GenomicFiles" version))
9726        (sha256
9727         (base32
9728          "0r0wmrs5jycf1kckhnc2sgjmp336srlcjdkpbb1ymm7kazdd0s9n"))))
9729     (properties `((upstream-name . "GenomicFiles")))
9730     (build-system r-build-system)
9731     (propagated-inputs
9732      `(("r-biocgenerics" ,r-biocgenerics)
9733        ("r-biocparallel" ,r-biocparallel)
9734        ("r-genomeinfodb" ,r-genomeinfodb)
9735        ("r-genomicalignments" ,r-genomicalignments)
9736        ("r-genomicranges" ,r-genomicranges)
9737        ("r-iranges" ,r-iranges)
9738        ("r-rsamtools" ,r-rsamtools)
9739        ("r-rtracklayer" ,r-rtracklayer)
9740        ("r-s4vectors" ,r-s4vectors)
9741        ("r-summarizedexperiment" ,r-summarizedexperiment)
9742        ("r-variantannotation" ,r-variantannotation)))
9743     (home-page "https://bioconductor.org/packages/GenomicFiles")
9744     (synopsis "Distributed computing by file or by range")
9745     (description
9746      "This package provides infrastructure for parallel computations
9747 distributed by file or by range.  User defined mapper and reducer functions
9748 provide added flexibility for data combination and manipulation.")
9749     (license license:artistic2.0)))
9750
9751 (define-public r-complexheatmap
9752   (package
9753     (name "r-complexheatmap")
9754     (version "1.17.1")
9755     (source
9756      (origin
9757        (method url-fetch)
9758        (uri (bioconductor-uri "ComplexHeatmap" version))
9759        (sha256
9760         (base32
9761          "1x6kp55iqqsd8bhdl3qch95nfiy2y46ldbbsx1sj1v8f0b0ywwcy"))))
9762     (properties
9763      `((upstream-name . "ComplexHeatmap")))
9764     (build-system r-build-system)
9765     (propagated-inputs
9766      `(("r-circlize" ,r-circlize)
9767        ("r-colorspace" ,r-colorspace)
9768        ("r-getoptlong" ,r-getoptlong)
9769        ("r-globaloptions" ,r-globaloptions)
9770        ("r-rcolorbrewer" ,r-rcolorbrewer)))
9771     (home-page
9772      "https://github.com/jokergoo/ComplexHeatmap")
9773     (synopsis "Making Complex Heatmaps")
9774     (description
9775      "Complex heatmaps are efficient to visualize associations between
9776 different sources of data sets and reveal potential structures.  This package
9777 provides a highly flexible way to arrange multiple heatmaps and supports
9778 self-defined annotation graphics.")
9779     (license license:gpl2+)))
9780
9781 (define-public r-dirichletmultinomial
9782   (package
9783     (name "r-dirichletmultinomial")
9784     (version "1.20.0")
9785     (source
9786      (origin
9787        (method url-fetch)
9788        (uri (bioconductor-uri "DirichletMultinomial" version))
9789        (sha256
9790         (base32
9791          "1c4s6x0qm20556grcd1xys9kkpnlzpasaai474malwcg6qvgi4x1"))))
9792     (properties
9793      `((upstream-name . "DirichletMultinomial")))
9794     (build-system r-build-system)
9795     (inputs
9796      `(("gsl" ,gsl)))
9797     (propagated-inputs
9798      `(("r-biocgenerics" ,r-biocgenerics)
9799        ("r-iranges" ,r-iranges)
9800        ("r-s4vectors" ,r-s4vectors)))
9801     (home-page "https://bioconductor.org/packages/DirichletMultinomial")
9802     (synopsis "Dirichlet-Multinomial mixture models for microbiome data")
9803     (description
9804      "Dirichlet-multinomial mixture models can be used to describe variability
9805 in microbial metagenomic data.  This package is an interface to code
9806 originally made available by Holmes, Harris, and Quince, 2012, PLoS ONE 7(2):
9807 1-15.")
9808     (license license:lgpl3)))
9809
9810 (define-public r-ensembldb
9811   (package
9812     (name "r-ensembldb")
9813     (version "2.2.2")
9814     (source
9815      (origin
9816        (method url-fetch)
9817        (uri (bioconductor-uri "ensembldb" version))
9818        (sha256
9819         (base32
9820          "1yngndkf3588z91z0a2fvkg423p26ajm6xv1p27x0l9mzhhaqq3k"))))
9821     (build-system r-build-system)
9822     (propagated-inputs
9823      `(("r-annotationdbi" ,r-annotationdbi)
9824        ("r-annotationfilter" ,r-annotationfilter)
9825        ("r-annotationhub" ,r-annotationhub)
9826        ("r-biobase" ,r-biobase)
9827        ("r-biocgenerics" ,r-biocgenerics)
9828        ("r-biostrings" ,r-biostrings)
9829        ("r-curl" ,r-curl)
9830        ("r-dbi" ,r-dbi)
9831        ("r-genomeinfodb" ,r-genomeinfodb)
9832        ("r-genomicfeatures" ,r-genomicfeatures)
9833        ("r-genomicranges" ,r-genomicranges)
9834        ("r-iranges" ,r-iranges)
9835        ("r-protgenerics" ,r-protgenerics)
9836        ("r-rsamtools" ,r-rsamtools)
9837        ("r-rsqlite" ,r-rsqlite)
9838        ("r-rtracklayer" ,r-rtracklayer)
9839        ("r-s4vectors" ,r-s4vectors)))
9840     (home-page "https://github.com/jotsetung/ensembldb")
9841     (synopsis "Utilities to create and use Ensembl-based annotation databases")
9842     (description
9843      "The package provides functions to create and use transcript-centric
9844 annotation databases/packages.  The annotation for the databases are directly
9845 fetched from Ensembl using their Perl API.  The functionality and data is
9846 similar to that of the TxDb packages from the @code{GenomicFeatures} package,
9847 but, in addition to retrieve all gene/transcript models and annotations from
9848 the database, the @code{ensembldb} package also provides a filter framework
9849 allowing to retrieve annotations for specific entries like genes encoded on a
9850 chromosome region or transcript models of lincRNA genes.")
9851     ;; No version specified
9852     (license license:lgpl3+)))
9853
9854 (define-public r-organismdbi
9855   (package
9856     (name "r-organismdbi")
9857     (version "1.20.0")
9858     (source
9859      (origin
9860        (method url-fetch)
9861        (uri (bioconductor-uri "OrganismDbi" version))
9862        (sha256
9863         (base32
9864          "0yxvhwn0m53wfwp0zi81x96argdf7cf1lpymc2as51apvfcnjdl8"))))
9865     (properties `((upstream-name . "OrganismDbi")))
9866     (build-system r-build-system)
9867     (propagated-inputs
9868      `(("r-annotationdbi" ,r-annotationdbi)
9869        ("r-biobase" ,r-biobase)
9870        ("r-biocgenerics" ,r-biocgenerics)
9871        ("r-biocinstaller" ,r-biocinstaller)
9872        ("r-dbi" ,r-dbi)
9873        ("r-genomicfeatures" ,r-genomicfeatures)
9874        ("r-genomicranges" ,r-genomicranges)
9875        ("r-graph" ,r-graph)
9876        ("r-iranges" ,r-iranges)
9877        ("r-rbgl" ,r-rbgl)
9878        ("r-s4vectors" ,r-s4vectors)))
9879     (home-page "https://bioconductor.org/packages/OrganismDbi")
9880     (synopsis "Software to enable the smooth interfacing of database packages")
9881     (description "The package enables a simple unified interface to several
9882 annotation packages each of which has its own schema by taking advantage of
9883 the fact that each of these packages implements a select methods.")
9884     (license license:artistic2.0)))
9885
9886 (define-public r-biovizbase
9887   (package
9888     (name "r-biovizbase")
9889     (version "1.26.0")
9890     (source
9891      (origin
9892        (method url-fetch)
9893        (uri (bioconductor-uri "biovizBase" version))
9894        (sha256
9895         (base32
9896          "14l4vhj0a4ssr9m9zdzz3qpd4qw1mhgq5bmxq7jhrq3j9kmd6i2f"))))
9897     (properties `((upstream-name . "biovizBase")))
9898     (build-system r-build-system)
9899     (propagated-inputs
9900      `(("r-annotationdbi" ,r-annotationdbi)
9901        ("r-annotationfilter" ,r-annotationfilter)
9902        ("r-biocgenerics" ,r-biocgenerics)
9903        ("r-biostrings" ,r-biostrings)
9904        ("r-dichromat" ,r-dichromat)
9905        ("r-ensembldb" ,r-ensembldb)
9906        ("r-genomeinfodb" ,r-genomeinfodb)
9907        ("r-genomicalignments" ,r-genomicalignments)
9908        ("r-genomicfeatures" ,r-genomicfeatures)
9909        ("r-genomicranges" ,r-genomicranges)
9910        ("r-hmisc" ,r-hmisc)
9911        ("r-iranges" ,r-iranges)
9912        ("r-rcolorbrewer" ,r-rcolorbrewer)
9913        ("r-rsamtools" ,r-rsamtools)
9914        ("r-s4vectors" ,r-s4vectors)
9915        ("r-scales" ,r-scales)
9916        ("r-summarizedexperiment" ,r-summarizedexperiment)
9917        ("r-variantannotation" ,r-variantannotation)))
9918     (home-page "https://bioconductor.org/packages/biovizBase")
9919     (synopsis "Basic graphic utilities for visualization of genomic data")
9920     (description
9921      "The biovizBase package is designed to provide a set of utilities, color
9922 schemes and conventions for genomic data.  It serves as the base for various
9923 high-level packages for biological data visualization.  This saves development
9924 effort and encourages consistency.")
9925     (license license:artistic2.0)))
9926
9927 (define-public r-ggbio
9928   (package
9929     (name "r-ggbio")
9930     (version "1.26.0")
9931     (source
9932      (origin
9933        (method url-fetch)
9934        (uri (bioconductor-uri "ggbio" version))
9935        (sha256
9936         (base32
9937          "1bqxfqy0hff87ax92z4lfbjz01ndrz7x8pzm6dlkdmi52p30krm9"))))
9938     (build-system r-build-system)
9939     (propagated-inputs
9940      `(("r-annotationdbi" ,r-annotationdbi)
9941        ("r-annotationfilter" ,r-annotationfilter)
9942        ("r-biobase" ,r-biobase)
9943        ("r-biocgenerics" ,r-biocgenerics)
9944        ("r-biostrings" ,r-biostrings)
9945        ("r-biovizbase" ,r-biovizbase)
9946        ("r-bsgenome" ,r-bsgenome)
9947        ("r-ensembldb" ,r-ensembldb)
9948        ("r-genomeinfodb" ,r-genomeinfodb)
9949        ("r-genomicalignments" ,r-genomicalignments)
9950        ("r-genomicfeatures" ,r-genomicfeatures)
9951        ("r-genomicranges" ,r-genomicranges)
9952        ("r-ggally" ,r-ggally)
9953        ("r-ggplot2" ,r-ggplot2)
9954        ("r-gridextra" ,r-gridextra)
9955        ("r-gtable" ,r-gtable)
9956        ("r-hmisc" ,r-hmisc)
9957        ("r-iranges" ,r-iranges)
9958        ("r-organismdbi" ,r-organismdbi)
9959        ("r-reshape2" ,r-reshape2)
9960        ("r-rsamtools" ,r-rsamtools)
9961        ("r-rtracklayer" ,r-rtracklayer)
9962        ("r-s4vectors" ,r-s4vectors)
9963        ("r-scales" ,r-scales)
9964        ("r-summarizedexperiment" ,r-summarizedexperiment)
9965        ("r-variantannotation" ,r-variantannotation)))
9966     (home-page "http://www.tengfei.name/ggbio/")
9967     (synopsis "Visualization tools for genomic data")
9968     (description
9969      "The ggbio package extends and specializes the grammar of graphics for
9970 biological data.  The graphics are designed to answer common scientific
9971 questions, in particular those often asked of high throughput genomics data.
9972 All core Bioconductor data structures are supported, where appropriate.  The
9973 package supports detailed views of particular genomic regions, as well as
9974 genome-wide overviews.  Supported overviews include ideograms and grand linear
9975 views.  High-level plots include sequence fragment length, edge-linked
9976 interval to data view, mismatch pileup, and several splicing summaries.")
