HCoop / jackhill / guix / guix.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
gnu: Move test packages from perl to perl-check.
[jackhill/guix/guix.git] / gnu / packages / bioinformatics.scm
CommitLineData
4e10a221 1;;; GNU Guix --- Functional package management for GNU
6193563a 2;;; Copyright © 2014, 2015, 2016, 2017 Ricardo Wurmus <rekado@elephly.net>
a9f754d7 3;;; Copyright © 2015, 2016, 2017 Ben Woodcroft <donttrustben@gmail.com>
8921841d 4;;; Copyright © 2015, 2016 Pjotr Prins <pjotr.guix@thebird.nl>
a5002ae7 5;;; Copyright © 2015 Andreas Enge <andreas@enge.fr>
a0a71439 6;;; Copyright © 2016 Roel Janssen <roel@gnu.org>
ce7e361f 7;;; Copyright © 2016, 2017 Efraim Flashner <efraim@flashner.co.il>
318c0aee 8;;; Copyright © 2016 Marius Bakke <mbakke@fastmail.com>
3fffabce 9;;; Copyright © 2016 Raoul Bonnal <ilpuccio.febo@gmail.com>
ce4dfde0 10;;; Copyright © 2017 Tobias Geerinckx-Rice <me@tobias.gr>
4e10a221
RW		 11;;;
12;;; This file is part of GNU Guix.
13;;;
14;;; GNU Guix is free software; you can redistribute it and/or modify it
15;;; under the terms of the GNU General Public License as published by
16;;; the Free Software Foundation; either version 3 of the License, or (at
17;;; your option) any later version.
18;;;
19;;; GNU Guix is distributed in the hope that it will be useful, but
20;;; WITHOUT ANY WARRANTY; without even the implied warranty of
21;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22;;; GNU General Public License for more details.
23;;;
24;;; You should have received a copy of the GNU General Public License
25;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
26
27(define-module (gnu packages bioinformatics)
28 #:use-module ((guix licenses) #:prefix license:)
29 #:use-module (guix packages)
8e913213 30 #:use-module (guix utils)
4e10a221 31 #:use-module (guix download)
2c16316e 32 #:use-module (guix git-download)
ec946638 33 #:use-module (guix hg-download)
10b4a969 34 #:use-module (guix build-system ant)
4e10a221 35 #:use-module (guix build-system gnu)
d7678942 36 #:use-module (guix build-system cmake)
c033f5d6 37 #:use-module (guix build-system ocaml)
365c8153 38 #:use-module (guix build-system perl)
8622a072 39 #:use-module (guix build-system python)
a5002ae7 40 #:use-module (guix build-system r)
9c38b540 41 #:use-module (guix build-system ruby)
d3517eda 42 #:use-module (guix build-system trivial)
4e10a221 43 #:use-module (gnu packages)
a2950fa4 44 #:use-module (gnu packages autotools)
684bf7c7 45 #:use-module (gnu packages algebra)
d3517eda 46 #:use-module (gnu packages base)
318c0aee 47 #:use-module (gnu packages bash)
a0a71439 48 #:use-module (gnu packages bison)
e4e5a4d8 49 #:use-module (gnu packages boost)
4e10a221 50 #:use-module (gnu packages compression)
82c370de 51 #:use-module (gnu packages cpio)
7cb61550 52 #:use-module (gnu packages cran)
1baee943 53 #:use-module (gnu packages curl)
99828fa7 54 #:use-module (gnu packages documentation)
94820951 55 #:use-module (gnu packages databases)
d29150b5 56 #:use-module (gnu packages datastructures)
75dd2424 57 #:use-module (gnu packages file)
99268755 58 #:use-module (gnu packages flex)
02f35bb5 59 #:use-module (gnu packages gawk)
2409f37f 60 #:use-module (gnu packages gcc)
66e40e00 61 #:use-module (gnu packages gd)
97b9da68 62 #:use-module (gnu packages gtk)
b16728b0 63 #:use-module (gnu packages glib)
18f5d2a7 64 #:use-module (gnu packages graph)
db7a3444 65 #:use-module (gnu packages groff)
50937297 66 #:use-module (gnu packages guile)
89984be4 67 #:use-module (gnu packages haskell)
66e40e00 68 #:use-module (gnu packages image)
97b9da68 69 #:use-module (gnu packages imagemagick)
15a3c3d4 70 #:use-module (gnu packages java)
5ded35d8 71 #:use-module (gnu packages ldc)
51c64999 72 #:use-module (gnu packages linux)
ec946638 73 #:use-module (gnu packages logging)
36742f43 74 #:use-module (gnu packages machine-learning)
db7a3444 75 #:use-module (gnu packages man)
c833ab55 76 #:use-module (gnu packages maths)
6c2b26e2 77 #:use-module (gnu packages mpi)
4e10a221 78 #:use-module (gnu packages ncurses)
c033f5d6 79 #:use-module (gnu packages ocaml)
81f3e0c1 80 #:use-module (gnu packages pcre)
ceb62d54 81 #:use-module (gnu packages parallel)
66e40e00 82 #:use-module (gnu packages pdf)
4e10a221 83 #:use-module (gnu packages perl)
5ccde207 84 #:use-module (gnu packages perl-check)
4e10a221 85 #:use-module (gnu packages pkg-config)
bfe3c685 86 #:use-module (gnu packages popt)
e4e5a4d8 87 #:use-module (gnu packages protobuf)
346a829a 88 #:use-module (gnu packages python)
ec946638 89 #:use-module (gnu packages readline)
9c38b540 90 #:use-module (gnu packages ruby)
84be3b99 91 #:use-module (gnu packages serialization)
94820951 92 #:use-module (gnu packages shells)
c833ab55 93 #:use-module (gnu packages statistics)
aa163424 94 #:use-module (gnu packages swig)
d7678942 95 #:use-module (gnu packages tbb)
97b9da68 96 #:use-module (gnu packages tex)
db7a3444 97 #:use-module (gnu packages texinfo)
2127cedb 98 #:use-module (gnu packages textutils)
43c565d2 99 #:use-module (gnu packages time)
a2950fa4 100 #:use-module (gnu packages tls)
ce7155d5 101 #:use-module (gnu packages vim)
365c8153 102 #:use-module (gnu packages web)
c833ab55 103 #:use-module (gnu packages xml)
66e40e00 104 #:use-module (gnu packages xorg)
2c9232ae 105 #:use-module (srfi srfi-1)
ce7e361f 106 #:use-module (ice-9 match))
4e10a221 107
036cd0cb
RJPB		 108(define-public r-ape
109 (package
110 (name "r-ape")
b12c6a66 111 (version "5.0")
036cd0cb
RJPB		 112 (source
113 (origin
114 (method url-fetch)
115 (uri (cran-uri "ape" version))
116 (sha256
117 (base32
b12c6a66 118 "0q59pmxawz498cb9mv5m49lhiwxib8ak94yyydz7qg8b6lpd4bn3"))))
036cd0cb 119 (build-system r-build-system)
aeb64f3c
RW		 120 (propagated-inputs
121 `(("r-lattice" ,r-lattice)
b12c6a66
RW		 122 ("r-nlme" ,r-nlme)
123 ("r-rcpp" ,r-rcpp)))
036cd0cb
RJPB		 124 (home-page "http://ape-package.ird.fr/")
125 (synopsis "Analyses of phylogenetics and evolution")
126 (description
127 "This package provides functions for reading, writing, plotting, and
128manipulating phylogenetic trees, analyses of comparative data in a
129phylogenetic framework, ancestral character analyses, analyses of
130diversification and macroevolution, computing distances from DNA sequences,
131and several other tools.")
132 (license license:gpl2+)))
133
8dc797fa
BW		 134(define-public aragorn
135 (package
136 (name "aragorn")
e990c81d 137 (version "1.2.38")
8dc797fa
BW		 138 (source (origin
139 (method url-fetch)
140 (uri (string-append
141 "http://mbio-serv2.mbioekol.lu.se/ARAGORN/Downloads/aragorn"
142 version ".tgz"))
143 (sha256
144 (base32
e990c81d 145 "09i1rg716smlbnixfm7q1ml2mfpaa2fpn3hwjg625ysmfwwy712b"))))
8dc797fa
BW		 146 (build-system gnu-build-system)
147 (arguments
148 `(#:tests? #f ; there are no tests
149 #:phases
150 (modify-phases %standard-phases
151 (delete 'configure)
152 (replace 'build
153 (lambda _
154 (zero? (system* "gcc"
155 "-O3"
156 "-ffast-math"
157 "-finline-functions"
158 "-o"
159 "aragorn"
160 (string-append "aragorn" ,version ".c")))))
161 (replace 'install
162 (lambda* (#:key outputs #:allow-other-keys)
163 (let* ((out (assoc-ref outputs "out"))
164 (bin (string-append out "/bin"))
165 (man (string-append out "/share/man/man1")))
166 (mkdir-p bin)
f3860753 167 (install-file "aragorn" bin)
8dc797fa 168 (mkdir-p man)
f3860753 169 (install-file "aragorn.1" man))
8dc797fa
BW		 170 #t)))))
171 (home-page "http://mbio-serv2.mbioekol.lu.se/ARAGORN")
172 (synopsis "Detect tRNA, mtRNA and tmRNA genes in nucleotide sequences")
173 (description
174 "Aragorn identifies transfer RNA, mitochondrial RNA and
175transfer-messenger RNA from nucleotide sequences, based on homology to known
176tRNA consensus sequences and RNA structure. It also outputs the secondary
177structure of the predicted RNA.")
178 (license license:gpl2)))
179
a12ba6e8
BW		 180(define-public bamm
181 (package
182 (name "bamm")
4b6da268 183 (version "1.7.3")
a12ba6e8
BW		 184 (source (origin
185 (method url-fetch)
186 ;; BamM is not available on pypi.
187 (uri (string-append
4b6da268 188 "https://github.com/Ecogenomics/BamM/archive/"
a12ba6e8
BW		 189 version ".tar.gz"))
190 (file-name (string-append name "-" version ".tar.gz"))
191 (sha256
192 (base32
4b6da268 193 "1f35yxp4pc8aadsvbpg6r4kg2jh4fkjci0iby4iyljm6980sac0s"))
a12ba6e8
BW		 194 (modules '((guix build utils)))
195 (snippet
196 `(begin
197 ;; Delete bundled htslib.
198 (delete-file-recursively "c/htslib-1.3.1")
199 #t))))
200 (build-system python-build-system)
201 (arguments
202 `(#:python ,python-2 ; BamM is Python 2 only.
203 ;; Do not use bundled libhts. Do use the bundled libcfu because it has
204 ;; been modified from its original form.
205 #:configure-flags
206 (let ((htslib (assoc-ref %build-inputs "htslib")))
207 (list "--with-libhts-lib" (string-append htslib "/lib")
208 "--with-libhts-inc" (string-append htslib "/include/htslib")))
209 #:phases
210 (modify-phases %standard-phases
211 (add-after 'unpack 'autogen
212 (lambda _
213 (with-directory-excursion "c"
214 (let ((sh (which "sh")))
215 ;; Use autogen so that 'configure' works.