9977     (license license:artistic2.0)))
9978
9979 (define-public r-gprofiler
9980   (package
9981     (name "r-gprofiler")
9982     (version "0.6.4")
9983     (source
9984      (origin
9985        (method url-fetch)
9986        (uri (cran-uri "gProfileR" version))
9987        (sha256
9988         (base32
9989          "1cka02zbz1rbppm782qpxk1xn9qxbrv2gp5rgf970j906hxm2y0b"))))
9990     (properties `((upstream-name . "gProfileR")))
9991     (build-system r-build-system)
9992     (propagated-inputs
9993      `(("r-plyr" ,r-plyr)
9994        ("r-rcurl" ,r-rcurl)))
9995     (home-page "https://cran.r-project.org/web/packages/gProfileR/")
9996     (synopsis "Interface to the g:Profiler toolkit")
9997     (description
9998      "This package provides tools for functional enrichment analysis,
9999 gene identifier conversion and mapping homologous genes across related
10000 organisms via the @code{g:Profiler} toolkit.")
10001     (license license:gpl2+)))
10002
10003 (define-public r-gqtlbase
10004   (package
10005     (name "r-gqtlbase")
10006     (version "1.10.0")
10007     (source
10008      (origin
10009        (method url-fetch)
10010        (uri (bioconductor-uri "gQTLBase" version))
10011        (sha256
10012         (base32
10013          "1756vfcj2dkkgcmfkkg7qdaig36dv9gfvpypn9rbrky56wm1p035"))))
10014     (properties `((upstream-name . "gQTLBase")))
10015     (build-system r-build-system)
10016     (propagated-inputs
10017      `(("r-batchjobs" ,r-batchjobs)
10018        ("r-bbmisc" ,r-bbmisc)
10019        ("r-biocgenerics" ,r-biocgenerics)
10020        ("r-bit" ,r-bit)
10021        ("r-doparallel" ,r-doparallel)
10022        ("r-ff" ,r-ff)
10023        ("r-ffbase" ,r-ffbase)
10024        ("r-foreach" ,r-foreach)
10025        ("r-genomicfiles" ,r-genomicfiles)
10026        ("r-genomicranges" ,r-genomicranges)
10027        ("r-rtracklayer" ,r-rtracklayer)
10028        ("r-s4vectors" ,r-s4vectors)
10029        ("r-summarizedexperiment" ,r-summarizedexperiment)))
10030     (home-page "https://bioconductor.org/packages/gQTLBase")
10031     (synopsis "Infrastructure for eQTL, mQTL and similar studies")
10032     (description
10033      "The purpose of this package is to simplify the storage and interrogation
10034 of @dfn{quantitative trait loci} (QTL) archives, such as eQTL, mQTL, dsQTL,
10035 and more.")
10036     (license license:artistic2.0)))
10037
10038 (define-public r-snpstats
10039   (package
10040     (name "r-snpstats")
10041     (version "1.28.0")
10042     (source
10043      (origin
10044        (method url-fetch)
10045        (uri (bioconductor-uri "snpStats" version))
10046        (sha256
10047         (base32
10048          "1x9qwynh2hwl24vq02naf4mchpch7xi2pkdrlgw896k28kx0lvir"))))
10049     (properties `((upstream-name . "snpStats")))
10050     (build-system r-build-system)
10051     (inputs `(("zlib" ,zlib)))
10052     (propagated-inputs
10053      `(("r-biocgenerics" ,r-biocgenerics)
10054        ("r-matrix" ,r-matrix)
10055        ("r-survival" ,r-survival)
10056        ("r-zlibbioc" ,r-zlibbioc)))
10057     (home-page "https://bioconductor.org/packages/snpStats")
10058     (synopsis "Methods for SNP association studies")
10059     (description
10060      "This package provides classes and statistical methods for large
10061 @dfn{single-nucleotide polymorphism} (SNP) association studies.  This extends
10062 the earlier snpMatrix package, allowing for uncertainty in genotypes.")
10063     (license license:gpl3)))
10064
10065 (define-public r-homo-sapiens
10066   (package
10067     (name "r-homo-sapiens")
10068     (version "1.3.1")
10069     (source (origin
10070               (method url-fetch)
10071               ;; We cannot use bioconductor-uri here because this tarball is
10072               ;; located under "data/annotation/" instead of "bioc/".
10073               (uri (string-append "http://www.bioconductor.org/packages/"
10074                                   "release/data/annotation/src/contrib/"
10075                                   "Homo.sapiens_"
10076                                   version ".tar.gz"))
10077               (sha256
10078                (base32
10079                 "151vj7h5p1c8yd5swrchk46z469p135wk50hvkl0nhgndvy0jj01"))))
10080     (properties
10081      `((upstream-name . "Homo.sapiens")))
10082     (build-system r-build-system)
10083     (propagated-inputs
10084      `(("r-genomicfeatures" ,r-genomicfeatures)
10085        ("r-go-db" ,r-go-db)
10086        ("r-org-hs-eg-db" ,r-org-hs-eg-db)
10087        ("r-txdb-hsapiens-ucsc-hg19-knowngene" ,r-txdb-hsapiens-ucsc-hg19-knowngene)
10088        ("r-organismdbi" ,r-organismdbi)
10089        ("r-annotationdbi" ,r-annotationdbi)))
10090     (home-page "https://bioconductor.org/packages/Homo.sapiens/")
10091     (synopsis "Annotation package for the Homo.sapiens object")
10092     (description
10093      "This package contains the Homo.sapiens object to access data from
10094 several related annotation packages.")
10095     (license license:artistic2.0)))
10096
10097 (define-public r-erma
10098   (package
10099     (name "r-erma")
10100     (version "0.10.0")
10101     (source
10102      (origin
10103        (method url-fetch)
10104        (uri (bioconductor-uri "erma" version))
10105        (sha256
10106         (base32
10107          "0gcfs9g8vvdv5vmq9b21kd8sq5mizjj49nfzd4in9zvp4b9v7x1g"))))
10108     (build-system r-build-system)
10109     (propagated-inputs
10110      `(("r-annotationdbi" ,r-annotationdbi)
10111        ("r-biobase" ,r-biobase)
10112        ("r-biocgenerics" ,r-biocgenerics)
10113        ("r-foreach" ,r-foreach)
10114        ("r-genomicfiles" ,r-genomicfiles)
10115        ("r-genomicranges" ,r-genomicranges)
10116        ("r-ggplot2" ,r-ggplot2)
10117        ("r-homo-sapiens" ,r-homo-sapiens)
10118        ("r-rtracklayer" ,r-rtracklayer)
10119        ("r-s4vectors" ,r-s4vectors)
10120        ("r-shiny" ,r-shiny)
10121        ("r-summarizedexperiment" ,r-summarizedexperiment)))
10122     (home-page "https://bioconductor.org/packages/erma")
10123     (synopsis "Epigenomic road map adventures")
10124     (description
10125      "The epigenomics road map describes locations of epigenetic marks in DNA
10126 from a variety of cell types.  Of interest are locations of histone
10127 modifications, sites of DNA methylation, and regions of accessible chromatin.
10128 This package presents a selection of elements of the road map including
10129 metadata and outputs of the ChromImpute procedure applied to ENCODE cell lines
10130 by Ernst and Kellis.")
10131     (license license:artistic2.0)))
10132
10133 (define-public r-ldblock
10134   (package
10135     (name "r-ldblock")
10136     (version "1.8.0")
10137     (source
10138      (origin
10139        (method url-fetch)
10140        (uri (bioconductor-uri "ldblock" version))
10141        (sha256
10142         (base32
10143          "18nfsixh6d2wfrb9laqsgly5w1frzihhak683k0p8fdf51h4aqba"))))
10144     (build-system r-build-system)
10145     (propagated-inputs
10146      `(("r-erma" ,r-erma)
10147        ("r-genomeinfodb" ,r-genomeinfodb)
10148        ("r-genomicfiles" ,r-genomicfiles)
10149        ("r-go-db" ,r-go-db)
10150        ("r-homo-sapiens" ,r-homo-sapiens)
10151        ("r-matrix" ,r-matrix)
10152        ("r-rsamtools" ,r-rsamtools)
10153        ("r-snpstats" ,r-snpstats)
10154        ("r-variantannotation" ,r-variantannotation)))
10155     (home-page "https://bioconductor.org/packages/ldblock")
10156     (synopsis "Data structures for linkage disequilibrium measures in populations")
10157     (description
10158      "This package defines data structures for @dfn{linkage
10159 disequilibrium} (LD) measures in populations.  Its purpose is to simplify
10160 handling of existing population-level data for the purpose of flexibly
10161 defining LD blocks.")
10162     (license license:artistic2.0)))
10163
10164 (define-public r-gqtlstats
10165   (package
10166     (name "r-gqtlstats")
10167     (version "1.10.1")
10168     (source
10169      (origin
10170        (method url-fetch)
10171        (uri (bioconductor-uri "gQTLstats" version))
10172        (sha256
10173         (base32
10174          "0gvq1sf2zjbkk431x40z6wql3c1rpclnnwa2f1hvykb8mmw70kmq"))))
10175     (properties `((upstream-name . "gQTLstats")))
10176     (build-system r-build-system)
10177     (propagated-inputs
10178      `(("r-annotationdbi" ,r-annotationdbi)
10179        ("r-batchjobs" ,r-batchjobs)
10180        ("r-bbmisc" ,r-bbmisc)
10181        ("r-beeswarm" ,r-beeswarm)
10182        ("r-biobase" ,r-biobase)
10183        ("r-biocgenerics" ,r-biocgenerics)
10184        ("r-doparallel" ,r-doparallel)
10185        ("r-dplyr" ,r-dplyr)
10186        ("r-erma" ,r-erma)
10187        ("r-ffbase" ,r-ffbase)
10188        ("r-foreach" ,r-foreach)
10189        ("r-genomeinfodb" ,r-genomeinfodb)
10190        ("r-genomicfeatures" ,r-genomicfeatures)
10191        ("r-genomicfiles" ,r-genomicfiles)
10192        ("r-genomicranges" ,r-genomicranges)
10193        ("r-ggbeeswarm" ,r-ggbeeswarm)
10194        ("r-ggplot2" ,r-ggplot2)
10195        ("r-gqtlbase" ,r-gqtlbase)
10196        ("r-hardyweinberg" ,r-hardyweinberg)
10197        ("r-iranges" ,r-iranges)
10198        ("r-ldblock" ,r-ldblock)
10199        ("r-limma" ,r-limma)
10200        ("r-mgcv" ,r-mgcv)
10201        ("r-plotly" ,r-plotly)
10202        ("r-reshape2" ,r-reshape2)
10203        ("r-s4vectors" ,r-s4vectors)
10204        ("r-shiny" ,r-shiny)
10205        ("r-snpstats" ,r-snpstats)
10206        ("r-summarizedexperiment" ,r-summarizedexperiment)
10207        ("r-variantannotation" ,r-variantannotation)))
10208     (home-page "https://bioconductor.org/packages/gQTLstats")
10209     (synopsis "Computationally efficient analysis for eQTL and allied studies")
10210     (description
10211      "This package provides tools for the computationally efficient analysis
10212 of @dfn{quantitative trait loci} (QTL) data, including eQTL, mQTL, dsQTL, etc.
10213 The software in this package aims to support refinements and functional
10214 interpretation of members of a collection of association statistics on a
10215 family of feature/genome hypotheses.")
10216     (license license:artistic2.0)))
10217
10218 (define-public r-gviz
10219   (package
10220     (name "r-gviz")
10221     (version "1.22.3")
10222     (source
10223      (origin
10224        (method url-fetch)
10225        (uri (bioconductor-uri "Gviz" version))
10226        (sha256
10227         (base32
10228          "1grjzrjpzkw572pbvpsvdnfkfgwybl0cnjd7nnk2xdr26wnbsi9a"))))
10229     (properties `((upstream-name . "Gviz")))
10230     (build-system r-build-system)
10231     (propagated-inputs
10232      `(("r-annotationdbi" ,r-annotationdbi)
10233        ("r-biobase" ,r-biobase)
10234        ("r-biocgenerics" ,r-biocgenerics)
10235        ("r-biomart" ,r-biomart)
10236        ("r-biostrings" ,r-biostrings)
10237        ("r-biovizbase" ,r-biovizbase)
10238        ("r-bsgenome" ,r-bsgenome)
10239        ("r-digest" ,r-digest)
10240        ("r-genomeinfodb" ,r-genomeinfodb)
10241        ("r-genomicalignments" ,r-genomicalignments)
10242        ("r-genomicfeatures" ,r-genomicfeatures)
10243        ("r-genomicranges" ,r-genomicranges)
10244        ("r-iranges" ,r-iranges)
10245        ("r-lattice" ,r-lattice)
10246        ("r-latticeextra" ,r-latticeextra)
10247        ("r-matrixstats" ,r-matrixstats)
10248        ("r-rcolorbrewer" ,r-rcolorbrewer)
10249        ("r-rsamtools" ,r-rsamtools)
10250        ("r-rtracklayer" ,r-rtracklayer)
10251        ("r-s4vectors" ,r-s4vectors)
10252        ("r-xvector" ,r-xvector)))
10253     (home-page "https://bioconductor.org/packages/Gviz")
10254     (synopsis "Plotting data and annotation information along genomic coordinates")
10255     (description
10256      "Genomic data analyses requires integrated visualization of known genomic
10257 information and new experimental data.  Gviz uses the biomaRt and the
10258 rtracklayer packages to perform live annotation queries to Ensembl and UCSC
10259 and translates this to e.g. gene/transcript structures in viewports of the
10260 grid graphics package.  This results in genomic information plotted together
10261 with your data.")
10262     (license license:artistic2.0)))
10263
10264 (define-public r-gwascat
10265   (package
10266     (name "r-gwascat")
10267     (version "2.10.0")
10268     (source
10269      (origin
10270        (method url-fetch)
10271        (uri (bioconductor-uri "gwascat" version))
10272        (sha256
10273         (base32
10274          "0n5x5i5v6a8wpn5mxmlpkl34b4kyypmymiwww6g61zch7xqrgywi"))))
10275     (build-system r-build-system)
10276     (propagated-inputs
10277      `(("r-annotationdbi" ,r-annotationdbi)
10278        ("r-annotationhub" ,r-annotationhub)
10279        ("r-biocgenerics" ,r-biocgenerics)
10280        ("r-biostrings" ,r-biostrings)
10281        ("r-genomeinfodb" ,r-genomeinfodb)
10282        ("r-genomicfeatures" ,r-genomicfeatures)
10283        ("r-genomicranges" ,r-genomicranges)
10284        ("r-ggbio" ,r-ggbio)
10285        ("r-ggplot2" ,r-ggplot2)
10286        ("r-gqtlstats" ,r-gqtlstats)
10287        ("r-graph" ,r-graph)
10288        ("r-gviz" ,r-gviz)
10289        ("r-homo-sapiens" ,r-homo-sapiens)
10290        ("r-iranges" ,r-iranges)
10291        ("r-rsamtools" ,r-rsamtools)
10292        ("r-rtracklayer" ,r-rtracklayer)
10293        ("r-s4vectors" ,r-s4vectors)
10294        ("r-snpstats" ,r-snpstats)
10295        ("r-summarizedexperiment" ,r-summarizedexperiment)
10296        ("r-variantannotation" ,r-variantannotation)))
10297     (home-page "https://bioconductor.org/packages/gwascat")
10298     (synopsis "Tools for data in the EMBL-EBI GWAS catalog")
10299     (description
10300      "This package provides tools for representing and modeling data in the
10301 EMBL-EBI GWAS catalog.")
10302     (license license:artistic2.0)))
10303
10304 (define-public r-sushi
10305   (package
10306     (name "r-sushi")
10307     (version "1.16.0")
10308     (source (origin
10309               (method url-fetch)
10310               (uri (bioconductor-uri "Sushi" version))
10311               (sha256
10312                (base32
10313                 "0axaqm480z8d0b2ldgxwm0swava1p4irc62bpl17p2k8k78g687g"))))
10314     (properties `((upstream-name . "Sushi")))
10315     (build-system r-build-system)
10316     (propagated-inputs
10317      `(("r-biomart" ,r-biomart)
10318        ("r-zoo" ,r-zoo)))
10319     (home-page "https://bioconductor.org/packages/Sushi")
10320     (synopsis "Tools for visualizing genomics data")
10321     (description
10322      "This package provides flexible, quantitative, and integrative genomic
10323 visualizations for publication-quality multi-panel figures.")
10324     (license license:gpl2+)))
10325
10326 (define-public r-fithic
10327   (package
10328     (name "r-fithic")
10329     (version "1.4.0")
10330     (source (origin
10331               (method url-fetch)
10332               (uri (bioconductor-uri "FitHiC" version))
10333               (sha256
10334                (base32
10335                 "12ylhrppi051m7nqsgq95kzd9g9wmp34i0zzfi55cjqawlpx7c6n"))))
10336     (properties `((upstream-name . "FitHiC")))
10337     (build-system r-build-system)
10338     (propagated-inputs
10339      `(("r-data-table" ,r-data-table)
10340        ("r-fdrtool" ,r-fdrtool)
10341        ("r-rcpp" ,r-rcpp)))
10342     (home-page "https://bioconductor.org/packages/FitHiC")
10343     (synopsis "Confidence estimation for intra-chromosomal contact maps")
10344     (description
10345      "Fit-Hi-C is a tool for assigning statistical confidence estimates to
10346 intra-chromosomal contact maps produced by genome-wide genome architecture
10347 assays such as Hi-C.")
10348     (license license:gpl2+)))
10349
10350 (define-public r-hitc
10351   (package
10352     (name "r-hitc")
10353     (version "1.22.0")
10354     (source (origin
10355               (method url-fetch)
10356               (uri (bioconductor-uri "HiTC" version))
10357               (sha256
10358                (base32
10359                 "0288xa1jy6nzvz2ha07csmp6dirjw5r7p9vy69q2wsbyzr02ymkp"))))
10360     (properties `((upstream-name . "HiTC")))
10361     (build-system r-build-system)
10362     (propagated-inputs
10363      `(("r-biostrings" ,r-biostrings)
10364        ("r-genomeinfodb" ,r-genomeinfodb)
10365        ("r-genomicranges" ,r-genomicranges)
10366        ("r-iranges" ,r-iranges)
10367        ("r-matrix" ,r-matrix)
10368        ("r-rcolorbrewer" ,r-rcolorbrewer)
10369        ("r-rtracklayer" ,r-rtracklayer)))
10370     (home-page "https://bioconductor.org/packages/HiTC")
10371     (synopsis "High throughput chromosome conformation capture analysis")
10372     (description
10373      "The HiTC package was developed to explore high-throughput \"C\" data
10374 such as 5C or Hi-C.  Dedicated R classes as well as standard methods for
10375 quality controls, normalization, visualization, and further analysis are also
10376 provided.")
10377     (license license:artistic2.0)))
10378
10379 (define-public r-qvalue
10380   (package
10381     (name "r-qvalue")
10382     (version "2.10.0")
10383     (source
10384      (origin
10385        (method url-fetch)
10386        (uri (bioconductor-uri "qvalue" version))
10387        (sha256
10388         (base32
10389          "1rd9rnf16kh8wc076kahd9hsb9rfwsbzmz3kjmp0pj6rbiq0051i"))))
10390     (build-system r-build-system)
10391     (propagated-inputs
10392      `(("r-ggplot2" ,r-ggplot2)
10393        ("r-reshape2" ,r-reshape2)))
10394     (home-page "http://github.com/jdstorey/qvalue")
10395     (synopsis "Q-value estimation for false discovery rate control")
10396     (description
10397      "This package takes a list of p-values resulting from the simultaneous
10398 testing of many hypotheses and estimates their q-values and local @dfn{false
10399 discovery rate} (FDR) values.  The q-value of a test measures the proportion
10400 of false positives incurred when that particular test is called significant.
10401 The local FDR measures the posterior probability the null hypothesis is true
10402 given the test's p-value.  Various plots are automatically generated, allowing
10403 one to make sensible significance cut-offs.  The software can be applied to
10404 problems in genomics, brain imaging, astrophysics, and data mining.")
10405     ;; Any version of the LGPL.
10406     (license license:lgpl3+)))
10407
10408 (define-public r-hdf5array
10409   (package
10410     (name "r-hdf5array")
10411     (version "1.6.0")
10412     (source
10413      (origin
10414        (method url-fetch)
10415        (uri (bioconductor-uri "HDF5Array" version))
10416        (sha256
10417         (base32
10418          "0kcdza41saqv6vlpvqd841awbiwkg84lh0plx6c7fmfgbqv7a0jh"))))
10419     (properties `((upstream-name . "HDF5Array")))
10420     (build-system r-build-system)
10421     (propagated-inputs
10422      `(("r-biocgenerics" ,r-biocgenerics)
10423        ("r-delayedarray" ,r-delayedarray)
10424        ("r-iranges" ,r-iranges)
10425        ("r-rhdf5" ,r-rhdf5)
10426        ("r-s4vectors" ,r-s4vectors)))
10427     (home-page "https://bioconductor.org/packages/HDF5Array")
10428     (synopsis "HDF5 back end for DelayedArray objects")
10429     (description "This package provides an array-like container for convenient
10430 access and manipulation of HDF5 datasets.  It supports delayed operations and
10431 block processing.")
10432     (license license:artistic2.0)))
10433
10434 (define-public r-rhdf5lib
10435   (package
10436     (name "r-rhdf5lib")
10437     (version "1.0.0")
10438     (source
10439      (origin
10440        (method url-fetch)
10441        (uri (bioconductor-uri "Rhdf5lib" version))
10442        (sha256
10443         (base32
10444          "0kkc4rprjbqn2wvbx4d49kk9l91vihccxbl4843qr1wqk6v33r1w"))))
10445     (properties `((upstream-name . "Rhdf5lib")))
10446     (build-system r-build-system)
10447     (arguments
10448      `(#:phases
10449        (modify-phases %standard-phases
10450          (add-after 'unpack 'do-not-use-bundled-hdf5
10451            (lambda* (#:key inputs #:allow-other-keys)
10452              (for-each delete-file '("configure" "configure.ac"))
10453              ;; Do not make other packages link with the proprietary libsz.
10454              (substitute* "R/zzz.R"
10455                (("'%s/libhdf5_cpp.a %s/libhdf5.a %s/libsz.a'")
10456                 "'%s/libhdf5_cpp.a %s/libhdf5.a %s/libhdf5.a'")
10457                (("'%s/libhdf5.a %s/libsz.a'")
10458                 "'%s/libhdf5.a %s/libhdf5.a'"))
10459              (with-directory-excursion "src"
10460                (invoke "tar" "xvf" (assoc-ref inputs "hdf5-source"))
10461                (rename-file (string-append "hdf5-" ,(package-version hdf5))
10462                             "hdf5")
10463                (rename-file "Makevars.in" "Makevars")
10464                (substitute* "Makevars"
10465                  (("HDF5_CXX_LIB=.*")
10466                   (string-append "HDF5_CXX_LIB="
10467                                  (assoc-ref inputs "hdf5") "/lib/libhdf5_cpp.a\n"))
10468                  (("HDF5_LIB=.*")
10469                   (string-append "HDF5_LIB="
10470                                  (assoc-ref inputs "hdf5") "/lib/libhdf5.a\n"))
10471                  (("HDF5_CXX_INCLUDE=.*") "HDF5_CXX_INCLUDE=./hdf5/c++/src\n")
10472                  (("HDF5_INCLUDE=.*") "HDF5_INCLUDE=./hdf5/src\n")
10473                  ;; szip is non-free software
10474                  (("cp \\$\\{SZIP_LIB\\}.*") "")
10475                  (("PKG_LIBS = \\$\\{HDF5_LIB\\} \\$\\{SZIP_LIB\\}")
10476                   "PKG_LIBS = ${HDF5_LIB}\n")))
10477              #t)))))
10478     (inputs
10479      `(("zlib" ,zlib)))
10480     (propagated-inputs
10481      `(("hdf5" ,hdf5)))
10482     (native-inputs
10483      `(("hdf5-source" ,(package-source hdf5))))
10484     (home-page "https://bioconductor.org/packages/Rhdf5lib")
10485     (synopsis "HDF5 library as an R package")
10486     (description "This package provides C and C++ HDF5 libraries for use in R
10487 packages.")
10488     (license license:artistic2.0)))
10489
10490 (define-public r-beachmat
10491   (package
10492     (name "r-beachmat")
10493     (version "1.0.2")
10494     (source
10495      (origin
10496        (method url-fetch)
10497        (uri (bioconductor-uri "beachmat" version))
10498        (sha256
10499         (base32
10500          "0b6dzja5fbx4dawb7ixj67mlhw4fy62pfp20mfp918fy96zmdwqz"))))
10501     (build-system r-build-system)
10502     (inputs
10503      `(("hdf5" ,hdf5)))
10504     (propagated-inputs
10505      `(("r-delayedarray" ,r-delayedarray)
10506        ("r-hdf5array" ,r-hdf5array)
10507        ("r-rcpp" ,r-rcpp)
10508        ("r-rhdf5" ,r-rhdf5)
10509        ("r-rhdf5lib" ,r-rhdf5lib)))
10510     (home-page "https://bioconductor.org/packages/beachmat")
10511     (synopsis "Compiling Bioconductor to handle each matrix type")
10512     (description "This package provides a consistent C++ class interface for a
10513 variety of commonly used matrix types, including sparse and HDF5-backed
10514 matrices.")