216 (substitute* "autogen.sh" (("/bin/sh") sh))
217 (setenv "CONFIG_SHELL" sh)
218 (substitute* "configure" (("/bin/sh") sh))
219 (zero? (system* "./autogen.sh"))))))
220 (delete 'build)
221 ;; Run tests after installation so compilation only happens once.
222 (delete 'check)
223 (add-after 'install 'wrap-executable
224 (lambda* (#:key outputs #:allow-other-keys)
225 (let* ((out (assoc-ref outputs "out"))
226 (path (getenv "PATH")))
227 (wrap-program (string-append out "/bin/bamm")
228 `("PATH" ":" prefix (,path))))
229 #t))
230 (add-after 'wrap-executable 'post-install-check
231 (lambda* (#:key inputs outputs #:allow-other-keys)
232 (setenv "PATH"
233 (string-append (assoc-ref outputs "out")
234 "/bin:"
235 (getenv "PATH")))
236 (setenv "PYTHONPATH"
237 (string-append
238 (assoc-ref outputs "out")
239 "/lib/python"
240 (string-take (string-take-right
241 (assoc-ref inputs "python") 5) 3)
242 "/site-packages:"
243 (getenv "PYTHONPATH")))
244 ;; There are 2 errors printed, but they are safe to ignore:
245 ;; 1) [E::hts_open_format] fail to open file ...
246 ;; 2) samtools view: failed to open ...
247 (zero? (system* "nosetests")))))))
248 (native-inputs
249 `(("autoconf" ,autoconf)
250 ("automake" ,automake)
251 ("libtool" ,libtool)
252 ("zlib" ,zlib)
253 ("python-nose" ,python2-nose)
f3b98f4f 254 ("python-pysam" ,python2-pysam)))
a12ba6e8 255 (inputs
bca2c576 256 `(("htslib" ,htslib-1.3) ; At least one test fails on htslib-1.4+.
a12ba6e8
BW		 257 ("samtools" ,samtools)
258 ("bwa" ,bwa)
259 ("grep" ,grep)
260 ("sed" ,sed)
261 ("coreutils" ,coreutils)))
262 (propagated-inputs
263 `(("python-numpy" ,python2-numpy)))
264 (home-page "http://ecogenomics.github.io/BamM/")
265 (synopsis "Metagenomics-focused BAM file manipulator")
266 (description
267 "BamM is a C library, wrapped in python, to efficiently generate and
268parse BAM files, specifically for the analysis of metagenomic data. For
269instance, it implements several methods to assess contig-wise read coverage.")
270 (license license:lgpl3+)))
271
9794180d
RW		 272(define-public bamtools
273 (package
274 (name "bamtools")
48d66a9c 275 (version "2.4.1")
9794180d
RW		 276 (source (origin
277 (method url-fetch)
278 (uri (string-append
279 "https://github.com/pezmaster31/bamtools/archive/v"
280 version ".tar.gz"))
281 (file-name (string-append name "-" version ".tar.gz"))
282 (sha256
283 (base32
48d66a9c 284 "0jr024kcrhjb82cm69i7p5fcg5375zlc1h3qh2n1v368hcd0qflk"))))
9794180d 285 (build-system cmake-build-system)
4702cec2
RW		 286 (arguments
287 `(#:tests? #f ;no "check" target
288 #:phases
289 (modify-phases %standard-phases
290 (add-before
291 'configure 'set-ldflags
292 (lambda* (#:key outputs #:allow-other-keys)
293 (setenv "LDFLAGS"
294 (string-append
295 "-Wl,-rpath="
296 (assoc-ref outputs "out") "/lib/bamtools")))))))
9794180d
RW		 297 (inputs `(("zlib" ,zlib)))
298 (home-page "https://github.com/pezmaster31/bamtools")
299 (synopsis "C++ API and command-line toolkit for working with BAM data")
300 (description
301 "BamTools provides both a C++ API and a command-line toolkit for handling
302BAM files.")
303 (license license:expat)))
304
bdc7be59
MB		 305(define-public bcftools
306 (package
307 (name "bcftools")
0620387a 308 (version "1.5")
bdc7be59
MB		 309 (source (origin
310 (method url-fetch)
311 (uri (string-append
312 "https://github.com/samtools/bcftools/releases/download/"
313 version "/bcftools-" version ".tar.bz2"))
314 (sha256
315 (base32
0620387a 316 "0093hkkvxmbwfaa7905s6185jymynvg42kq6sxv7fili11l5mxwz"))
3837108e 317 (patches (search-patches "bcftools-regidx-unsigned-char.patch"))
bdc7be59
MB		 318 (modules '((guix build utils)))
319 (snippet
320 ;; Delete bundled htslib.
0620387a 321 '(delete-file-recursively "htslib-1.5"))))
bdc7be59
MB		 322 (build-system gnu-build-system)
323 (arguments
324 `(#:test-target "test"
0620387a 325 #:configure-flags (list "--with-htslib=system")
bdc7be59
MB		 326 #:make-flags
327 (list
328 "USE_GPL=1"
0620387a 329 "LIBS=-lgsl -lgslcblas"
bdc7be59
MB		 330 (string-append "prefix=" (assoc-ref %outputs "out"))
331 (string-append "HTSDIR=" (assoc-ref %build-inputs "htslib") "/include")
98593f9f 332 (string-append "HTSLIB=" (assoc-ref %build-inputs "htslib") "/lib/libhts.so")
bdc7be59 333 (string-append "BGZIP=" (assoc-ref %build-inputs "htslib") "/bin/bgzip")
98593f9f
BW		 334 (string-append "TABIX=" (assoc-ref %build-inputs "htslib") "/bin/tabix")
335 (string-append "PACKAGE_VERSION=" ,version))
bdc7be59
MB		 336 #:phases
337 (modify-phases %standard-phases
bdc7be59
MB		 338 (add-before 'check 'patch-tests
339 (lambda _
340 (substitute* "test/test.pl"
341 (("/bin/bash") (which "bash")))
342 #t)))))
343 (native-inputs
344 `(("htslib" ,htslib)
345 ("perl" ,perl)))
346 (inputs
347 `(("gsl" ,gsl)
348 ("zlib" ,zlib)))
349 (home-page "https://samtools.github.io/bcftools/")
350 (synopsis "Utilities for variant calling and manipulating VCFs and BCFs")
351 (description
352 "BCFtools is a set of utilities that manipulate variant calls in the
353Variant Call Format (VCF) and its binary counterpart BCF. All commands work
354transparently with both VCFs and BCFs, both uncompressed and BGZF-compressed.")
355 ;; The sources are dual MIT/GPL, but becomes GPL-only when USE_GPL=1.
356 (license (list license:gpl3+ license:expat))))
357
8dd4ff11
RW		 358(define-public bedops
359 (package
360 (name "bedops")
1bbc3b1d 361 (version "2.4.14")
8dd4ff11
RW		 362 (source (origin
363 (method url-fetch)
364 (uri (string-append "https://github.com/bedops/bedops/archive/v"
365 version ".tar.gz"))
f586c877 366 (file-name (string-append name "-" version ".tar.gz"))
8dd4ff11
RW		 367 (sha256
368 (base32
1bbc3b1d 369 "1kqbac547wyqma81cyky9n7mkgikjpsfd3nnmcm6hpqwanqgh10v"))))
8dd4ff11
RW		 370 (build-system gnu-build-system)
371 (arguments
372 '(#:tests? #f
373 #:make-flags (list (string-append "BINDIR=" %output "/bin"))
374 #:phases
dc1d3cde
KK		 375 (modify-phases %standard-phases
376 (add-after 'unpack 'unpack-tarballs
377 (lambda _
378 ;; FIXME: Bedops includes tarballs of minimally patched upstream
379 ;; libraries jansson, zlib, and bzip2. We cannot just use stock
380 ;; libraries because at least one of the libraries (zlib) is
381 ;; patched to add a C++ function definition (deflateInit2cpp).
382 ;; Until the Bedops developers offer a way to link against system
383 ;; libraries we have to build the in-tree copies of these three
384 ;; libraries.
385
386 ;; See upstream discussion:
387 ;; https://github.com/bedops/bedops/issues/124
388
389 ;; Unpack the tarballs to benefit from shebang patching.
390 (with-directory-excursion "third-party"
391 (and (zero? (system* "tar" "xvf" "jansson-2.6.tar.bz2"))
392 (zero? (system* "tar" "xvf" "zlib-1.2.7.tar.bz2"))
393 (zero? (system* "tar" "xvf" "bzip2-1.0.6.tar.bz2"))))
394 ;; Disable unpacking of tarballs in Makefile.
395 (substitute* "system.mk/Makefile.linux"
396 (("^\tbzcat .*") "\t@echo \"not unpacking\"\n")
397 (("\\./configure") "CONFIG_SHELL=bash ./configure"))
398 (substitute* "third-party/zlib-1.2.7/Makefile.in"
399 (("^SHELL=.*$") "SHELL=bash\n"))
400 #t))
401 (delete 'configure))))
8dd4ff11
RW		 402 (home-page "https://github.com/bedops/bedops")
403 (synopsis "Tools for high-performance genomic feature operations")
404 (description
405 "BEDOPS is a suite of tools to address common questions raised in genomic
406studies---mostly with regard to overlap and proximity relationships between
407data sets. It aims to be scalable and flexible, facilitating the efficient
408and accurate analysis and management of large-scale genomic data.
409
410BEDOPS provides tools that perform highly efficient and scalable Boolean and
411other set operations, statistical calculations, archiving, conversion and
412other management of genomic data of arbitrary scale. Tasks can be easily
413split by chromosome for distributing whole-genome analyses across a
414computational cluster.")
415 (license license:gpl2+)))
416
81de5647
RW		 417(define-public bedtools
418 (package
419 (name "bedtools")
d285657e 420 (version "2.26.0")
81de5647
RW		 421 (source (origin
422 (method url-fetch)
423 (uri (string-append "https://github.com/arq5x/bedtools2/archive/v"
424 version ".tar.gz"))
f586c877 425 (file-name (string-append name "-" version ".tar.gz"))
81de5647
RW		 426 (sha256
427 (base32
d285657e 428 "0xvri5hnp2iim1cx6mcd5d9f102p5ql41x69rd6106x1c17pinqm"))))
81de5647
RW		 429 (build-system gnu-build-system)
430 (native-inputs `(("python" ,python-2)))
431 (inputs `(("samtools" ,samtools)
432 ("zlib" ,zlib)))
433 (arguments
434 '(#:test-target "test"
435 #:phases
6573ac82 436 (modify-phases %standard-phases
6573ac82
BW		 437 (delete 'configure)
438 (replace 'install
439 (lambda* (#:key outputs #:allow-other-keys)
440 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
441 (for-each (lambda (file)
442 (install-file file bin))
443 (find-files "bin" ".*")))
444 #t)))))
81de5647
RW		 445 (home-page "https://github.com/arq5x/bedtools2")
446 (synopsis "Tools for genome analysis and arithmetic")
447 (description
448 "Collectively, the bedtools utilities are a swiss-army knife of tools for
449a wide-range of genomics analysis tasks. The most widely-used tools enable
450genome arithmetic: that is, set theory on the genome. For example, bedtools
451allows one to intersect, merge, count, complement, and shuffle genomic
452intervals from multiple files in widely-used genomic file formats such as BAM,
453BED, GFF/GTF, VCF.")