10515     (license license:gpl3)))
10516
10517 (define-public r-singlecellexperiment
10518   (package
10519     (name "r-singlecellexperiment")
10520     (version "1.0.0")
10521     (source
10522      (origin
10523        (method url-fetch)
10524        (uri (bioconductor-uri "SingleCellExperiment" version))
10525        (sha256
10526         (base32
10527          "1r276i97w64a5vdlg6952gkj7bls909p42zl8fn8yz87cdwyaars"))))
10528     (properties
10529      `((upstream-name . "SingleCellExperiment")))
10530     (build-system r-build-system)
10531     (propagated-inputs
10532      `(("r-biocgenerics" ,r-biocgenerics)
10533        ("r-s4vectors" ,r-s4vectors)
10534        ("r-summarizedexperiment" ,r-summarizedexperiment)))
10535     (home-page "https://bioconductor.org/packages/SingleCellExperiment")
10536     (synopsis "S4 classes for single cell data")
10537     (description "This package defines an S4 class for storing data from
10538 single-cell experiments.  This includes specialized methods to store and
10539 retrieve spike-in information, dimensionality reduction coordinates and size
10540 factors for each cell, along with the usual metadata for genes and
10541 libraries.")
10542     (license license:gpl3)))
10543
10544 (define-public r-scater
10545   (package
10546     (name "r-scater")
10547     (version "1.6.3")
10548     (source (origin
10549               (method url-fetch)
10550               (uri (bioconductor-uri "scater" version))
10551               (sha256
10552                (base32
10553                 "0q3s96gf8saa1dq2fvmpl0jyj7bx3wrdfck3hanb8pxkcir2p7dn"))))
10554     (build-system r-build-system)
10555     (propagated-inputs
10556      `(("r-beachmat" ,r-beachmat)
10557        ("r-biobase" ,r-biobase)
10558        ("r-biocgenerics" ,r-biocgenerics)
10559        ("r-biomart" ,r-biomart)
10560        ("r-data-table" ,r-data-table)
10561        ("r-dplyr" ,r-dplyr)
10562        ("r-edger" ,r-edger)
10563        ("r-ggbeeswarm" ,r-ggbeeswarm)
10564        ("r-ggplot2" ,r-ggplot2)
10565        ("r-limma" ,r-limma)
10566        ("r-matrix" ,r-matrix)
10567        ("r-matrixstats" ,r-matrixstats)
10568        ("r-plyr" ,r-plyr)
10569        ("r-rcpp" ,r-rcpp)
10570        ("r-reshape2" ,r-reshape2)
10571        ("r-rhdf5" ,r-rhdf5)
10572        ("r-rhdf5lib" ,r-rhdf5lib)
10573        ("r-rjson" ,r-rjson)
10574        ("r-s4vectors" ,r-s4vectors)
10575        ("r-shiny" ,r-shiny)
10576        ("r-shinydashboard" ,r-shinydashboard)
10577        ("r-singlecellexperiment" ,r-singlecellexperiment)
10578        ("r-summarizedexperiment" ,r-summarizedexperiment)
10579        ("r-tximport" ,r-tximport)
10580        ("r-viridis" ,r-viridis)))
10581     (home-page "https://github.com/davismcc/scater")
10582     (synopsis "Single-cell analysis toolkit for gene expression data in R")
10583     (description "This package provides a collection of tools for doing
10584 various analyses of single-cell RNA-seq gene expression data, with a focus on
10585 quality control.")
10586     (license license:gpl2+)))
10587
10588 (define-public r-scran
10589   (package
10590     (name "r-scran")
10591     (version "1.6.7")
10592     (source
10593      (origin
10594        (method url-fetch)
10595        (uri (bioconductor-uri "scran" version))
10596        (sha256
10597         (base32
10598          "0zfm5i8xrnzb3894xygqd6zkbwaa2x3z74wbxw39fcjhyhxv2hmb"))))
10599     (build-system r-build-system)
10600     (propagated-inputs
10601      `(("r-beachmat" ,r-beachmat)
10602        ("r-biocgenerics" ,r-biocgenerics)
10603        ("r-biocparallel" ,r-biocparallel)
10604        ("r-dt" ,r-dt)
10605        ("r-dynamictreecut" ,r-dynamictreecut)
10606        ("r-edger" ,r-edger)
10607        ("r-fnn" ,r-fnn)
10608        ("r-ggplot2" ,r-ggplot2)
10609        ("r-igraph" ,r-igraph)
10610        ("r-limma" ,r-limma)
10611        ("r-matrix" ,r-matrix)
10612        ("r-rcpp" ,r-rcpp)
10613        ("r-rhdf5lib" ,r-rhdf5lib)
10614        ("r-s4vectors" ,r-s4vectors)
10615        ("r-scater" ,r-scater)
10616        ("r-shiny" ,r-shiny)
10617        ("r-singlecellexperiment" ,r-singlecellexperiment)
10618        ("r-statmod" ,r-statmod)
10619        ("r-summarizedexperiment" ,r-summarizedexperiment)
10620        ("r-viridis" ,r-viridis)
10621        ("r-zoo" ,r-zoo)))
10622     (home-page "https://bioconductor.org/packages/scran")
10623     (synopsis "Methods for single-cell RNA-Seq data analysis")
10624     (description "This package implements a variety of low-level analyses of
10625 single-cell RNA-seq data.  Methods are provided for normalization of
10626 cell-specific biases, assignment of cell cycle phase, and detection of highly
10627 variable and significantly correlated genes.")
10628     (license license:gpl3)))
10629
10630 (define-public r-delayedmatrixstats
10631   (package
10632     (name "r-delayedmatrixstats")
10633     (version "1.0.3")
10634     (source
10635      (origin
10636        (method url-fetch)
10637        (uri (bioconductor-uri "DelayedMatrixStats" version))
10638        (sha256
10639         (base32
10640          "1cxjbjdq9hg9cm95rci0al7a4pk2h73ym276ahw9q4977zbg6381"))))
10641     (properties
10642      `((upstream-name . "DelayedMatrixStats")))
10643     (build-system r-build-system)
10644     (propagated-inputs
10645      `(("r-delayedarray" ,r-delayedarray)
10646        ("r-iranges" ,r-iranges)
10647        ("r-matrix" ,r-matrix)
10648        ("r-matrixstats" ,r-matrixstats)
10649        ("r-s4vectors" ,r-s4vectors)))
10650     (home-page "https://github.com/PeteHaitch/DelayedMatrixStats")
10651     (synopsis "Functions that apply to rows and columns of DelayedMatrix objects")
10652     (description
10653      "This package provides a port of the @code{matrixStats} API for use with
10654 @code{DelayedMatrix} objects from the @code{DelayedArray} package.  It
10655 contains high-performing functions operating on rows and columns of
10656 @code{DelayedMatrix} objects, e.g. @code{colMedians}, @code{rowMedians},
10657 @code{colRanks}, @code{rowRanks}, @code{colSds}, and @code{rowSds}.  Functions
10658 are optimized per data type and for subsetted calculations such that both
10659 memory usage and processing time is minimized.")
10660     (license license:expat)))
10661
10662 (define-public r-dropbead
10663   (let ((commit "cf0be5ae5302684bd03e78ab65b142900bbbb840")
10664         (revision "1"))
10665     (package
10666       (name "r-dropbead")
10667       (version (string-append "0-" revision "." (string-take commit 7)))
10668       (source
10669        (origin
10670          (method git-fetch)
10671          (uri (git-reference
10672                (url "https://github.com/rajewsky-lab/dropbead.git")
10673                (commit commit)))
10674          (sha256
10675           (base32
10676            "1b2lphsc236s1rdzlijxg8yl1jnqpwcvj4x938r89rqpj93jb780"))))
10677       (build-system r-build-system)
10678       (propagated-inputs
10679        `(("r-ggplot2" ,r-ggplot2)
10680          ("r-rcolorbrewer" ,r-rcolorbrewer)
10681          ("r-gridextra" ,r-gridextra)
10682          ("r-gplots" ,r-gplots)
10683          ("r-plyr" ,r-plyr)))
10684       (home-page "https://github.com/rajewsky-lab/dropbead")
10685       (synopsis "Basic exploration and analysis of Drop-seq data")
10686       (description "This package offers a quick and straight-forward way to
10687 explore and perform basic analysis of single cell sequencing data coming from
10688 droplet sequencing.  It has been particularly tailored for Drop-seq.")
10689       (license license:gpl3))))
10690
10691 (define htslib-for-sambamba
10692   (let ((commit "2f3c3ea7b301f9b45737a793c0b2dcf0240e5ee5"))
10693     (package
10694       (inherit htslib)
10695       (name "htslib-for-sambamba")
10696       (version (string-append "1.3.1-1." (string-take commit 9)))
10697       (source
10698        (origin
10699          (method git-fetch)
10700          (uri (git-reference
10701                (url "https://github.com/lomereiter/htslib.git")
10702                (commit commit)))
10703          (file-name (string-append "htslib-" version "-checkout"))
10704          (sha256
10705           (base32
10706            "0g38g8s3npr0gjm9fahlbhiskyfws9l5i0x1ml3rakzj7az5l9c9"))))
10707       (native-inputs
10708        `(("autoconf" ,autoconf)
10709          ("automake" ,automake)
10710          ,@(package-native-inputs htslib))))))
10711
10712 (define-public sambamba
10713   (package
10714     (name "sambamba")
10715     (version "0.6.5")
10716     (source
10717      (origin
10718        (method url-fetch)
10719        (uri (string-append "https://github.com/lomereiter/sambamba/"
10720                            "archive/v" version ".tar.gz"))
10721        (file-name (string-append name "-" version ".tar.gz"))
10722        (sha256
10723         (base32
10724          "17076gijd65a3f07zns2gvbgahiz5lriwsa6dq353ss3jl85d8vy"))))
10725     (build-system gnu-build-system)
10726     (arguments
10727      `(#:tests? #f ; there is no test target
10728        #:make-flags
10729        '("D_COMPILER=ldc2"
10730          ;; Override "--compiler" flag only.
10731          "D_FLAGS=--compiler=ldc2 -IBioD -g -d"
10732          "sambamba-ldmd2-64")
10733        #:phases
10734        (modify-phases %standard-phases
10735          (delete 'configure)
10736          (add-after 'unpack 'place-biod
10737            (lambda* (#:key inputs #:allow-other-keys)
10738              (copy-recursively (assoc-ref inputs "biod") "BioD")
10739              #t))
10740          (add-after 'unpack 'unbundle-prerequisites
10741            (lambda _
10742              (substitute* "Makefile"
10743                ((" htslib-static lz4-static") ""))
10744              #t))
10745          (replace 'install
10746            (lambda* (#:key outputs #:allow-other-keys)
10747              (let* ((out   (assoc-ref outputs "out"))
10748                     (bin   (string-append out "/bin")))
10749                (mkdir-p bin)
10750                (install-file "build/sambamba" bin)
10751                #t))))))
10752     (native-inputs
10753      `(("ldc" ,ldc)
10754        ("rdmd" ,rdmd)
10755        ("biod"
10756         ,(let ((commit "1248586b54af4bd4dfb28ebfebfc6bf012e7a587"))
10757            (origin
10758              (method git-fetch)
10759              (uri (git-reference
10760                    (url "https://github.com/biod/BioD.git")
10761                    (commit commit)))
10762              (file-name (string-append "biod-"
10763                                        (string-take commit 9)
10764                                        "-checkout"))
10765              (sha256
10766               (base32
10767                "1m8hi1n7x0ri4l6s9i0x6jg4z4v94xrfdzp7mbizdipfag0m17g3")))))))
10768     (inputs
10769      `(("lz4" ,lz4)
10770        ("htslib" ,htslib-for-sambamba)))
10771     (home-page "http://lomereiter.github.io/sambamba")
10772     (synopsis "Tools for working with SAM/BAM data")
10773     (description "Sambamba is a high performance modern robust and
10774 fast tool (and library), written in the D programming language, for
10775 working with SAM and BAM files.  Current parallelised functionality is
10776 an important subset of samtools functionality, including view, index,
10777 sort, markdup, and depth.")