454 (license license:gpl2)))
455
9a8f309c
RW		 456;; Later releases of bedtools produce files with more columns than
457;; what Ribotaper expects.
458(define-public bedtools-2.18
459 (package (inherit bedtools)
460 (name "bedtools")
461 (version "2.18.0")
462 (source (origin
463 (method url-fetch)
464 (uri (string-append "https://github.com/arq5x/bedtools2/"
465 "archive/v" version ".tar.gz"))
466 (file-name (string-append name "-" version ".tar.gz"))
467 (sha256
468 (base32
469 "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))))
470
17dc32a4
RW		 471(define-public ribotaper
472 (package
473 (name "ribotaper")
474 (version "1.3.1")
475 (source (origin
476 (method url-fetch)
477 (uri (string-append "https://ohlerlab.mdc-berlin.de/"
478 "files/RiboTaper/RiboTaper_Version_"
479 version ".tar.gz"))
480 (sha256
481 (base32
482 "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
483 (build-system gnu-build-system)
484 (inputs
485 `(("bedtools" ,bedtools-2.18)
486 ("samtools" ,samtools-0.1)
2d7c4ae3 487 ("r-minimal" ,r-minimal)
17dc32a4
RW		 488 ("r-foreach" ,r-foreach)
489 ("r-xnomial" ,r-xnomial)
490 ("r-domc" ,r-domc)
491 ("r-multitaper" ,r-multitaper)
492 ("r-seqinr" ,r-seqinr)))
493 (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
494 (synopsis "Define translated ORFs using ribosome profiling data")
495 (description
496 "Ribotaper is a method for defining translated @dfn{open reading
497frames} (ORFs) using ribosome profiling (ribo-seq) data. This package
498provides the Ribotaper pipeline.")
499 (license license:gpl3+)))
500
769fc6bb
RW		 501(define-public ribodiff
502 (package
503 (name "ribodiff")
504 (version "0.2.2")
505 (source
506 (origin
507 (method url-fetch)
508 (uri (string-append "https://github.com/ratschlab/RiboDiff/"
509 "archive/v" version ".tar.gz"))
510 (file-name (string-append name "-" version ".tar.gz"))
511 (sha256
512 (base32
513 "0wpbwmfv05wdjxv7ikm664f7s7p7cqr8jnw99zrda0q67rl50aaj"))))
514 (build-system python-build-system)
515 (arguments
516 `(#:python ,python-2
517 #:phases
518 (modify-phases %standard-phases
519 ;; Generate an installable executable script wrapper.
520 (add-after 'unpack 'patch-setup.py
521 (lambda _
522 (substitute* "setup.py"
523 (("^(.*)packages=.*" line prefix)
524 (string-append line "\n"
525 prefix "scripts=['scripts/TE.py'],\n")))
526 #t)))))
527 (inputs
528 `(("python-numpy" ,python2-numpy)
529 ("python-matplotlib" ,python2-matplotlib)
530 ("python-scipy" ,python2-scipy)
531 ("python-statsmodels" ,python2-statsmodels)))
2bb12f5a
MB		 532 (native-inputs
533 `(("python-mock" ,python2-mock)
534 ("python-nose" ,python2-nose)))
769fc6bb
RW		 535 (home-page "http://public.bmi.inf.ethz.ch/user/zhongy/RiboDiff/")
536 (synopsis "Detect translation efficiency changes from ribosome footprints")
537 (description "RiboDiff is a statistical tool that detects the protein
538translational efficiency change from Ribo-Seq (ribosome footprinting) and
539RNA-Seq data. It uses a generalized linear model to detect genes showing
540difference in translational profile taking mRNA abundance into account. It
541facilitates us to decipher the translational regulation that behave
542independently with transcriptional regulation.")
543 (license license:gpl3+)))
544
a0a71439
RJ		 545(define-public bioawk
546 (package
547 (name "bioawk")
548 (version "1.0")
549 (source (origin
550 (method url-fetch)
551 (uri (string-append "https://github.com/lh3/bioawk/archive/v"
552 version ".tar.gz"))
553 (file-name (string-append name "-" version ".tar.gz"))
554 (sha256
555 (base32 "1daizxsk17ahi9n58fj8vpgwyhzrzh54bzqhanjanp88kgrz7gjw"))))
556 (build-system gnu-build-system)
557 (inputs
558 `(("zlib" ,zlib)))
559 (native-inputs
560 `(("bison" ,bison)))
561 (arguments
562 `(#:tests? #f ; There are no tests to run.
563 ;; Bison must generate files, before other targets can build.
564 #:parallel-build? #f
565 #:phases
566 (modify-phases %standard-phases
567 (delete 'configure) ; There is no configure phase.
568 (replace 'install
569 (lambda* (#:key outputs #:allow-other-keys)
570 (let* ((out (assoc-ref outputs "out"))
571 (bin (string-append out "/bin"))
572 (man (string-append out "/share/man/man1")))
573 (mkdir-p man)
574 (copy-file "awk.1" (string-append man "/bioawk.1"))
575 (install-file "bioawk" bin)))))))
576 (home-page "https://github.com/lh3/bioawk")
577 (synopsis "AWK with bioinformatics extensions")
578 (description "Bioawk is an extension to Brian Kernighan's awk, adding the
579support of several common biological data formats, including optionally gzip'ed
580BED, GFF, SAM, VCF, FASTA/Q and TAB-delimited formats with column names. It
581also adds a few built-in functions and a command line option to use TAB as the
582input/output delimiter. When the new functionality is not used, bioawk is
583intended to behave exactly the same as the original BWK awk.")
584 (license license:x11)))
585
a2fb1492
RW		 586(define-public python2-pybedtools
587 (package
588 (name "python2-pybedtools")
589 (version "0.6.9")
590 (source (origin
591 (method url-fetch)
592 (uri (string-append
593 "https://pypi.python.org/packages/source/p/pybedtools/pybedtools-"
594 version ".tar.gz"))
595 (sha256
596 (base32
597 "1ldzdxw1p4y3g2ignmggsdypvqkcwqwzhdha4rbgpih048z5p4an"))))
598 (build-system python-build-system)
599 (arguments `(#:python ,python-2)) ; no Python 3 support
600 (inputs
f2516de2 601 `(("python-matplotlib" ,python2-matplotlib)))
a2fb1492
RW		 602 (propagated-inputs
603 `(("bedtools" ,bedtools)
604 ("samtools" ,samtools)))
605 (native-inputs
f2516de2
HG		 606 `(("python-cython" ,python2-cython)
607 ("python-pyyaml" ,python2-pyyaml)
f3b98f4f 608 ("python-nose" ,python2-nose)))
a2fb1492
RW		 609 (home-page "https://pythonhosted.org/pybedtools/")
610 (synopsis "Python wrapper for BEDtools programs")
611 (description
612 "pybedtools is a Python wrapper for Aaron Quinlan's BEDtools programs,
613which are widely used for genomic interval manipulation or \"genome algebra\".
614pybedtools extends BEDTools by offering feature-level manipulations from with
615Python.")
616 (license license:gpl2+)))
617
9e12eba8
BW		 618(define-public python-biom-format
619 (package
620 (name "python-biom-format")
abc08cba 621 (version "2.1.6")
9e12eba8
BW		 622 (source
623 (origin
624 (method url-fetch)
625 ;; Use GitHub as source because PyPI distribution does not contain
626 ;; test data: https://github.com/biocore/biom-format/issues/693
627 (uri (string-append "https://github.com/biocore/biom-format/archive/"
628 version ".tar.gz"))
629 (file-name (string-append name "-" version ".tar.gz"))
630 (sha256
631 (base32
abc08cba 632 "08cr7wpahk6zb31h4bs7jmzpvxcqv9s13xz40h6y2h656jvdvnpj"))))
9e12eba8 633 (build-system python-build-system)
de96ea28 634 (propagated-inputs
9e12eba8
BW		 635 `(("python-numpy" ,python-numpy)
636 ("python-scipy" ,python-scipy)
637 ("python-future" ,python-future)
638 ("python-click" ,python-click)
abc08cba
BW		 639 ("python-h5py" ,python-h5py)
640 ("python-pandas" ,python-pandas)))
da5ebd10
MB		 641 (native-inputs
642 `(("python-nose" ,python-nose)))
9e12eba8
BW		 643 (home-page "http://www.biom-format.org")
644 (synopsis "Biological Observation Matrix (BIOM) format utilities")
645 (description
646 "The BIOM file format is designed to be a general-use format for
647representing counts of observations e.g. operational taxonomic units, KEGG
648orthology groups or lipid types, in one or more biological samples
649e.g. microbiome samples, genomes, metagenomes.")
650 (license license:bsd-3)
651 (properties `((python2-variant . ,(delay python2-biom-format))))))
652
653(define-public python2-biom-format
654 (let ((base (package-with-python2 (strip-python2-variant python-biom-format))))
655 (package
656 (inherit base)
657 (arguments
658 `(#:phases
659 (modify-phases %standard-phases
660 ;; Do not require the unmaintained pyqi library.
661 (add-after 'unpack 'remove-pyqi
662 (lambda _
663 (substitute* "setup.py"
664 (("install_requires.append\\(\"pyqi\"\\)") "pass"))
665 #t)))
00e10c6e 666 ,@(package-arguments base))))))
9e12eba8 667
f7283db3
RW		 668(define-public bioperl-minimal
669 (let* ((inputs `(("perl-module-build" ,perl-module-build)
670 ("perl-data-stag" ,perl-data-stag)
671 ("perl-libwww" ,perl-libwww)
672 ("perl-uri" ,perl-uri)))
673 (transitive-inputs
674 (map (compose package-name cadr)
675 (delete-duplicates
676 (concatenate
677 (map (compose package-transitive-target-inputs cadr) inputs))))))
678 (package
679 (name "bioperl-minimal")
c70271ec 680 (version "1.7.0")
f7283db3
RW		 681 (source
682 (origin
683 (method url-fetch)
c70271ec
RW		 684 (uri (string-append "https://github.com/bioperl/bioperl-live/"
685 "archive/release-"
686 (string-map (lambda (c)
687 (if (char=? c #\.)