10778     (license license:gpl2+)))
10779
10780 (define-public ritornello
10781   (package
10782     (name "ritornello")
10783     (version "1.0.0")
10784     (source (origin
10785               (method url-fetch)
10786               (uri (string-append "https://github.com/KlugerLab/"
10787                                   "Ritornello/archive/v"
10788                                   version ".tar.gz"))
10789               (file-name (string-append name "-" version ".tar.gz"))
10790               (sha256
10791                (base32
10792                 "02nik86gq9ljjriv6pamwlmqnfky3ads1fpklx6mc3hx6k40pg38"))))
10793     (build-system gnu-build-system)
10794     (arguments
10795      `(#:tests? #f                      ; there are no tests
10796        #:phases
10797        (modify-phases %standard-phases
10798          (add-after 'unpack 'patch-samtools-references
10799            (lambda* (#:key inputs #:allow-other-keys)
10800              (substitute* '("src/SamStream.h"
10801                             "src/BufferedGenomeReader.h")
10802                (("<sam.h>") "<samtools/sam.h>"))
10803              #t))
10804          (delete 'configure)
10805          (replace 'install
10806            (lambda* (#:key inputs outputs #:allow-other-keys)
10807              (let* ((out    (assoc-ref outputs "out"))
10808                     (bin    (string-append out "/bin/")))
10809                (mkdir-p bin)
10810                (install-file "bin/Ritornello" bin)
10811                #t))))))
10812     (inputs
10813      `(("samtools" ,samtools-0.1)
10814        ("fftw" ,fftw)
10815        ("boost" ,boost)
10816        ("zlib" ,zlib)))
10817     (home-page "https://github.com/KlugerLab/Ritornello")
10818     (synopsis "Control-free peak caller for ChIP-seq data")
10819     (description "Ritornello is a ChIP-seq peak calling algorithm based on
10820 signal processing that can accurately call binding events without the need to
10821 do a pair total DNA input or IgG control sample.  It has been tested for use
10822 with narrow binding events such as transcription factor ChIP-seq.")
10823     (license license:gpl3+)))
10824
10825 (define-public trim-galore
10826   (package
10827     (name "trim-galore")
10828     (version "0.4.2")
10829     (source
10830      (origin
10831        (method url-fetch)
10832        (uri (string-append "http://www.bioinformatics.babraham.ac.uk/"
10833                            "projects/trim_galore/trim_galore_v"
10834                            version ".zip"))
10835        (sha256
10836         (base32
10837          "0b9qdxi4521gsrjvbhgky8g7kry9b5nx3byzaxkgxz7p4k8bn1mn"))))
10838     (build-system gnu-build-system)
10839     (arguments
10840      `(#:tests? #f                      ; no tests
10841        #:phases
10842        (modify-phases %standard-phases
10843          ;; The archive contains plain files.
10844          (replace 'unpack
10845            (lambda* (#:key source #:allow-other-keys)
10846              (zero? (system* "unzip" source))))
10847          (delete 'configure)
10848          (delete 'build)
10849          (add-after 'unpack 'hardcode-tool-references
10850            (lambda* (#:key inputs #:allow-other-keys)
10851              (substitute* "trim_galore"
10852                (("\\$path_to_cutadapt = 'cutadapt'")
10853                 (string-append "$path_to_cutadapt = '"
10854                                (assoc-ref inputs "cutadapt")
10855                                "/bin/cutadapt'"))
10856                (("\\| gzip")
10857                 (string-append "| "
10858                                (assoc-ref inputs "gzip")
10859                                "/bin/gzip"))
10860                (("\"gunzip")
10861                 (string-append "\""
10862                                (assoc-ref inputs "gzip")
10863                                "/bin/gunzip")))
10864              #t))
10865          (replace 'install
10866            (lambda* (#:key outputs #:allow-other-keys)
10867              (let ((bin (string-append (assoc-ref outputs "out")
10868                                        "/bin")))
10869                (mkdir-p bin)
10870                (install-file "trim_galore" bin)
10871                #t))))))
10872     (inputs
10873      `(("gzip" ,gzip)
10874        ("perl" ,perl)
10875        ("cutadapt" ,cutadapt)))
10876     (native-inputs
10877      `(("unzip" ,unzip)))
10878     (home-page "http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/")
10879     (synopsis "Wrapper around Cutadapt and FastQC")
10880     (description "Trim Galore! is a wrapper script to automate quality and
10881 adapter trimming as well as quality control, with some added functionality to
10882 remove biased methylation positions for RRBS sequence files.")
10883     (license license:gpl3+)))
10884
10885 (define-public gess
10886   (package
10887     (name "gess")
10888     (version "1.0")
10889     (source (origin
10890               (method url-fetch)
10891               (uri (string-append "http://compbio.uthscsa.edu/"
10892                                   "GESS_Web/files/"
10893                                   "gess-" version ".src.tar.gz"))
10894               (sha256
10895                (base32
10896                 "0hyk403kxscclzfs24pvdgiv0wm03kjcziqdrp5w46cb049gz0d7"))))
10897     (build-system gnu-build-system)
10898     (arguments
10899      `(#:tests? #f                      ; no tests
10900        #:phases
10901        (modify-phases %standard-phases
10902          (delete 'configure)
10903          (delete 'build)
10904          (replace 'install
10905            (lambda* (#:key inputs outputs #:allow-other-keys)
10906              (let* ((python (assoc-ref inputs "python"))
10907                     (out    (assoc-ref outputs "out"))
10908                     (bin    (string-append out "/bin/"))
10909                     (target (string-append
10910                              out "/lib/python2.7/site-packages/gess/")))
10911                (mkdir-p target)
10912                (copy-recursively "." target)
10913                ;; Make GESS.py executable
10914                (chmod (string-append target "GESS.py") #o555)
10915                ;; Add Python shebang to the top and make Matplotlib
10916                ;; usable.
10917                (substitute* (string-append target "GESS.py")
10918                  (("\"\"\"Description:" line)
10919                   (string-append "#!" (which "python") "
10920 import matplotlib
10921 matplotlib.use('Agg')
10922 " line)))
10923                ;; Make sure GESS has all modules in its path
10924                (wrap-program (string-append target "GESS.py")
10925                  `("PYTHONPATH" ":" prefix (,target ,(getenv "PYTHONPATH"))))
10926                (mkdir-p bin)
10927                (symlink (string-append target "GESS.py")
10928                         (string-append bin "GESS.py"))
10929                #t))))))
10930     (inputs
10931      `(("python" ,python-2)
10932        ("python2-pysam" ,python2-pysam)
10933        ("python2-scipy" ,python2-scipy)
10934        ("python2-numpy" ,python2-numpy)
10935        ("python2-networkx" ,python2-networkx)
10936        ("python2-biopython" ,python2-biopython)))
10937     (home-page "http://compbio.uthscsa.edu/GESS_Web/")
10938     (synopsis "Detect exon-skipping events from raw RNA-seq data")
10939     (description
10940      "GESS is an implementation of a novel computational method to detect de
10941 novo exon-skipping events directly from raw RNA-seq data without the prior
10942 knowledge of gene annotation information.  GESS stands for the graph-based
10943 exon-skipping scanner detection scheme.")
10944     (license license:bsd-3)))
10945
10946 (define-public phylip
10947   (package
10948     (name "phylip")
10949     (version "3.696")
10950     (source
10951      (origin
10952        (method url-fetch)
10953        (uri (string-append "http://evolution.gs.washington.edu/phylip/"
10954                            "download/phylip-" version ".tar.gz"))
10955        (sha256
10956         (base32
10957          "01jar1rayhr2gba2pgbw49m56rc5z4p5wn3ds0m188hrlln4a2nd"))))
10958     (build-system gnu-build-system)
10959     (arguments
10960      `(#:tests? #f                      ; no check target
10961        #:make-flags (list "-f" "Makefile.unx" "install")
10962        #:parallel-build? #f             ; not supported
10963        #:phases
10964        (modify-phases %standard-phases
10965          (add-after 'unpack 'enter-dir
10966            (lambda _ (chdir "src") #t))
10967          (delete 'configure)
10968          (replace 'install
10969            (lambda* (#:key inputs outputs #:allow-other-keys)
10970              (let ((target (string-append (assoc-ref outputs "out")
10971                                           "/bin")))
10972                (mkdir-p target)
10973                (for-each (lambda (file)
10974                            (install-file file target))
10975                          (find-files "../exe" ".*")))
10976              #t)))))
10977     (home-page "http://evolution.genetics.washington.edu/phylip/")
10978     (synopsis "Tools for inferring phylogenies")
10979     (description "PHYLIP (the PHYLogeny Inference Package) is a package of
10980 programs for inferring phylogenies (evolutionary trees).")
10981     (license license:bsd-2)))
10982
10983 (define-public imp
10984   (package
10985     (name "imp")
10986     (version "2.6.2")
10987     (source
10988      (origin
10989        (method url-fetch)
10990        (uri (string-append "https://integrativemodeling.org/"
10991                            version "/download/imp-" version ".tar.gz"))
10992        (sha256
10993         (base32
10994          "0lxqx7vh79d771svr611dkilp6sn30qrbw8zvscbrm37v38d2j6h"))))
10995     (build-system cmake-build-system)
10996     (arguments
10997      `(;; FIXME: Some tests fail because they produce warnings, others fail
10998        ;; because the PYTHONPATH does not include the modeller's directory.
10999        #:tests? #f))
11000     (inputs
11001      `(("boost" ,boost)
11002        ("gsl" ,gsl)
11003        ("swig" ,swig)
11004        ("hdf5" ,hdf5)
11005        ("fftw" ,fftw)
11006        ("python" ,python-2)))
11007     (propagated-inputs
11008      `(("python2-numpy" ,python2-numpy)
11009        ("python2-scipy" ,python2-scipy)
11010        ("python2-pandas" ,python2-pandas)
11011        ("python2-scikit-learn" ,python2-scikit-learn)
11012        ("python2-networkx" ,python2-networkx)))
11013     (home-page "https://integrativemodeling.org")
11014     (synopsis "Integrative modeling platform")
11015     (description "IMP's broad goal is to contribute to a comprehensive
11016 structural characterization of biomolecules ranging in size and complexity
11017 from small peptides to large macromolecular assemblies, by integrating data
11018 from diverse biochemical and biophysical experiments.  IMP provides a C++ and
11019 Python toolbox for solving complex modeling problems, and a number of
11020 applications for tackling some common problems in a user-friendly way.")
11021     ;; IMP is largely available under the GNU Lesser GPL; see the file
11022     ;; COPYING.LGPL for the full text of this license. Some IMP modules are
11023     ;; available under the GNU GPL (see the file COPYING.GPL).
11024     (license (list license:lgpl2.1+
11025                    license:gpl3+))))
11026
11027 (define-public tadbit
11028   (package
11029     (name "tadbit")
11030     (version "0.2")
11031     (source (origin
11032               (method url-fetch)
11033               (uri (string-append "https://github.com/3DGenomes/TADbit/"
11034                                   "archive/v" version ".tar.gz"))
11035               (file-name (string-append name "-" version ".tar.gz"))
11036               (sha256
11037                (base32
11038                 "1cnfqrl4685zar4nnw94j94nhvl2h29jm448nadqi1h05z6fdk4f"))))
11039     (build-system python-build-system)
11040     (arguments
11041      `(;; Tests are included and must be run after installation, but
11042        ;; they are incomplete and thus cannot be run.
11043        #:tests? #f
11044        #:python ,python-2
11045        #:phases
11046        (modify-phases %standard-phases
11047          (add-after 'unpack 'fix-problems-with-setup.py
11048            (lambda* (#:key outputs #:allow-other-keys)
11049              ;; setup.py opens these files for writing
11050              (chmod "_pytadbit/_version.py" #o664)
11051              (chmod "README.rst" #o664)
11052
11053              ;; Don't attempt to install the bash completions to
11054              ;; the home directory.