688 #\- c)) version)
689 ".tar.gz"))
f7283db3
RW		 690 (sha256
691 (base32
c70271ec 692 "12phgpxwgkqflkwfb9dcqg7a31dpjlfhar8wcgv0aj5ln4akfz06"))))
f7283db3
RW		 693 (build-system perl-build-system)
694 (arguments
695 `(#:phases
696 (modify-phases %standard-phases
697 (add-after
698 'install 'wrap-programs
699 (lambda* (#:key outputs #:allow-other-keys)
700 ;; Make sure all executables in "bin" find the required Perl
701 ;; modules at runtime. As the PERL5LIB variable contains also
702 ;; the paths of native inputs, we pick the transitive target
703 ;; inputs from %build-inputs.
704 (let* ((out (assoc-ref outputs "out"))
705 (bin (string-append out "/bin/"))
706 (path (string-join
707 (cons (string-append out "/lib/perl5/site_perl")
708 (map (lambda (name)
709 (assoc-ref %build-inputs name))
710 ',transitive-inputs))
711 ":")))
712 (for-each (lambda (file)
713 (wrap-program file
714 `("PERL5LIB" ":" prefix (,path))))
715 (find-files bin "\\.pl$"))
716 #t))))))
717 (inputs inputs)
718 (native-inputs
719 `(("perl-test-most" ,perl-test-most)))
720 (home-page "http://search.cpan.org/dist/BioPerl")
721 (synopsis "Bioinformatics toolkit")
722 (description
723 "BioPerl is the product of a community effort to produce Perl code which
724is useful in biology. Examples include Sequence objects, Alignment objects
725and database searching objects. These objects not only do what they are
726advertised to do in the documentation, but they also interact - Alignment
727objects are made from the Sequence objects, Sequence objects have access to
728Annotation and SeqFeature objects and databases, Blast objects can be
729converted to Alignment objects, and so on. This means that the objects
730provide a coordinated and extensible framework to do computational biology.")
2f3108ad 731 (license license:perl-license))))
f7283db3 732
85c37e29
RW		 733(define-public python-biopython
734 (package
735 (name "python-biopython")
af6ce610 736 (version "1.70")
85c37e29
RW		 737 (source (origin
738 (method url-fetch)
e815c094
BW		 739 ;; use PyPi rather than biopython.org to ease updating
740 (uri (pypi-uri "biopython" version))
85c37e29
RW		 741 (sha256
742 (base32
af6ce610 743 "0nz4n9d2y2dg849gn1z0vjlkwcpzzkzy3fij7x94a6ixy2c54z2a"))))
85c37e29 744 (build-system python-build-system)
4ce60305
BW		 745 (arguments
746 `(#:phases
747 (modify-phases %standard-phases
748 (add-before 'check 'set-home
749 ;; Some tests require a home directory to be set.
750 (lambda _ (setenv "HOME" "/tmp") #t)))))
f22efa01 751 (propagated-inputs
85c37e29 752 `(("python-numpy" ,python-numpy)))
85c37e29
RW		 753 (home-page "http://biopython.org/")
754 (synopsis "Tools for biological computation in Python")
755 (description
756 "Biopython is a set of tools for biological computation including parsers
757for bioinformatics files into Python data structures; interfaces to common
758bioinformatics programs; a standard sequence class and tools for performing
759common operations on them; code to perform data classification; code for
760dealing with alignments; code making it easy to split up parallelizable tasks
761into separate processes; and more.")
5c31f4aa 762 (license (license:non-copyleft "http://www.biopython.org/DIST/LICENSE"))))
85c37e29
RW		 763
764(define-public python2-biopython
5c31f4aa 765 (package-with-python2 python-biopython))
85c37e29 766
4b1a1528
BW		 767;; An outdated version of biopython is required for seqmagick, see
768;; https://github.com/fhcrc/seqmagick/issues/59
769;; When that issue has been resolved this package should be removed.
770(define python2-biopython-1.66
771 (package
772 (inherit python2-biopython)
773 (version "1.66")
774 (source (origin
775 (method url-fetch)
776 (uri (pypi-uri "biopython" version))
777 (sha256
778 (base32
779 "1gdv92593klimg22icf5j9by7xiq86jnwzkpz4abaa05ylkdf6hp"))))))
780
985d8411
BW		 781(define-public bpp-core
782 ;; The last release was in 2014 and the recommended way to install from source
783 ;; is to clone the git repository, so we do this.
784 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
785 (let ((commit "7d8bced0d1a87291ea8dd7046b7fb5ff9c35c582"))
786 (package
787 (name "bpp-core")
788 (version (string-append "2.2.0-1." (string-take commit 7)))
789 (source (origin
790 (method git-fetch)
791 (uri (git-reference
792 (url "http://biopp.univ-montp2.fr/git/bpp-core")
793 (commit commit)))
794 (file-name (string-append name "-" version "-checkout"))
795 (sha256
796 (base32
797 "10djsq5vlnkilv436gnmh4irpk49v29pa69r6xiryg32xmvn909j"))))
798 (build-system cmake-build-system)
799 (arguments
800 `(#:parallel-build? #f))
801 (inputs
802 `(("gcc" ,gcc-5))) ; Compilation of bpp-phyl fails with GCC 4.9 so we
803 ; compile all of the bpp packages with GCC 5.
804 (home-page "http://biopp.univ-montp2.fr")
805 (synopsis "C++ libraries for Bioinformatics")
806 (description
807 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
808analysis, phylogenetics, molecular evolution and population genetics. It is
809Object Oriented and is designed to be both easy to use and computer efficient.
810Bio++ intends to help programmers to write computer expensive programs, by
811providing them a set of re-usable tools.")
812 (license license:cecill-c))))
813
8b5f4d57
BW		 814(define-public bpp-phyl
815 ;; The last release was in 2014 and the recommended way to install from source
816 ;; is to clone the git repository, so we do this.
817 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
818 (let ((commit "0c07167b629f68b569bf274d1ad0c4af83276ae2"))
819 (package
820 (name "bpp-phyl")
821 (version (string-append "2.2.0-1." (string-take commit 7)))
822 (source (origin
823 (method git-fetch)
824 (uri (git-reference
825 (url "http://biopp.univ-montp2.fr/git/bpp-phyl")
826 (commit commit)))
827 (file-name (string-append name "-" version "-checkout"))
828 (sha256
829 (base32
830 "1ssjgchzwj3iai26kyly7gwkdv8sk59nqhkb1wpap3sf5m6kyllh"))))
831 (build-system cmake-build-system)
832 (arguments
833 `(#:parallel-build? #f
834 ;; If out-of-source, test data is not copied into the build directory
835 ;; so the tests fail.
836 #:out-of-source? #f))
837 (inputs
838 `(("bpp-core" ,bpp-core)
839 ("bpp-seq" ,bpp-seq)
840 ;; GCC 4.8 fails due to an 'internal compiler error', so we use a more
841 ;; modern GCC.
842 ("gcc" ,gcc-5)))
843 (home-page "http://biopp.univ-montp2.fr")
844 (synopsis "Bio++ phylogenetic Library")
845 (description
846 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
847analysis, phylogenetics, molecular evolution and population genetics. This
848library provides phylogenetics-related modules.")
849 (license license:cecill-c))))
850
159a7016
BW		 851(define-public bpp-popgen
852 ;; The last release was in 2014 and the recommended way to install from source
853 ;; is to clone the git repository, so we do this.
854 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
855 (let ((commit "e472bac9b1a148803895d747cd6d0c5904f85d9f"))
856 (package
857 (name "bpp-popgen")
858 (version (string-append "2.2.0-1." (string-take commit 7)))
859 (source (origin
860 (method git-fetch)
861 (uri (git-reference
862 (url "http://biopp.univ-montp2.fr/git/bpp-popgen")
863 (commit commit)))
864 (file-name (string-append name "-" version "-checkout"))
865 (sha256
866 (base32
867 "0yn82dzn1n5629nzja68xfrhi655709rjanyryb36vzkmymy6dw5"))))
868 (build-system cmake-build-system)
869 (arguments
870 `(#:parallel-build? #f
871 #:tests? #f)) ; There are no tests.
872 (inputs
873 `(("bpp-core" ,bpp-core)
874 ("bpp-seq" ,bpp-seq)
875 ("gcc" ,gcc-5)))
876 (home-page "http://biopp.univ-montp2.fr")
877 (synopsis "Bio++ population genetics library")
878 (description
879 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
880analysis, phylogenetics, molecular evolution and population genetics. This
881library provides population genetics-related modules.")
882 (license license:cecill-c))))
883
70f1bc05
BW		 884(define-public bpp-seq
885 ;; The last release was in 2014 and the recommended way to install from source
886 ;; is to clone the git repository, so we do this.
887 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
888 (let ((commit "6cfa07965ce152e5598a89df2fa80a75973bfa33"))
889 (package
890 (name "bpp-seq")
891 (version (string-append "2.2.0-1." (string-take commit 7)))
892 (source (origin
893 (method git-fetch)
894 (uri (git-reference
895 (url "http://biopp.univ-montp2.fr/git/bpp-seq")
896 (commit commit)))
897 (file-name (string-append name "-" version "-checkout"))
898 (sha256
899 (base32
900 "1nys5jq7jqvdg40d91wsmj3q2yzy4276cp7sp44n67p468f27zf2"))))
901 (build-system cmake-build-system)
902 (arguments
903 `(#:parallel-build? #f
904 ;; If out-of-source, test data is not copied into the build directory
905 ;; so the tests fail.
906 #:out-of-source? #f))
907 (inputs
908 `(("bpp-core" ,bpp-core)
909 ("gcc" ,gcc-5))) ; Use GCC 5 as per 'bpp-core'.
910 (home-page "http://biopp.univ-montp2.fr")
911 (synopsis "Bio++ sequence library")
912 (description
913 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
914analysis, phylogenetics, molecular evolution and population genetics. This
915library provides sequence-related modules.")
916 (license license:cecill-c))))
917
db7a3444
BW		 918(define-public bppsuite
919 ;; The last release was in 2014 and the recommended way to install from source
920 ;; is to clone the git repository, so we do this.
921 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
922 (let ((commit "c516147f57aa50961121cd505bed52cd7603698b"))
923 (package
924 (name "bppsuite")
925 (version (string-append "2.2.0-1." (string-take commit 7)))
926 (source (origin
927 (method git-fetch)
928 (uri (git-reference
929 (url "http://biopp.univ-montp2.fr/git/bppsuite")
930 (commit commit)))
931 (file-name (string-append name "-" version "-checkout"))
932 (sha256
933 (base32
934 "1y87pxvw0jxjizhq2dr9g2r91md45k1p9ih2sl1yy1y3p934l2kb"))))
935 (build-system cmake-build-system)
936 (arguments
937 `(#:parallel-build? #f
938 #:tests? #f)) ; There are no tests.
939 (native-inputs
940 `(("groff" ,groff)
941 ("man-db" ,man-db)
942 ("texinfo" ,texinfo)))
943 (inputs
944 `(("bpp-core" ,bpp-core)
945 ("bpp-seq" ,bpp-seq)
946 ("bpp-phyl" ,bpp-phyl)
947 ("bpp-phyl" ,bpp-popgen)
948 ("gcc" ,gcc-5)))
949 (home-page "http://biopp.univ-montp2.fr")
950 (synopsis "Bioinformatics tools written with the Bio++ libraries")
951 (description
952 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
953analysis, phylogenetics, molecular evolution and population genetics. This
954package provides command line tools using the Bio++ library.")
955 (license license:cecill-c))))
956
82c370de
RW		 957(define-public blast+
958 (package
959 (name "blast+")
4732e6ee 960 (version "2.6.0")
82c370de
RW		 961 (source (origin
962 (method url-fetch)
963 (uri (string-append
964 "ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/"
965 version "/ncbi-blast-" version "+-src.tar.gz"))
966 (sha256
967 (base32
4732e6ee
BW		 968 "15n937pw5aqmyfjb6l387d18grqbb96l63d5xj4l7yyh0zbf2405"))
969 (patches (search-patches "blast+-fix-makefile.patch"))
82c370de
RW		 970 (modules '((guix build utils)))
971 (snippet
972 '(begin
4732e6ee 973 ;; Remove bundled bzip2, zlib and pcre.
82c370de
RW		 974 (delete-file-recursively "c++/src/util/compress/bzip2")
975 (delete-file-recursively "c++/src/util/compress/zlib")
4732e6ee 976 (delete-file-recursively "c++/src/util/regexp")
82c370de
RW		 977 (substitute* "c++/src/util/compress/Makefile.in"
978 (("bzip2 zlib api") "api"))
979 ;; Remove useless msbuild directory
980 (delete-file-recursively
981 "c++/src/build-system/project_tree_builder/msbuild")
982 #t))))
983 (build-system gnu-build-system)
984 (arguments
4732e6ee 985 `(;; There are two(!) tests for this massive library, and both fail with
82c370de 986 ;; "unparsable timing stats".
82c370de
RW		 987 ;; ERR [127] -- [serial/datatool] datatool.sh (unparsable timing stats)
988 ;; ERR [127] -- [serial/datatool] datatool_xml.sh (unparsable timing stats)
989 #:tests? #f
990 #:out-of-source? #t
991 #:parallel-build? #f ; not supported
992 #:phases
993 (modify-phases %standard-phases
994 (add-before
995 'configure 'set-HOME
996 ;; $HOME needs to be set at some point during the configure phase
997 (lambda _ (setenv "HOME" "/tmp") #t))
998 (add-after
999 'unpack 'enter-dir
1000 (lambda _ (chdir "c++") #t))
1001 (add-after
1002 'enter-dir 'fix-build-system
1003 (lambda _
1004 (define (which* cmd)
1005 (cond ((string=? cmd "date")
1006 ;; make call to "date" deterministic
1007 "date -d @0")
1008 ((which cmd)
1009 => identity)
1010 (else
1011 (format (current-error-port)
1012 "WARNING: Unable to find absolute path for ~s~%"
1013 cmd)
1014 #f)))
1015
1016 ;; Rewrite hardcoded paths to various tools
1017 (substitute* (append '("src/build-system/configure.ac"
1018 "src/build-system/configure"
4732e6ee 1019 "src/build-system/helpers/run_with_lock.c"
82c370de
RW		 1020 "scripts/common/impl/if_diff.sh"
1021 "scripts/common/impl/run_with_lock.sh"
1022 "src/build-system/Makefile.configurables.real"
1023 "src/build-system/Makefile.in.top"
1024 "src/build-system/Makefile.meta.gmake=no"
1025 "src/build-system/Makefile.meta.in"
1026 "src/build-system/Makefile.meta_l"
1027 "src/build-system/Makefile.meta_p"
1028 "src/build-system/Makefile.meta_r"
1029 "src/build-system/Makefile.mk.in"
1030 "src/build-system/Makefile.requirements"
1031 "src/build-system/Makefile.rules_with_autodep.in")
1032 (find-files "scripts/common/check" "\\.sh$"))
1033 (("(/usr/bin/|/bin/)([a-z][-_.a-z]*)" all dir cmd)
1034 (or (which* cmd) all)))
1035
1036 (substitute* (find-files "src/build-system" "^config.*")
1037 (("LN_S=/bin/\\$LN_S") (string-append "LN_S=" (which "ln")))
1038 (("^PATH=.*") ""))
1039
1040 ;; rewrite "/var/tmp" in check script
1041 (substitute* "scripts/common/check/check_make_unix.sh"
1042 (("/var/tmp") "/tmp"))
1043
1044 ;; do not reset PATH
1045 (substitute* (find-files "scripts/common/impl/" "\\.sh$")
1046 (("^ *PATH=.*") "")
1047 (("action=/bin/") "action=")
1048 (("export PATH") ":"))
1049 #t))
1050 (replace
1051 'configure
1052 (lambda* (#:key inputs outputs #:allow-other-keys)
1053 (let ((out (assoc-ref outputs "out"))
1054 (lib (string-append (assoc-ref outputs "lib") "/lib"))
1055 (include (string-append (assoc-ref outputs "include")
1056 "/include/ncbi-tools++")))
1057 ;; The 'configure' script doesn't recognize things like
1058 ;; '--enable-fast-install'.
1059 (zero? (system* "./configure.orig"
1060 (string-append "--with-build-root=" (getcwd) "/build")
1061 (string-append "--prefix=" out)
1062 (string-append "--libdir=" lib)
1063 (string-append "--includedir=" include)
1064 (string-append "--with-bz2="
1065 (assoc-ref inputs "bzip2"))
1066 (string-append "--with-z="
1067 (assoc-ref inputs "zlib"))
4732e6ee
BW		 1068 (string-append "--with-pcre="
1069 (assoc-ref inputs "pcre"))
82c370de
RW		 1070 ;; Each library is built twice by default, once
1071 ;; with "-static" in its name, and again
1072 ;; without.
1073 "--without-static"
1074 "--with-dll"))))))))
4732e6ee
BW		 1075 (outputs '("out" ; 21 MB
1076 "lib" ; 226 MB
1077 "include")) ; 33 MB
82c370de
RW		 1078 (inputs
1079 `(("bzip2" ,bzip2)
4732e6ee
BW		 1080 ("zlib" ,zlib)
1081 ("pcre" ,pcre)
1082 ("perl" ,perl)
1083 ("python" ,python-wrapper)))
82c370de
RW		 1084 (native-inputs
1085 `(("cpio" ,cpio)))
1086 (home-page "http://blast.ncbi.nlm.nih.gov")
1087 (synopsis "Basic local alignment search tool")
1088 (description
1089 "BLAST is a popular method of performing a DNA or protein sequence
1090similarity search, using heuristics to produce results quickly. It also
1091calculates an “expect value” that estimates how many matches would have
1092occurred at a given score by chance, which can aid a user in judging how much
1093confidence to have in an alignment.")
1094 ;; Most of the sources are in the public domain, with the following
1095 ;; exceptions:
1096 ;; * Expat:
1097 ;; * ./c++/include/util/bitset/
1098 ;; * ./c++/src/html/ncbi_menu*.js
1099 ;; * Boost license:
1100 ;; * ./c++/include/util/impl/floating_point_comparison.hpp
1101 ;; * LGPL 2+:
1102 ;; * ./c++/include/dbapi/driver/odbc/unix_odbc/
1103 ;; * ASL 2.0:
1104 ;; * ./c++/src/corelib/teamcity_*
1105 (license (list license:public-domain
1106 license:expat
1107 license:boost1.0
1108 license:lgpl2.0+
1109 license:asl2.0))))
1110
6c2b26e2
RW		 1111(define-public bless
1112 (package
1113 (name "bless")
1114 (version "1p02")
1115 (source (origin
1116 (method url-fetch)
1117 (uri (string-append "mirror://sourceforge/bless-ec/bless.v"
1118 version ".tgz"))
1119 (sha256
1120 (base32
4d75e03a
RW		 1121 "0rm0gw2s18dqwzzpl3c2x1z05ni2v0xz5dmfk3d33j6g4cgrlrdd"))
1122 (modules '((guix build utils)))
6c2b26e2
RW		 1123 (snippet
1124 `(begin
1125 ;; Remove bundled boost, pigz, zlib, and .git directory
953c1223
RW		 1126 ;; FIXME: also remove bundled sources for murmurhash3 and
1127 ;; kmc once packaged.
6c2b26e2
RW		 1128 (delete-file-recursively "boost")
1129 (delete-file-recursively "pigz")
953c1223 1130 (delete-file-recursively "google-sparsehash")
6c2b26e2
RW		 1131 (delete-file-recursively "zlib")
1132 (delete-file-recursively ".git")
1133 #t))))
1134 (build-system gnu-build-system)
1135 (arguments
1136 '(#:tests? #f ;no "check" target
1137 #:make-flags
1138 (list (string-append "ZLIB="
1139 (assoc-ref %build-inputs "zlib")
1140 "/lib/libz.a")
1141 (string-append "LDFLAGS="
1142 (string-join '("-lboost_filesystem"
1143 "-lboost_system"
1144 "-lboost_iostreams"
1145 "-lz"
1146 "-fopenmp"
1147 "-std=c++11"))))
1148 #:phases
1149 (modify-phases %standard-phases
1150 (add-after 'unpack 'do-not-build-bundled-pigz
1151 (lambda* (#:key inputs outputs #:allow-other-keys)
1152 (substitute* "Makefile"
1153 (("cd pigz/pigz-2.3.3; make") ""))
1154 #t))
1155 (add-after 'unpack 'patch-paths-to-executables
1156 (lambda* (#:key inputs outputs #:allow-other-keys)
1157 (substitute* "parse_args.cpp"
1158 (("kmc_binary = .*")
1159 (string-append "kmc_binary = \""
1160 (assoc-ref outputs "out")
1161 "/bin/kmc\";"))
1162 (("pigz_binary = .*")
1163 (string-append "pigz_binary = \""
1164 (assoc-ref inputs "pigz")
1165 "/bin/pigz\";")))
1166 #t))
1167 (replace 'install
1168 (lambda* (#:key outputs #:allow-other-keys)
1169 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
1170 (for-each (lambda (file)
1171 (install-file file bin))
1172 '("bless" "kmc/bin/kmc"))
1173 #t)))
1174 (delete 'configure))))
1175 (native-inputs
1176 `(("perl" ,perl)))
1177 (inputs
1178 `(("openmpi" ,openmpi)
1179 ("boost" ,boost)
953c1223 1180 ("sparsehash" ,sparsehash)
6c2b26e2
RW		 1181 ("pigz" ,pigz)
1182 ("zlib" ,zlib)))
9641a899 1183 (supported-systems '("x86_64-linux"))
3b3b60d0 1184 (home-page "https://sourceforge.net/p/bless-ec/wiki/Home/")
6c2b26e2
RW		 1185 (synopsis "Bloom-filter-based error correction tool for NGS reads")
1186 (description
1187 "@dfn{Bloom-filter-based error correction solution for high-throughput
1188sequencing reads} (BLESS) uses a single minimum-sized bloom filter is a
1189correction tool for genomic reads produced by @dfn{Next-generation
1190sequencing} (NGS). BLESS produces accurate correction results with much less
1191memory compared with previous solutions and is also able to tolerate a higher
1192false-positive rate. BLESS can extend reads like DNA assemblers to correct
1193errors at the end of reads.")