11055              (rename-file "extras/.bash_completion"
11056                           "extras/tadbit")
11057              (substitute* "setup.py"
11058                (("\\(path.expanduser\\('~'\\)")
11059                 (string-append "(\""
11060                                (assoc-ref outputs "out")
11061                                "/etc/bash_completion.d\""))
11062                (("extras/\\.bash_completion")
11063                 "extras/tadbit"))
11064              #t)))))
11065     (inputs
11066      ;; TODO: add Chimera for visualization
11067      `(("imp" ,imp)
11068        ("mcl" ,mcl)
11069        ("python2-scipy" ,python2-scipy)
11070        ("python2-numpy" ,python2-numpy)
11071        ("python2-matplotlib" ,python2-matplotlib)
11072        ("python2-pysam" ,python2-pysam)))
11073     (home-page "http://3dgenomes.github.io/TADbit/")
11074     (synopsis "Analyze, model, and explore 3C-based data")
11075     (description
11076      "TADbit is a complete Python library to deal with all steps to analyze,
11077 model, and explore 3C-based data.  With TADbit the user can map FASTQ files to
11078 obtain raw interaction binned matrices (Hi-C like matrices), normalize and
11079 correct interaction matrices, identify and compare the so-called
11080 @dfn{Topologically Associating Domains} (TADs), build 3D models from the
11081 interaction matrices, and finally, extract structural properties from the
11082 models.  TADbit is complemented by TADkit for visualizing 3D models.")
11083     (license license:gpl3+)))
11084
11085 (define-public kentutils
11086   (package
11087     (name "kentutils")
11088     ;; 302.1.0 is out, but the only difference is the inclusion of
11089     ;; pre-built binaries.
11090     (version "302.0.0")
11091     (source
11092      (origin
11093        (method url-fetch)
11094        (uri (string-append "https://github.com/ENCODE-DCC/kentUtils/"
11095                            "archive/v" version ".tar.gz"))
11096        (file-name (string-append name "-" version ".tar.gz"))
11097        (sha256
11098         (base32
11099          "134aja3k1cj32kbk1nnw0q9gxjb2krr15q6sga8qldzvc0585rmm"))
11100        (modules '((guix build utils)
11101                   (srfi srfi-26)
11102                   (ice-9 ftw)))
11103        (snippet
11104         '(begin
11105            ;; Only the contents of the specified directories are free
11106            ;; for all uses, so we remove the rest.  "hg/autoSql" and
11107            ;; "hg/autoXml" are nominally free, but they depend on a
11108            ;; library that is built from the sources in "hg/lib",
11109            ;; which is nonfree.
11110            (let ((free (list "." ".."
11111                              "utils" "lib" "inc" "tagStorm"
11112                              "parasol" "htslib"))
11113                  (directory? (lambda (file)
11114                                (eq? 'directory (stat:type (stat file))))))
11115              (for-each (lambda (file)
11116                          (and (directory? file)
11117                               (delete-file-recursively file)))
11118                        (map (cut string-append "src/" <>)
11119                             (scandir "src"
11120                                      (lambda (file)
11121                                        (not (member file free)))))))
11122            ;; Only make the utils target, not the userApps target,
11123            ;; because that requires libraries we won't build.
11124            (substitute* "Makefile"
11125              ((" userApps") " utils"))
11126            ;; Only build libraries that are free.
11127            (substitute* "src/makefile"
11128              (("DIRS =.*") "DIRS =\n")
11129              (("cd jkOwnLib.*") "")
11130              ((" hgLib") "")
11131              (("cd hg.*") ""))
11132            (substitute* "src/utils/makefile"
11133              ;; These tools depend on "jkhgap.a", which is part of the
11134              ;; nonfree "src/hg/lib" directory.
11135              (("raSqlQuery") "")
11136              (("pslLiftSubrangeBlat") "")
11137
11138              ;; Do not build UCSC tools, which may require nonfree
11139              ;; components.
11140              (("ALL_APPS =.*") "ALL_APPS = $(UTILS_APPLIST)\n"))
11141            #t))))
11142     (build-system gnu-build-system)
11143     (arguments
11144      `( ;; There is no global test target and the test target for
11145        ;; individual tools depends on input files that are not
11146        ;; included.
11147        #:tests? #f
11148        #:phases
11149        (modify-phases %standard-phases
11150          (add-after 'unpack 'fix-paths
11151            (lambda _
11152              (substitute* "Makefile"
11153                (("/bin/echo") (which "echo")))
11154              #t))
11155          (add-after 'unpack 'prepare-samtabix
11156            (lambda* (#:key inputs #:allow-other-keys)
11157              (copy-recursively (assoc-ref inputs "samtabix")
11158                                "samtabix")
11159              #t))
11160          (delete 'configure)
11161          (replace 'install
11162            (lambda* (#:key outputs #:allow-other-keys)
11163              (let ((bin (string-append (assoc-ref outputs "out")
11164                                        "/bin")))
11165                (copy-recursively "bin" bin))
11166              #t)))))
11167     (native-inputs
11168      `(("samtabix"
11169         ,(origin
11170            (method git-fetch)
11171            (uri (git-reference
11172                  (url "http://genome-source.cse.ucsc.edu/samtabix.git")
11173                  (commit "10fd107909c1ac4d679299908be4262a012965ba")))
11174            (sha256
11175             (base32
11176              "0c1nj64l42v395sa84n7az43xiap4i6f9n9dfz4058aqiwkhkmma"))))))
11177     (inputs
11178      `(("zlib" ,zlib)
11179        ("tcsh" ,tcsh)
11180        ("perl" ,perl)
11181        ("libpng" ,libpng)
11182        ("mysql" ,mysql)
11183        ("openssl" ,openssl)))
11184     (home-page "http://genome.cse.ucsc.edu/index.html")
11185     (synopsis "Assorted bioinformatics utilities")
11186     (description "This package provides the kentUtils, a selection of
11187 bioinformatics utilities used in combination with the UCSC genome
11188 browser.")
11189     ;; Only a subset of the sources are released under a non-copyleft
11190     ;; free software license.  All other sources are removed in a
11191     ;; snippet.  See this bug report for an explanation of how the
11192     ;; license statements apply:
11193     ;; https://github.com/ENCODE-DCC/kentUtils/issues/12
11194     (license (license:non-copyleft
11195               "http://genome.ucsc.edu/license/"
11196               "The contents of this package are free for all uses."))))
11197
11198 (define-public f-seq
11199   (let ((commit "6ccded34cff38cf432deed8503648b4a66953f9b")
11200         (revision "1"))
11201     (package
11202       (name "f-seq")
11203       (version (string-append "1.1-" revision "." (string-take commit 7)))
11204       (source (origin
11205                 (method git-fetch)
11206                 (uri (git-reference
11207                       (url "https://github.com/aboyle/F-seq.git")
11208                       (commit commit)))
11209                 (file-name (string-append name "-" version))
11210                 (sha256
11211                  (base32
11212                   "1nk33k0yajg2id4g59bc4szr58r2q6pdq42vgcw054m8ip9wv26h"))
11213                 (modules '((guix build utils)))
11214                 ;; Remove bundled Java library archives.
11215                 (snippet
11216                  '(begin
11217                     (for-each delete-file (find-files "lib" ".*"))
11218                     #t))))
11219       (build-system ant-build-system)
11220       (arguments
11221        `(#:tests? #f ; no tests included
11222          #:phases
11223          (modify-phases %standard-phases
11224            (replace 'install
11225              (lambda* (#:key inputs outputs #:allow-other-keys)
11226                (let* ((target (assoc-ref outputs "out"))
11227                       (doc (string-append target "/share/doc/f-seq/")))
11228                  (mkdir-p target)
11229                  (mkdir-p doc)
11230                  (substitute* "bin/linux/fseq"
11231                    (("java") (which "java"))
11232                    (("\\$REALDIR/../lib/commons-cli-1.1.jar")
11233                     (string-append (assoc-ref inputs "java-commons-cli")
11234                                    "/share/java/commons-cli.jar"))
11235                    (("REALDIR=.*")
11236                     (string-append "REALDIR=" target "/bin\n")))
11237                  (install-file "README.txt" doc)
11238                  (install-file "bin/linux/fseq" (string-append target "/bin"))
11239                  (install-file "build~/fseq.jar" (string-append target "/lib"))
11240                  (copy-recursively "lib" (string-append target "/lib"))
11241                  #t))))))
11242       (inputs
11243        `(("perl" ,perl)
11244          ("java-commons-cli" ,java-commons-cli)))
11245       (home-page "http://fureylab.web.unc.edu/software/fseq/")
11246       (synopsis "Feature density estimator for high-throughput sequence tags")
11247       (description
11248        "F-Seq is a software package that generates a continuous tag sequence
11249 density estimation allowing identification of biologically meaningful sites
11250 such as transcription factor binding sites (ChIP-seq) or regions of open
11251 chromatin (DNase-seq).  Output can be displayed directly in the UCSC Genome
11252 Browser.")
11253       (license license:gpl3+))))
11254
11255 (define-public bismark
11256   (package
11257     (name "bismark")
11258     (version "0.16.3")
11259     (source
11260      (origin
11261        (method url-fetch)
11262        (uri (string-append "https://github.com/FelixKrueger/Bismark/"
11263                            "archive/" version ".tar.gz"))
11264        (file-name (string-append name "-" version ".tar.gz"))
11265        (sha256
11266         (base32
11267          "1204i0pa02ll2jn5pnxypkclnskvv7a2nwh5nxhagmhxk9wfv9sq"))))
11268     (build-system perl-build-system)
11269     (arguments
11270      `(#:tests? #f ; there are no tests
11271        #:phases
11272        (modify-phases %standard-phases
11273          (delete 'configure)
11274          (delete 'build)
11275          (replace 'install
11276            (lambda* (#:key outputs #:allow-other-keys)
11277              (let ((bin (string-append (assoc-ref outputs "out")
11278                                        "/bin"))
11279                    (docdir  (string-append (assoc-ref outputs "out")
11280                                            "/share/doc/bismark"))
11281                    (docs    '("Bismark_User_Guide.pdf"
11282                               "RELEASE_NOTES.txt"))
11283                    (scripts '("bismark"
11284                               "bismark_genome_preparation"
11285                               "bismark_methylation_extractor"
11286                               "bismark2bedGraph"
11287                               "bismark2report"
11288                               "coverage2cytosine"
11289                               "deduplicate_bismark"
11290                               "bismark_sitrep.tpl"
11291                               "bam2nuc"
11292                               "bismark2summary")))
11293                (mkdir-p docdir)
11294                (mkdir-p bin)
11295                (for-each (lambda (file) (install-file file bin))
11296                          scripts)
11297                (for-each (lambda (file) (install-file file docdir))
11298                          docs)
11299                #t))))))
11300     (home-page "http://www.bioinformatics.babraham.ac.uk/projects/bismark/")
11301     (synopsis "Map bisulfite treated sequence reads and analyze methylation")
11302     (description "Bismark is a program to map bisulfite treated sequencing
11303 reads to a genome of interest and perform methylation calls in a single step.
11304 The output can be easily imported into a genome viewer, such as SeqMonk, and
11305 enables a researcher to analyse the methylation levels of their samples
11306 straight away.  Its main features are:
11307
11308 @itemize
11309 @item Bisulfite mapping and methylation calling in one single step
11310 @item Supports single-end and paired-end read alignments
11311 @item Supports ungapped and gapped alignments
11312 @item Alignment seed length, number of mismatches etc are adjustable
11313 @item Output discriminates between cytosine methylation in CpG, CHG
11314   and CHH context
11315 @end itemize\n")
11316     (license license:gpl3+)))
11317
11318 (define-public paml
11319   (package
11320     (name "paml")
11321     (version "4.9e")
11322     (source (origin
11323               (method url-fetch)
11324               (uri (string-append "http://abacus.gene.ucl.ac.uk/software/"
11325                                   "paml" version ".tgz"))
11326               (sha256
11327                (base32
11328                 "13zf6h9fiqghwhch2h06x1zdr6s42plsnqahflp5g7myr3han3s6"))
11329               (modules '((guix build utils)))
11330               ;; Remove Windows binaries
11331               (snippet
11332                '(begin
11333                   (for-each delete-file (find-files "." "\\.exe$"))
11334                   #t))))
11335     (build-system gnu-build-system)
11336     (arguments
11337      `(#:tests? #f ; there are no tests
11338        #:make-flags '("CC=gcc")
11339        #:phases
11340        (modify-phases %standard-phases
11341          (replace 'configure
11342            (lambda _
11343              (substitute* "src/BFdriver.c"
11344                (("/bin/bash") (which "bash")))
11345              (chdir "src")
11346              #t))
11347          (replace 'install
11348            (lambda* (#:key outputs #:allow-other-keys)
11349              (let ((tools '("baseml" "basemlg" "codeml"
11350                             "pamp" "evolver" "yn00" "chi2"))
11351                    (bin    (string-append (assoc-ref outputs "out") "/bin"))
11352                    (docdir (string-append (assoc-ref outputs "out")
11353                                            "/share/doc/paml")))
11354                (mkdir-p bin)
11355                (for-each (lambda (file) (install-file file bin)) tools)
11356                (copy-recursively "../doc" docdir)
11357                #t))))))
11358     (home-page "http://abacus.gene.ucl.ac.uk/software/paml.html")
11359     (synopsis "Phylogentic analysis by maximum likelihood")
11360     (description "PAML (for Phylogentic Analysis by Maximum Likelihood)
11361 contains a few programs for model fitting and phylogenetic tree reconstruction
11362 using nucleotide or amino-acid sequence data.")