1194 (license license:gpl3+)))
1195
2c7ee167
RW		 1196(define-public bowtie
1197 (package
1198 (name "bowtie")
d6e63cf3 1199 (version "2.3.2")
2c7ee167
RW		 1200 (source (origin
1201 (method url-fetch)
1202 (uri (string-append "https://github.com/BenLangmead/bowtie2/archive/v"
1203 version ".tar.gz"))
f586c877 1204 (file-name (string-append name "-" version ".tar.gz"))
2c7ee167
RW		 1205 (sha256
1206 (base32
d6e63cf3 1207 "0hwa5r9qbglppb7sz5z79rlmmddr3n51n468jb3wh8rwjgn3yr90"))
2c7ee167
RW		 1208 (modules '((guix build utils)))
1209 (snippet
1210 '(substitute* "Makefile"
2c7ee167
RW		 1211 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
1212 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
0047d26a 1213 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\"")))))
2c7ee167 1214 (build-system gnu-build-system)
d6e63cf3
BW		 1215 (inputs
1216 `(("perl" ,perl)
1217 ("perl-clone" ,perl-clone)
1218 ("perl-test-deep" ,perl-test-deep)
1219 ("perl-test-simple" ,perl-test-simple)
1220 ("python" ,python-2)
1221 ("tbb" ,tbb)
1222 ("zlib" ,zlib)))
2c7ee167 1223 (arguments
0047d26a
RW		 1224 '(#:make-flags
1225 (list "allall"
1226 "WITH_TBB=1"
1227 (string-append "prefix=" (assoc-ref %outputs "out")))
2c7ee167 1228 #:phases
06e37236
BW		 1229 (modify-phases %standard-phases
1230 (delete 'configure)
1231 (replace 'check
1232 (lambda* (#:key outputs #:allow-other-keys)
1233 (zero? (system* "perl"
1234 "scripts/test/simple_tests.pl"
1235 "--bowtie2=./bowtie2"
1236 "--bowtie2-build=./bowtie2-build")))))))
2c7ee167
RW		 1237 (home-page "http://bowtie-bio.sourceforge.net/bowtie2/index.shtml")
1238 (synopsis "Fast and sensitive nucleotide sequence read aligner")
1239 (description
1240 "Bowtie 2 is a fast and memory-efficient tool for aligning sequencing
1241reads to long reference sequences. It is particularly good at aligning reads
1242of about 50 up to 100s or 1,000s of characters, and particularly good at
1243aligning to relatively long (e.g. mammalian) genomes. Bowtie 2 indexes the
1244genome with an FM Index to keep its memory footprint small: for the human
1245genome, its memory footprint is typically around 3.2 GB. Bowtie 2 supports
1246gapped, local, and paired-end alignment modes.")
241e1221 1247 (supported-systems '("x86_64-linux"))
2c7ee167
RW		 1248 (license license:gpl3+)))
1249
94ce537e
RW		 1250(define-public tophat
1251 (package
1252 (name "tophat")
1253 (version "2.1.0")
1254 (source (origin
1255 (method url-fetch)
1256 (uri (string-append
1257 "http://ccb.jhu.edu/software/tophat/downloads/tophat-"
1258 version ".tar.gz"))
1259 (sha256
1260 (base32
1261 "168zlzykq622zbgkh90a90f1bdgsxkscq2zxzbj8brq80hbjpyp7"))
fc1adab1 1262 (patches (search-patches "tophat-build-with-later-seqan.patch"))
94ce537e
RW		 1263 (modules '((guix build utils)))
1264 (snippet
1265 '(begin
1266 ;; Remove bundled SeqAn and samtools
1267 (delete-file-recursively "src/SeqAn-1.3")
1268 (delete-file-recursively "src/samtools-0.1.18")
1269 #t))))
1270 (build-system gnu-build-system)
1271 (arguments
1272 '(#:parallel-build? #f ; not supported
1273 #:phases
1274 (modify-phases %standard-phases
1275 (add-after 'unpack 'use-system-samtools
1276 (lambda* (#:key inputs #:allow-other-keys)
1277 (substitute* "src/Makefile.in"
1278 (("(noinst_LIBRARIES = )\\$\\(SAMLIB\\)" _ prefix) prefix)
1279 (("\\$\\(SAMPROG\\): \\$\\(SAMLIB\\)") "")
1280 (("SAMPROG = samtools_0\\.1\\.18") "")
1281 (("\\$\\(samtools_0_1_18_SOURCES\\)") "")
1282 (("am__EXEEXT_1 = samtools_0\\.1\\.18\\$\\(EXEEXT\\)") ""))
1283 (substitute* '("src/common.cpp"
1284 "src/tophat.py")
1285 (("samtools_0.1.18") (which "samtools")))
1286 (substitute* '("src/common.h"
1287 "src/bam2fastx.cpp")
1288 (("#include \"bam.h\"") "#include <samtools/bam.h>")
1289 (("#include \"sam.h\"") "#include <samtools/sam.h>"))
1290 (substitute* '("src/bwt_map.h"
1291 "src/map2gtf.h"
1292 "src/align_status.h")
1293 (("#include <bam.h>") "#include <samtools/bam.h>")
1294 (("#include <sam.h>") "#include <samtools/sam.h>"))
1295 #t)))))
1296 (inputs
1297 `(("boost" ,boost)
1298 ("bowtie" ,bowtie)
1299 ("samtools" ,samtools-0.1)
1300 ("ncurses" ,ncurses)
1301 ("python" ,python-2)
1302 ("perl" ,perl)
1303 ("zlib" ,zlib)
1304 ("seqan" ,seqan)))
1305 (home-page "http://ccb.jhu.edu/software/tophat/index.shtml")
1306 (synopsis "Spliced read mapper for RNA-Seq data")
1307 (description
1308 "TopHat is a fast splice junction mapper for nucleotide sequence
1309reads produced by the RNA-Seq method. It aligns RNA-Seq reads to
1310mammalian-sized genomes using the ultra high-throughput short read
1311aligner Bowtie, and then analyzes the mapping results to identify
1312splice junctions between exons.")
1313 ;; TopHat is released under the Boost Software License, Version 1.0
1314 ;; See https://github.com/infphilo/tophat/issues/11#issuecomment-121589893
1315 (license license:boost1.0)))
1316
9a8336d8
RW		 1317(define-public bwa
1318 (package
1319 (name "bwa")
6f141eff 1320 (version "0.7.17")
9a8336d8
RW		 1321 (source (origin
1322 (method url-fetch)
ae6e00f6
BW		 1323 (uri (string-append
1324 "https://github.com/lh3/bwa/releases/download/v"
1325 version "/bwa-" version ".tar.bz2"))
9a8336d8
RW		 1326 (sha256
1327 (base32
6f141eff 1328 "1zfhv2zg9v1icdlq4p9ssc8k01mca5d1bd87w71py2swfi74s6yy"))))
9a8336d8
RW		 1329 (build-system gnu-build-system)
1330 (arguments
1331 '(#:tests? #f ;no "check" target
1332 #:phases
dc1d3cde
KK		 1333 (modify-phases %standard-phases
1334 (replace 'install
1335 (lambda* (#:key outputs #:allow-other-keys)
1336 (let ((bin (string-append
1337 (assoc-ref outputs "out") "/bin"))
1338 (doc (string-append
1339 (assoc-ref outputs "out") "/share/doc/bwa"))
1340 (man (string-append
1341 (assoc-ref outputs "out") "/share/man/man1")))
1342 (install-file "bwa" bin)
1343 (install-file "README.md" doc)
1344 (install-file "bwa.1" man))
1345 #t))
1346 ;; no "configure" script
1347 (delete 'configure))))
9a8336d8 1348 (inputs `(("zlib" ,zlib)))
db94f8c7
RW		 1349 ;; Non-portable SSE instructions are used so building fails on platforms
1350 ;; other than x86_64.
1351 (supported-systems '("x86_64-linux"))
9a8336d8
RW		 1352 (home-page "http://bio-bwa.sourceforge.net/")
1353 (synopsis "Burrows-Wheeler sequence aligner")
1354 (description
1355 "BWA is a software package for mapping low-divergent sequences against a
1356large reference genome, such as the human genome. It consists of three
1357algorithms: BWA-backtrack, BWA-SW and BWA-MEM. The first algorithm is
1358designed for Illumina sequence reads up to 100bp, while the rest two for
1359longer sequences ranged from 70bp to 1Mbp. BWA-MEM and BWA-SW share similar
1360features such as long-read support and split alignment, but BWA-MEM, which is
1361the latest, is generally recommended for high-quality queries as it is faster
1362and more accurate. BWA-MEM also has better performance than BWA-backtrack for
136370-100bp Illumina reads.")
1364 (license license:gpl3+)))
1365
d29150b5
RW		 1366(define-public bwa-pssm
1367 (package (inherit bwa)
1368 (name "bwa-pssm")
1369 (version "0.5.11")
1370 (source (origin
1371 (method url-fetch)
1372 (uri (string-append "https://github.com/pkerpedjiev/bwa-pssm/"
1373 "archive/" version ".tar.gz"))
1374 (file-name (string-append name "-" version ".tar.gz"))
1375 (sha256
1376 (base32
1377 "02p7mpbs4mlxmn84g2x4ghak638vbj4lqix2ipx5g84pz9bhdavg"))))
1378 (build-system gnu-build-system)
1379 (inputs
1380 `(("gdsl" ,gdsl)
1381 ("zlib" ,zlib)
1382 ("perl" ,perl)))
1383 (home-page "http://bwa-pssm.binf.ku.dk/")
1384 (synopsis "Burrows-Wheeler transform-based probabilistic short read mapper")
1385 (description
1386 "BWA-PSSM is a probabilistic short genomic sequence read aligner based on
1387the use of @dfn{position specific scoring matrices} (PSSM). Like many of the
1388existing aligners it is fast and sensitive. Unlike most other aligners,
1389however, it is also adaptible in the sense that one can direct the alignment
1390based on known biases within the data set. It is coded as a modification of
1391the original BWA alignment program and shares the genome index structure as
1392well as many of the command line options.")