11363     ;; GPLv3 only
11364     (license license:gpl3)))
11365
11366 (define-public kallisto
11367   (package
11368     (name "kallisto")
11369     (version "0.43.1")
11370     (source (origin
11371               (method url-fetch)
11372               (uri (string-append "https://github.com/pachterlab/"
11373                                   "kallisto/archive/v" version ".tar.gz"))
11374               (file-name (string-append name "-" version ".tar.gz"))
11375               (sha256
11376                (base32
11377                 "03j3iqhvq7ya3c91gidly3k3jvgm97vjq4scihrlxh315j696r11"))))
11378     (build-system cmake-build-system)
11379     (arguments `(#:tests? #f))          ; no "check" target
11380     (inputs
11381      `(("hdf5" ,hdf5)
11382        ("zlib" ,zlib)))
11383     (home-page "http://pachterlab.github.io/kallisto/")
11384     (synopsis "Near-optimal RNA-Seq quantification")
11385     (description
11386      "Kallisto is a program for quantifying abundances of transcripts from
11387 RNA-Seq data, or more generally of target sequences using high-throughput
11388 sequencing reads.  It is based on the novel idea of pseudoalignment for
11389 rapidly determining the compatibility of reads with targets, without the need
11390 for alignment.  Pseudoalignment of reads preserves the key information needed
11391 for quantification, and kallisto is therefore not only fast, but also as
11392 accurate as existing quantification tools.")
11393     (license license:bsd-2)))
11394
11395 (define-public libgff
11396   (package
11397     (name "libgff")
11398     (version "1.0")
11399     (source (origin
11400               (method url-fetch)
11401               (uri (string-append
11402                     "https://github.com/Kingsford-Group/"
11403                     "libgff/archive/v" version ".tar.gz"))
11404               (file-name (string-append name "-" version ".tar.gz"))
11405               (sha256
11406                (base32
11407                 "0vc4nxyhlm6g9vvmx5l4lfs5pnvixsv1hiiy4kddf2y3p6jna8ls"))))
11408     (build-system cmake-build-system)
11409     (arguments `(#:tests? #f))          ; no tests included
11410     (home-page "https://github.com/Kingsford-Group/libgff")
11411     (synopsis "Parser library for reading/writing GFF files")
11412     (description "This is a simple \"libraryfication\" of the GFF/GTF parsing
11413 code that is used in the Cufflinks codebase.  The goal of this library is to
11414 provide this functionality without the necessity of drawing in a heavy-weight
11415 dependency like SeqAn.")
11416     (license (license:x11-style "http://www.boost.org/LICENSE_1_0.txt"))))
11417
11418 (define-public libdivsufsort
11419   (package
11420     (name "libdivsufsort")
11421     (version "2.0.1")
11422     (source (origin
11423               (method git-fetch)
11424               (uri (git-reference
11425                     (url "https://github.com/y-256/libdivsufsort.git")
11426                     (commit version)))
11427               (sha256
11428                (base32
11429                 "0fgdz9fzihlvjjrxy01md1bv9vh12rkgkwbm90b1hj5xpbaqp7z2"))))
11430     (build-system cmake-build-system)
11431     (arguments
11432      '(#:tests? #f                      ; there are no tests
11433        #:configure-flags
11434        ;; Needed for rapmap and sailfish.
11435        '("-DBUILD_DIVSUFSORT64=ON")))
11436     (home-page "https://github.com/y-256/libdivsufsort")
11437     (synopsis "Lightweight suffix-sorting library")
11438     (description "libdivsufsort is a software library that implements a
11439 lightweight suffix array construction algorithm.  This library provides a
11440 simple and an efficient C API to construct a suffix array and a
11441 Burrows-Wheeler transformed string from a given string over a constant-size
11442 alphabet.  The algorithm runs in O(n log n) worst-case time using only 5n+O(1)
11443 bytes of memory space, where n is the length of the string.")
11444     (license license:expat)))
11445
11446 (define-public sailfish
11447   (package
11448     (name "sailfish")
11449     (version "0.10.1")
11450     (source (origin
11451               (method url-fetch)
11452               (uri
11453                (string-append "https://github.com/kingsfordgroup/"
11454                               "sailfish/archive/v" version ".tar.gz"))
11455               (file-name (string-append name "-" version ".tar.gz"))
11456               (sha256
11457                (base32
11458                 "1inn60dxiwsz8g9w7kvfhjxj4bwfb0r12dyhpzzhfbig712dkmm0"))
11459               (modules '((guix build utils)))
11460               (snippet
11461                '(begin
11462                   ;; Delete bundled headers for eigen3.
11463                   (delete-file-recursively "include/eigen3/")
11464                   #t))))
11465     (build-system cmake-build-system)
11466     (arguments
11467      `(#:configure-flags
11468        (list (string-append "-DBOOST_INCLUDEDIR="
11469                             (assoc-ref %build-inputs "boost")
11470                             "/include/")
11471              (string-append "-DBOOST_LIBRARYDIR="
11472                             (assoc-ref %build-inputs "boost")
11473                             "/lib/")
11474              (string-append "-DBoost_LIBRARIES="
11475                             "-lboost_iostreams "
11476                             "-lboost_filesystem "
11477                             "-lboost_system "
11478                             "-lboost_thread "
11479                             "-lboost_timer "
11480                             "-lboost_chrono "
11481                             "-lboost_program_options")
11482              "-DBoost_FOUND=TRUE"
11483              ;; Don't download RapMap---we already have it!
11484              "-DFETCHED_RAPMAP=1")
11485        ;; Tests must be run after installation and the location of the test
11486        ;; data file must be overridden.  But the tests fail.  It looks like
11487        ;; they are not really meant to be run.
11488        #:tests? #f
11489        #:phases
11490        (modify-phases %standard-phases
11491          ;; Boost cannot be found, even though it's right there.
11492          (add-after 'unpack 'do-not-look-for-boost
11493            (lambda* (#:key inputs #:allow-other-keys)
11494              (substitute* "CMakeLists.txt"
11495                (("find_package\\(Boost 1\\.53\\.0") "#"))))
11496          (add-after 'unpack 'do-not-assign-to-macro
11497            (lambda _
11498              (substitute* "include/spdlog/details/format.cc"
11499                (("const unsigned CHAR_WIDTH = 1;") ""))))
11500          (add-after 'unpack 'prepare-rapmap
11501            (lambda* (#:key inputs #:allow-other-keys)
11502              (let ((src "external/install/src/rapmap/")
11503                    (include "external/install/include/rapmap/")
11504                    (rapmap (assoc-ref inputs "rapmap")))
11505                (mkdir-p "/tmp/rapmap")
11506                (system* "tar" "xf"
11507                         (assoc-ref inputs "rapmap")
11508                         "-C" "/tmp/rapmap"
11509                         "--strip-components=1")
11510                (mkdir-p src)
11511                (mkdir-p include)
11512                (for-each (lambda (file)
11513                            (install-file file src))
11514                          (find-files "/tmp/rapmap/src" "\\.(c|cpp)"))
11515                (copy-recursively "/tmp/rapmap/include" include))))
11516          (add-after 'unpack 'use-system-libraries
11517            (lambda* (#:key inputs #:allow-other-keys)
11518              (substitute* '("src/SailfishIndexer.cpp"
11519                             "src/SailfishUtils.cpp"
11520                             "src/SailfishQuantify.cpp"
11521                             "src/FASTAParser.cpp"
11522                             "include/PCA.hpp"
11523                             "include/SailfishUtils.hpp"
11524                             "include/SailfishIndex.hpp"
11525                             "include/CollapsedEMOptimizer.hpp"
11526                             "src/CollapsedEMOptimizer.cpp")
11527                (("#include \"jellyfish/config.h\"") ""))
11528              (substitute* "src/CMakeLists.txt"
11529                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/include/jellyfish-2.2..")
11530                 (string-append (assoc-ref inputs "jellyfish")
11531                                "/include/jellyfish-" ,(package-version jellyfish)))
11532                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libjellyfish-2.0.a")
11533                 (string-append (assoc-ref inputs "jellyfish")
11534                                "/lib/libjellyfish-2.0.a"))
11535                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libdivsufsort.a")
11536                 (string-append (assoc-ref inputs "libdivsufsort")
11537                                "/lib/libdivsufsort.so"))
11538                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libdivsufsort64.a")
11539                 (string-append (assoc-ref inputs "libdivsufsort")
11540                                "/lib/libdivsufsort64.so")))
11541              (substitute* "CMakeLists.txt"
11542                ;; Don't prefer static libs
11543                (("SET\\(CMAKE_FIND_LIBRARY_SUFFIXES.*") "")
11544                (("find_package\\(Jellyfish.*") "")
11545                (("ExternalProject_Add\\(libjellyfish") "message(")
11546                (("ExternalProject_Add\\(libgff") "message(")
11547                (("ExternalProject_Add\\(libsparsehash") "message(")
11548                (("ExternalProject_Add\\(libdivsufsort") "message("))
11549
11550              ;; Ensure that Eigen headers can be found
11551              (setenv "CPLUS_INCLUDE_PATH"
11552                      (string-append (getenv "CPLUS_INCLUDE_PATH")
11553                                     ":"
11554                                     (assoc-ref inputs "eigen")
11555                                     "/include/eigen3")))))))
11556     (inputs
11557      `(("boost" ,boost)
11558        ("eigen" ,eigen)
11559        ("jemalloc" ,jemalloc)
11560        ("jellyfish" ,jellyfish)
11561        ("sparsehash" ,sparsehash)
11562        ("rapmap" ,(origin
11563                     (method git-fetch)
11564                     (uri (git-reference
11565                           (url "https://github.com/COMBINE-lab/RapMap.git")
11566                           (commit (string-append "sf-v" version))))
11567                     (file-name (string-append "rapmap-sf-v" version "-checkout"))
11568                     (sha256
11569                      (base32
11570                       "1hv79l5i576ykv5a1srj2p0q36yvyl5966m0fcy2lbi169ipjakf"))
11571                     (modules '((guix build utils)))
11572                     ;; These files are expected to be excluded.
11573                     (snippet
11574                      '(begin (delete-file-recursively "include/spdlog")
11575                              (for-each delete-file '("include/xxhash.h"
11576                                                      "src/xxhash.c"))))))
11577        ("libdivsufsort" ,libdivsufsort)
11578        ("libgff" ,libgff)
11579        ("tbb" ,tbb)
11580        ("zlib" ,zlib)))
11581     (native-inputs
11582      `(("pkg-config" ,pkg-config)))
11583     (home-page "http://www.cs.cmu.edu/~ckingsf/software/sailfish")
11584     (synopsis "Mapping-based isoform quantification from RNA-Seq reads")
11585     (description "Sailfish is a tool for genomic transcript quantification
11586 from RNA-seq data.  It requires a set of target transcripts (either from a
11587 reference or de-novo assembly) to quantify.  All you need to run sailfish is a
11588 fasta file containing your reference transcripts and a (set of) fasta/fastq
11589 file(s) containing your reads.")
11590     (license license:gpl3+)))
11591
11592 (define libstadenio-for-salmon
11593   (package
11594     (name "libstadenio")
11595     (version "1.14.8")
11596     (source (origin
11597               (method git-fetch)
11598               (uri (git-reference
11599                     (url "https://github.com/COMBINE-lab/staden-io_lib.git")
11600                     (commit (string-append "v" version))))
11601               (file-name (string-append name "-" version "-checkout"))
11602               (sha256
11603                (base32
11604                 "1x8kxxqxl892vwfbprlbyfwkkv7c34ggkc94892x9x0g37x5nbwx"))))
11605     (build-system gnu-build-system)
11606     (arguments '(#:parallel-tests? #f)) ; not supported
11607     (inputs
11608      `(("zlib" ,zlib)))
11609     (native-inputs
11610      `(("perl" ,perl)))                 ; for tests
11611     (home-page "https://github.com/COMBINE-lab/staden-io_lib")
11612     (synopsis "General purpose trace and experiment file library")
11613     (description "This package provides a library of file reading and writing
11614 code to provide a general purpose Trace file (and Experiment File) reading
11615 interface.