1393 (license license:gpl3+)))
1394
ad641d53
RW		 1395(define-public python2-bx-python
1396 (package
1397 (name "python2-bx-python")
c1dfe8c3 1398 (version "0.7.3")
ad641d53
RW		 1399 (source (origin
1400 (method url-fetch)
c1dfe8c3 1401 (uri (pypi-uri "bx-python" version))
ad641d53
RW		 1402 (sha256
1403 (base32
c1dfe8c3 1404 "15z2w3bvnc0n4qmb9bd6d8ylc2h2nj883x2w9iixf4x3vki9b22i"))
ad641d53
RW		 1405 (modules '((guix build utils)))
1406 (snippet
1407 '(substitute* "setup.py"
1408 ;; remove dependency on outdated "distribute" module
1409 (("^from distribute_setup import use_setuptools") "")
1410 (("^use_setuptools\\(\\)") "")))))
1411 (build-system python-build-system)
1412 (arguments
1413 `(#:tests? #f ;tests fail because test data are not included
1414 #:python ,python-2))
1415 (inputs
1416 `(("python-numpy" ,python2-numpy)
1417 ("zlib" ,zlib)))
1418 (native-inputs
f3b98f4f 1419 `(("python-nose" ,python2-nose)))
ad641d53
RW		 1420 (home-page "http://bitbucket.org/james_taylor/bx-python/")
1421 (synopsis "Tools for manipulating biological data")
1422 (description
1423 "bx-python provides tools for manipulating biological data, particularly
1424multiple sequence alignments.")
1425 (license license:expat)))
1426
55a9a8c2
RW		 1427(define-public python-pysam
1428 (package
1429 (name "python-pysam")
60739965 1430 (version "0.11.2.2")
d454640c
RW		 1431 (source (origin
1432 (method url-fetch)
f536dce5
MB		 1433 ;; Test data is missing on PyPi.
1434 (uri (string-append
1435 "https://github.com/pysam-developers/pysam/archive/v"
1436 version ".tar.gz"))
1437 (file-name (string-append name "-" version ".tar.gz"))
d454640c
RW		 1438 (sha256
1439 (base32
60739965 1440 "1cfqdxsqs3xhacns9n0271ck6wkc76px66ddjm91wfw2jxxfklvc"))
dff26b23
MB		 1441 (modules '((guix build utils)))
1442 (snippet
1443 ;; Drop bundled htslib. TODO: Also remove samtools and bcftools.
1444 '(delete-file-recursively "htslib"))))
55a9a8c2
RW		 1445 (build-system python-build-system)
1446 (arguments
71dbf592
RW		 1447 `(#:modules ((ice-9 ftw)
1448 (srfi srfi-26)
1449 (guix build python-build-system)
1450 (guix build utils))
1451 #:phases
397d463a
MB		 1452 (modify-phases %standard-phases
1453 (add-before 'build 'set-flags
dff26b23
MB		 1454 (lambda* (#:key inputs #:allow-other-keys)
1455 (setenv "HTSLIB_MODE" "external")
1456 (setenv "HTSLIB_LIBRARY_DIR"
1457 (string-append (assoc-ref inputs "htslib") "/lib"))
1458 (setenv "HTSLIB_INCLUDE_DIR"
1459 (string-append (assoc-ref inputs "htslib") "/include"))
397d463a
MB		 1460 (setenv "LDFLAGS" "-lncurses")
1461 (setenv "CFLAGS" "-D_CURSES_LIB=1")
f536dce5 1462 #t))
71dbf592 1463 (replace 'check
f536dce5 1464 (lambda* (#:key inputs outputs #:allow-other-keys)
71dbf592 1465 ;; Add first subdirectory of "build" directory to PYTHONPATH.
f536dce5
MB		 1466 (setenv "PYTHONPATH"
1467 (string-append
1468 (getenv "PYTHONPATH")
71dbf592
RW		 1469 ":" (getcwd) "/build/"
1470 (car (scandir "build"
e1f02f92 1471 (negate (cut string-prefix? "." <>))))))
f536dce5 1472 ;; Step out of source dir so python does not import from CWD.
71dbf592
RW		 1473 (with-directory-excursion "tests"
1474 (setenv "HOME" "/tmp")
1475 (and (zero? (system* "make" "-C" "pysam_data"))
1476 (zero? (system* "make" "-C" "cbcf_data"))
b2955e22
RW		 1477 ;; Running nosetests without explicitly asking for a
1478 ;; single process leads to a crash. Running with multiple
1479 ;; processes fails because the tests are not designed to
1480 ;; run in parallel.
31c374e0
RW		 1481
1482 ;; FIXME: tests keep timing out on some systems.
1483 ;; (zero? (system* "nosetests" "-v"
1484 ;; "--processes" "1"))
1485 )))))))
dff26b23
MB		 1486 (propagated-inputs
1487 `(("htslib" ,htslib))) ; Included from installed header files.
55a9a8c2 1488 (inputs
649e9b3b 1489 `(("ncurses" ,ncurses)
55a9a8c2 1490 ("zlib" ,zlib)))
649e9b3b
RW		 1491 (native-inputs
1492 `(("python-cython" ,python-cython)
f536dce5
MB		 1493 ;; Dependencies below are are for tests only.
1494 ("samtools" ,samtools)
1495 ("bcftools" ,bcftools)
1496 ("python-nose" ,python-nose)))
55a9a8c2
RW		 1497 (home-page "https://github.com/pysam-developers/pysam")
1498 (synopsis "Python bindings to the SAMtools C API")
1499 (description
1500 "Pysam is a Python module for reading and manipulating files in the
1501SAM/BAM format. Pysam is a lightweight wrapper of the SAMtools C API. It
1502also includes an interface for tabix.")
1503 (license license:expat)))
1504
1505(define-public python2-pysam
1506 (package-with-python2 python-pysam))
1507
4db9433a
RW		 1508(define-public python-twobitreader
1509 (package
1510 (name "python-twobitreader")
044ac8d2 1511 (version "3.1.4")
4db9433a
RW		 1512 (source (origin
1513 (method url-fetch)
1514 (uri (pypi-uri "twobitreader" version))
1515 (sha256
1516 (base32
044ac8d2 1517 "1q8wnj2kga9nz1lwc4w7qv52smfm536hp6mc8w6s53lhyj0mpi22"))))
4db9433a 1518 (build-system python-build-system)
900fb8d0
LF		 1519 (arguments
1520 '(;; Tests are not distributed in the PyPi release.
1521 ;; TODO Try building from the Git repo or asking the upstream maintainer
1522 ;; to distribute the tests on PyPi.
1523 #:tests? #f))
4db9433a
RW		 1524 (native-inputs
1525 `(("python-sphinx" ,python-sphinx)))
1526 (home-page "https://github.com/benjschiller/twobitreader")
1527 (synopsis "Python library for reading .2bit files")
1528 (description
1529 "twobitreader is a Python library for reading .2bit files as used by the
1530UCSC genome browser.")
1531 (license license:artistic2.0)))
1532
1533(define-public python2-twobitreader
5c31f4aa 1534 (package-with-python2 python-twobitreader))
4db9433a 1535
f94bf198
RW		 1536(define-public python-plastid
1537 (package
1538 (name "python-plastid")
897ab082 1539 (version "0.4.8")
f94bf198
RW		 1540 (source (origin
1541 (method url-fetch)
1542 (uri (pypi-uri "plastid" version))
1543 (sha256
1544 (base32
897ab082 1545 "0l24dd3q66if8yj042m4s0g95n6acn7im1imqd3p6h8ns43kxhj8"))))
f94bf198
RW		 1546 (build-system python-build-system)
1547 (arguments
1548 ;; Some test files are not included.
1549 `(#:tests? #f))
1550 (propagated-inputs
1551 `(("python-numpy" ,python-numpy)
1552 ("python-scipy" ,python-scipy)
1553 ("python-pandas" ,python-pandas)
1554 ("python-pysam" ,python-pysam)
1555 ("python-matplotlib" ,python-matplotlib)
1556 ("python-biopython" ,python-biopython)
99caa6f7
BW		 1557 ("python-twobitreader" ,python-twobitreader)
1558 ("python-termcolor" ,python-termcolor)))
f94bf198
RW		 1559 (native-inputs
1560 `(("python-cython" ,python-cython)
1561 ("python-nose" ,python-nose)))
1562 (home-page "https://github.com/joshuagryphon/plastid")
1563 (synopsis "Python library for genomic analysis")
1564 (description
1565 "plastid is a Python library for genomic analysis – in particular,
1566high-throughput sequencing data – with an emphasis on simplicity.")
1567 (license license:bsd-3)))
1568
1569(define-public python2-plastid
5c31f4aa 1570 (package-with-python2 python-plastid))
f94bf198 1571
6c1305f9
RW		 1572(define-public cd-hit
1573 (package
1574 (name "cd-hit")
ba773f65 1575 (version "4.6.8")
6c1305f9
RW		 1576 (source (origin
1577 (method url-fetch)
1578 (uri (string-append "https://github.com/weizhongli/cdhit"
1579 "/releases/download/V" version
ba773f65
BW		 1580 "/cd-hit-v" version
1581 "-2017-0621-source.tar.gz"))
6c1305f9
RW		 1582 (sha256
1583 (base32
d4735e8c 1584 "1b4mwm2520ixjbw57sil20f9iixzw4bkdqqwgg1fc3pzm6rz4zmn"))))
6c1305f9
RW		 1585 (build-system gnu-build-system)
1586 (arguments
1587 `(#:tests? #f ; there are no tests
1588 #:make-flags
1589 ;; Executables are copied directly to the PREFIX.
1590 (list (string-append "PREFIX=" (assoc-ref %outputs "out") "/bin"))
1591 #:phases
1592 (modify-phases %standard-phases
1593 ;; No "configure" script
1594 (delete 'configure)
1595 ;; Remove sources of non-determinism
1596 (add-after 'unpack 'be-timeless
1597 (lambda _
1598 (substitute* "cdhit-utility.c++"
1599 ((" \\(built on \" __DATE__ \"\\)") ""))
1600 (substitute* "cdhit-common.c++"
1601 (("__DATE__") "\"0\"")
1602 (("\", %s, \" __TIME__ \"\\\\n\", date") ""))
1603 #t))
ba773f65 1604 ;; The "install" target does not create the target directory.
6c1305f9
RW		 1605 (add-before 'install 'create-target-dir
1606 (lambda* (#:key outputs #:allow-other-keys)
1607 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
1608 #t)))))
1609 (inputs
1610 `(("perl" ,perl)))
1611 (home-page "http://weizhongli-lab.org/cd-hit/")
1612 (synopsis "Cluster and compare protein or nucleotide sequences")
1613 (description
1614 "CD-HIT is a program for clustering and comparing protein or nucleotide
1615sequences. CD-HIT is designed to be fast and handle extremely large
1616databases.")
1617 ;; The manual says: "It can be copied under the GNU General Public License
1618 ;; version 2 (GPLv2)."