11616
11617 The following file formats are supported:
11618
11619 @enumerate
11620 @item SCF trace files
11621 @item ABI trace files
11622 @item ALF trace files
11623 @item ZTR trace files
11624 @item SFF trace archives
11625 @item SRF trace archives
11626 @item Experiment files
11627 @item Plain text files
11628 @item SAM/BAM sequence files
11629 @item CRAM sequence files
11630 @end enumerate\n")
11631     (license license:bsd-3)))
11632
11633 (define spdlog-for-salmon
11634   (package
11635     (name "spdlog")
11636     (version "0.14.0")
11637     (source (origin
11638               (method git-fetch)
11639               (uri (git-reference
11640                     (url "https://github.com/COMBINE-lab/spdlog.git")
11641                     (commit (string-append "v" version))))
11642               (file-name (string-append name "-" version "-checkout"))
11643               (sha256
11644                (base32
11645                 "13730429gwlabi432ilpnja3sfvy0nn2719vnhhmii34xcdyc57q"))))
11646     (build-system cmake-build-system)
11647     (home-page "https://github.com/COMBINE-lab/spdlog")
11648     (synopsis "Very fast C++ logging library")
11649     (description "Spdlog is a very fast header-only C++ logging library with
11650 performance as its primary goal.")
11651     (license license:expat)))
11652
11653 ;; This is a modified variant of bwa for use with Salmon. It installs a
11654 ;; library to avoid having to build this as part of Salmon.
11655 (define bwa-for-salmon
11656   (package (inherit bwa)
11657     (name "bwa")
11658     (version "0.7.12.5")
11659     (source (origin
11660               (method git-fetch)
11661               (uri (git-reference
11662                     (url "https://github.com/COMBINE-lab/bwa.git")
11663                     (commit (string-append "v" version))))
11664               (file-name (string-append "bwa-for-salmon-" version "-checkout"))
11665               (sha256
11666                (base32
11667                 "1z2qa64y0c5hky10510x137mnzlhz6k8qf27csw4w9j6qihq95gb"))))
11668     (build-system gnu-build-system)
11669     (arguments
11670      '(#:tests? #f ;no "check" target
11671        #:phases
11672        (modify-phases %standard-phases
11673          (replace 'install
11674            (lambda* (#:key outputs #:allow-other-keys)
11675              (let* ((out (assoc-ref outputs "out"))
11676                     (bin (string-append out "/bin"))
11677                     (lib (string-append out "/lib"))
11678                     (doc (string-append out "/share/doc/bwa"))
11679                     (man (string-append out "/share/man/man1"))
11680                     (inc (string-append out "/include/bwa")))
11681                (install-file "bwa" bin)
11682                (install-file "README.md" doc)
11683                (install-file "bwa.1" man)
11684                (install-file "libbwa.a" lib)
11685                (mkdir-p lib)
11686                (mkdir-p inc)
11687                (for-each (lambda (file)
11688                            (install-file file inc))
11689                          (find-files "." "\\.h$")))
11690              #t))
11691          ;; no "configure" script
11692          (delete 'configure))))))
11693
11694 (define-public salmon
11695   (package
11696     (name "salmon")
11697     (version "0.9.1")
11698     (source (origin
11699               (method git-fetch)
11700               (uri (git-reference
11701                     (url "https://github.com/COMBINE-lab/salmon.git")
11702                     (commit (string-append "v" version))))
11703               (file-name (string-append name "-" version "-checkout"))
11704               (sha256
11705                (base32
11706                 "1zi1ff4i7y2ykk0vdzysgwzzzv166vg2x77pj1mf4baclavxj87a"))
11707               (modules '((guix build utils)))
11708               (snippet
11709                '(begin
11710                   ;; Delete bundled headers for eigen3.
11711                   (delete-file-recursively "include/eigen3/")
11712                   #t))))
11713     (build-system cmake-build-system)
11714     (arguments
11715      `(#:configure-flags
11716        (list (string-append "-DBOOST_INCLUDEDIR="
11717                             (assoc-ref %build-inputs "boost")
11718                             "/include/")
11719              (string-append "-DBOOST_LIBRARYDIR="
11720                             (assoc-ref %build-inputs "boost")
11721                             "/lib/")
11722              (string-append "-DBoost_LIBRARIES="
11723                             "-lboost_iostreams "
11724                             "-lboost_filesystem "
11725                             "-lboost_system "
11726                             "-lboost_thread "
11727                             "-lboost_timer "
11728                             "-lboost_chrono "
11729                             "-lboost_program_options")
11730              "-DBoost_FOUND=TRUE"
11731              "-DTBB_LIBRARIES=tbb tbbmalloc"
11732              ;; Don't download RapMap---we already have it!
11733              "-DFETCHED_RAPMAP=1")
11734        #:phases
11735        (modify-phases %standard-phases
11736          ;; Boost cannot be found, even though it's right there.
11737          (add-after 'unpack 'do-not-look-for-boost
11738            (lambda* (#:key inputs #:allow-other-keys)
11739              (substitute* "CMakeLists.txt"
11740                (("find_package\\(Boost 1\\.53\\.0") "#"))))
11741          (add-after 'unpack 'do-not-phone-home
11742            (lambda _
11743              (substitute* "src/Salmon.cpp"
11744                (("getVersionMessage\\(\\)") "\"\""))))
11745          (add-after 'unpack 'prepare-rapmap
11746            (lambda* (#:key inputs #:allow-other-keys)
11747              (let ((src "external/install/src/rapmap/")
11748                    (include "external/install/include/rapmap/")
11749                    (rapmap (assoc-ref inputs "rapmap")))
11750                (mkdir-p src)
11751                (mkdir-p include)
11752                (for-each (lambda (file)
11753                            (install-file file src))
11754                          (find-files (string-append rapmap "/src") "\\.(c|cpp)"))
11755                (copy-recursively (string-append rapmap "/include") include)
11756                (for-each delete-file '("external/install/include/rapmap/xxhash.h"
11757                                        "external/install/include/rapmap/FastxParser.hpp"
11758                                        "external/install/include/rapmap/concurrentqueue.h"
11759                                        "external/install/include/rapmap/FastxParserThreadUtils.hpp"
11760                                        "external/install/src/rapmap/FastxParser.cpp"
11761                                        "external/install/src/rapmap/xxhash.c")))))
11762          (add-after 'unpack 'use-system-libraries
11763            (lambda* (#:key inputs #:allow-other-keys)
11764              (substitute* "src/CMakeLists.txt"
11765                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/include/jellyfish-2.2..")
11766                 (string-append (assoc-ref inputs "jellyfish")
11767                                "/include/jellyfish-" ,(package-version jellyfish)))
11768                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libjellyfish-2.0.a")
11769                 (string-append (assoc-ref inputs "jellyfish")
11770                                "/lib/libjellyfish-2.0.a"))
11771                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libdivsufsort.a")
11772                 (string-append (assoc-ref inputs "libdivsufsort")
11773                                "/lib/libdivsufsort.so"))
11774                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libstaden-read.a")
11775                 (string-append (assoc-ref inputs "libstadenio-for-salmon")
11776                                "/lib/libstaden-read.a"))
11777                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libbwa.a")
11778                 (string-append (assoc-ref inputs "bwa") "/lib/libbwa.a"))
11779                (("\\$\\{GAT_SOURCE_DIR\\}/external/install/lib/libdivsufsort64.a")
11780                 (string-append (assoc-ref inputs "libdivsufsort")
11781                                "/lib/libdivsufsort64.so")))
11782              (substitute* "CMakeLists.txt"
11783                ;; Don't prefer static libs
11784                (("SET\\(CMAKE_FIND_LIBRARY_SUFFIXES.*") "")
11785                (("set\\(TBB_LIBRARIES") "message(")
11786                (("find_package\\(Jellyfish.*") "")
11787                (("ExternalProject_Add\\(libcereal") "message(")
11788                (("ExternalProject_Add\\(libbwa") "message(")
11789                (("ExternalProject_Add\\(libjellyfish") "message(")
11790                (("ExternalProject_Add\\(libgff") "message(")
11791                (("ExternalProject_Add\\(libtbb") "message(")
11792                (("ExternalProject_Add\\(libspdlog") "message(")
11793                (("ExternalProject_Add\\(libdivsufsort") "message(")
11794                (("ExternalProject_Add\\(libstadenio") "message(")
11795                (("ExternalProject_Add_Step\\(") "message("))
11796
11797              ;; Ensure that all headers can be found
11798              (setenv "CPLUS_INCLUDE_PATH"
11799                      (string-append (getenv "CPLUS_INCLUDE_PATH")
11800                                     ":"
11801                                     (assoc-ref inputs "bwa")
11802                                     "/include/bwa"
11803                                     ":"
11804                                     (assoc-ref inputs "eigen")
11805                                     "/include/eigen3"))
11806              (setenv "CPATH"
11807                      (string-append (assoc-ref inputs "bwa")
11808                                     "/include/bwa"
11809                                     ":"
11810                                     (assoc-ref inputs "eigen")
11811                                     "/include/eigen3"))
11812              #t))
11813          ;; CMAKE_INSTALL_PREFIX does not exist when the tests are
11814          ;; run.  It only exists after the install phase.
11815          (add-after 'unpack 'fix-tests
11816            (lambda _
11817              (substitute* "src/CMakeLists.txt"
11818                (("DTOPLEVEL_DIR=\\$\\{CMAKE_INSTALL_PREFIX")
11819                 "DTOPLEVEL_DIR=${GAT_SOURCE_DIR"))
11820              #t)))))
11821     (inputs
11822      `(("boost" ,boost)
11823        ("bwa" ,bwa-for-salmon)
11824        ("bzip2" ,bzip2)
11825        ("cereal" ,cereal)
11826        ("eigen" ,eigen)
11827        ("rapmap" ,(origin
11828                     (method git-fetch)
11829                     (uri (git-reference
11830                           (url "https://github.com/COMBINE-lab/RapMap.git")
11831                           (commit (string-append "salmon-v" version))))
11832                     (file-name (string-append "rapmap-salmon-v" version "-checkout"))
11833                     (sha256
11834                      (base32
11835                       "1yc12yqsz6f0r8sg1qnk57xg34aqwc9jbqq6gd5ys28xw3plj98p"))))
11836        ("jemalloc" ,jemalloc)
11837        ("jellyfish" ,jellyfish)
11838        ("libgff" ,libgff)
11839        ("tbb" ,tbb)
11840        ("libdivsufsort" ,libdivsufsort)
11841        ("libstadenio-for-salmon" ,libstadenio-for-salmon)
11842        ("spdlog-for-salmon" ,spdlog-for-salmon)
11843        ("xz" ,xz)
11844        ("zlib" ,zlib)))
11845     (home-page "https://github.com/COMBINE-lab/salmon")
11846     (synopsis "Quantification from RNA-seq reads using lightweight alignments")
11847     (description "Salmon is a program to produce highly-accurate,
11848 transcript-level quantification estimates from RNA-seq data.  Salmon achieves
11849 its accuracy and speed via a number of different innovations, including the
11850 use of lightweight alignments (accurate but fast-to-compute proxies for
11851 traditional read alignments) and massively-parallel stochastic collapsed
11852 variational inference.")
11853     (license license:gpl3+)))
11854
11855 (define-public python-loompy
11856   (package
11857     (name "python-loompy")
11858     (version "2.0.2")
11859     (source
11860      (origin
11861        (method url-fetch)
11862        (uri (pypi-uri "loompy" version))
11863        (sha256
11864         (base32
11865          "1drgv8j1hxqzzpnfg272x9djb6j8qr798w1pc2x8ikmfgyd9gh51"))))
11866     (build-system python-build-system)
11867     ;; There are no tests
11868     (arguments '(#:tests? #f))
11869     (propagated-inputs
11870      `(("python-h5py" ,python-h5py)
11871        ("python-numpy" ,python-numpy)
11872        ("python-scipy" ,python-scipy)
11873        ("python-typing" ,python-typing)))
11874     (home-page "https://github.com/linnarsson-lab/loompy")
11875     (synopsis "Work with .loom files for single-cell RNA-seq data")
11876     (description "The loom file format is an efficient format for very large
11877 omics datasets, consisting of a main matrix, optional additional layers, a
11878 variable number of row and column annotations.  Loom also supports sparse
11879 graphs.  This library makes it easy to work with @file{.loom} files for
11880 single-cell RNA-seq data.")
11881     (license license:bsd-3)))