1619 (license license:gpl2)))
1620
810cff85
RW		 1621(define-public clipper
1622 (package
1623 (name "clipper")
433530a5 1624 (version "1.1")
810cff85
RW		 1625 (source (origin
1626 (method url-fetch)
1627 (uri (string-append
1628 "https://github.com/YeoLab/clipper/archive/"
1629 version ".tar.gz"))
9ab5ea44 1630 (file-name (string-append name "-" version ".tar.gz"))
810cff85
RW		 1631 (sha256
1632 (base32
433530a5 1633 "0pflmsvhbf8izbgwhbhj1i7349sw1f55qpqj8ljmapp16hb0p0qi"))
810cff85
RW		 1634 (modules '((guix build utils)))
1635 (snippet
433530a5
RW		 1636 '(begin
1637 ;; remove unnecessary setup dependency
1638 (substitute* "setup.py"
1639 (("setup_requires = .*") ""))
1640 (for-each delete-file
1641 '("clipper/src/peaks.so"
1642 "clipper/src/readsToWiggle.so"))
1643 (delete-file-recursively "dist/")
1644 #t))))
810cff85
RW		 1645 (build-system python-build-system)
1646 (arguments `(#:python ,python-2)) ; only Python 2 is supported
1647 (inputs
92971d68 1648 `(("htseq" ,python2-htseq)
810cff85
RW		 1649 ("python-pybedtools" ,python2-pybedtools)
1650 ("python-cython" ,python2-cython)
1651 ("python-scikit-learn" ,python2-scikit-learn)
1652 ("python-matplotlib" ,python2-matplotlib)
433530a5 1653 ("python-pandas" ,python2-pandas)
810cff85
RW		 1654 ("python-pysam" ,python2-pysam)
1655 ("python-numpy" ,python2-numpy)
1656 ("python-scipy" ,python2-scipy)))
1657 (native-inputs
f3b98f4f 1658 `(("python-mock" ,python2-mock) ; for tests
d281be18 1659 ("python-nose" ,python2-nose) ; for tests
f3b98f4f 1660 ("python-pytz" ,python2-pytz))) ; for tests
810cff85
RW		 1661 (home-page "https://github.com/YeoLab/clipper")
1662 (synopsis "CLIP peak enrichment recognition")
1663 (description
1664 "CLIPper is a tool to define peaks in CLIP-seq datasets.")
1665 (license license:gpl2)))
1666
6a35566d
RS		 1667(define-public codingquarry
1668 (package
1669 (name "codingquarry")
1670 (version "2.0")
1671 (source (origin
1672 (method url-fetch)
1673 (uri (string-append
1674 "mirror://sourceforge/codingquarry/CodingQuarry_v"
1675 version ".tar.gz"))
1676 (sha256
1677 (base32
1678 "0115hkjflsnfzn36xppwf9h9avfxlavr43djqmshkkzbgjzsz60i"))))
1679 (build-system gnu-build-system)
1680 (arguments
1681 '(#:tests? #f ; no "check" target
1682 #:phases
1683 (modify-phases %standard-phases
1684 (delete 'configure)
1685 (replace 'install
1686 (lambda* (#:key outputs #:allow-other-keys)
1687 (let* ((out (assoc-ref outputs "out"))
1688 (bin (string-append out "/bin"))
1689 (doc (string-append out "/share/doc/codingquarry")))
1690 (install-file "INSTRUCTIONS.pdf" doc)
1691 (copy-recursively "QuarryFiles"
1692 (string-append out "/QuarryFiles"))
1693 (install-file "CodingQuarry" bin)
1694 (install-file "CufflinksGTF_to_CodingQuarryGFF3.py" bin)))))))
1695 (inputs `(("openmpi" ,openmpi)))
1696 (native-search-paths
1697 (list (search-path-specification
1698 (variable "QUARRY_PATH")
1699 (files '("QuarryFiles")))))
1700 (native-inputs `(("python" ,python-2))) ; Only Python 2 is supported
1701 (synopsis "Fungal gene predictor")
1702 (description "CodingQuarry is a highly accurate, self-training GHMM fungal
1703gene predictor designed to work with assembled, aligned RNA-seq transcripts.")
1704 (home-page "https://sourceforge.net/projects/codingquarry/")
1705 (license license:gpl3+)))
1706
36742f43
RW		 1707(define-public couger
1708 (package
1709 (name "couger")
1710 (version "1.8.2")
1711 (source (origin
1712 (method url-fetch)
1713 (uri (string-append
1714 "http://couger.oit.duke.edu/static/assets/COUGER"
1715 version ".zip"))
1716 (sha256
1717 (base32
1718 "04p2b14nmhzxw5h72mpzdhalv21bx4w9b87z0wpw0xzxpysyncmq"))))
1719 (build-system gnu-build-system)
1720 (arguments
1721 `(#:tests? #f
1722 #:phases
1723 (modify-phases %standard-phases
1724 (delete 'configure)
1725 (delete 'build)
1726 (replace
1727 'install
1728 (lambda* (#:key outputs #:allow-other-keys)
f3860753
TGR		 1729 (let* ((out (assoc-ref outputs "out"))
1730 (bin (string-append out "/bin")))
36742f43 1731 (copy-recursively "src" (string-append out "/src"))
f3860753 1732 (mkdir bin)
36742f43
RW		 1733 ;; Add "src" directory to module lookup path.
1734 (substitute* "couger"
1735 (("from argparse")
1736 (string-append "import sys\nsys.path.append(\""
1737 out "\")\nfrom argparse")))
f3860753 1738 (install-file "couger" bin))
36742f43
RW		 1739 #t))
1740 (add-after
1741 'install 'wrap-program
1742 (lambda* (#:key inputs outputs #:allow-other-keys)
1743 ;; Make sure 'couger' runs with the correct PYTHONPATH.
1744 (let* ((out (assoc-ref outputs "out"))
1745 (path (getenv "PYTHONPATH")))
1746 (wrap-program (string-append out "/bin/couger")
1747 `("PYTHONPATH" ":" prefix (,path))))
1748 #t)))))
1749 (inputs
1750 `(("python" ,python-2)
1751 ("python2-pillow" ,python2-pillow)
1752 ("python2-numpy" ,python2-numpy)
1753 ("python2-scipy" ,python2-scipy)
1754 ("python2-matplotlib" ,python2-matplotlib)))
1755 (propagated-inputs
2d7c4ae3 1756 `(("r-minimal" ,r-minimal)
36742f43
RW		 1757 ("libsvm" ,libsvm)
1758 ("randomjungle" ,randomjungle)))
1759 (native-inputs
1760 `(("unzip" ,unzip)))
1761 (home-page "http://couger.oit.duke.edu")
1762 (synopsis "Identify co-factors in sets of genomic regions")
1763 (description
1764 "COUGER can be applied to any two sets of genomic regions bound by
1765paralogous TFs (e.g., regions derived from ChIP-seq experiments) to identify
1766putative co-factors that provide specificity to each TF. The framework
1767determines the genomic targets uniquely-bound by each TF, and identifies a
1768small set of co-factors that best explain the in vivo binding differences
1769between the two TFs.
1770
1771COUGER uses classification algorithms (support vector machines and random
1772forests) with features that reflect the DNA binding specificities of putative
1773co-factors. The features are generated either from high-throughput TF-DNA
1774binding data (from protein binding microarray experiments), or from large
1775collections of DNA motifs.")
1776 (license license:gpl3+)))
1777
bfe3c685
RW		 1778(define-public clustal-omega
1779 (package
1780 (name "clustal-omega")
b3936f35 1781 (version "1.2.4")
bfe3c685
RW		 1782 (source (origin
1783 (method url-fetch)
b3936f35
RW		 1784 (uri (string-append "http://www.clustal.org/omega/clustal-omega-"
1785 version ".tar.gz"))
bfe3c685
RW		 1786 (sha256
1787 (base32
b3936f35 1788 "1vm30mzncwdv881vrcwg11vzvrsmwy4wg80j5i0lcfk6dlld50w6"))))
bfe3c685
RW		 1789 (build-system gnu-build-system)
1790 (inputs
1791 `(("argtable" ,argtable)))
1792 (home-page "http://www.clustal.org/omega/")
1793 (synopsis "Multiple sequence aligner for protein and DNA/RNA")
1794 (description
1795 "Clustal-Omega is a general purpose multiple sequence alignment (MSA)
1796program for protein and DNA/RNA. It produces high quality MSAs and is capable
1797of handling data-sets of hundreds of thousands of sequences in reasonable
1798time.")
1799 (license license:gpl2+)))
1800
191c7101
RW		 1801(define-public crossmap
1802 (package
1803 (name "crossmap")
61d5fd03 1804 (version "0.2.1")
191c7101
RW		 1805 (source (origin
1806 (method url-fetch)
1807 (uri (string-append "mirror://sourceforge/crossmap/CrossMap-"
1808 version ".tar.gz"))
1809 (sha256
1810 (base32
61d5fd03
RW		 1811 "07y179f63d7qnzdvkqcziwk9bs3k4zhp81q392fp1hwszjdvy22f"))
1812 ;; This patch has been sent upstream already and is available
1813 ;; for download from Sourceforge, but it has not been merged.
fc1adab1 1814 (patches (search-patches "crossmap-allow-system-pysam.patch"))
191c7101
RW		 1815 (modules '((guix build utils)))
1816 ;; remove bundled copy of pysam
1817 (snippet
1818 '(delete-file-recursively "lib/pysam"))))
1819 (build-system python-build-system)
1820 (arguments
1821 `(#:python ,python-2
1822 #:phases
dc1d3cde
KK		 1823 (modify-phases %standard-phases
1824 (add-after 'unpack 'set-env
1825 (lambda _ (setenv "CROSSMAP_USE_SYSTEM_PYSAM" "1") #t)))))
191c7101
RW		 1826 (inputs
1827 `(("python-numpy" ,python2-numpy)
1828 ("python-pysam" ,python2-pysam)
1829 ("zlib" ,zlib)))
1830 (native-inputs
1831 `(("python-cython" ,python2-cython)
f3b98f4f 1832 ("python-nose" ,python2-nose)))
191c7101
RW		 1833 (home-page "http://crossmap.sourceforge.net/")
1834 (synopsis "Convert genome coordinates between assemblies")
1835 (description
1836 "CrossMap is a program for conversion of genome coordinates or annotation
1837files between different genome assemblies. It supports most commonly used
1838file formats including SAM/BAM, Wiggle/BigWig, BED, GFF/GTF, VCF.")
1839 (license license:gpl2+)))
1840
8e913213
RW		 1841(define-public cutadapt
1842 (package
1843 (name "cutadapt")
0af4df7e 1844 (version "1.14")
8e913213
RW		 1845 (source (origin
1846 (method url-fetch)
1847 (uri (string-append
1848 "https://github.com/marcelm/cutadapt/archive/v"
1849 version ".tar.gz"))
1850 (file-name (string-append name "-" version ".tar.gz"))
1851 (sha256
1852 (base32
0af4df7e 1853 "16gbpiwy4m48vq2h5wqar3i8vr6vcj9gcl2qvqim19x6ya9dp8kd"))))
8e913213
RW		 1854 (build-system python-build-system)
1855 (arguments
33d5b246
TGR		 1856 `(#:phases
1857 (modify-phases %standard-phases
1858 ;; The tests must be run after installation.
1859 (delete 'check)
1860 (add-after 'install 'check
1861 (lambda* (#:key inputs outputs #:allow-other-keys)
1862 (setenv "PYTHONPATH"
1863 (string-append
1864 (getenv "PYTHONPATH")
1865 ":" (assoc-ref outputs "out")
1866 "/lib/python"
1867 (string-take (string-take-right
1868 (assoc-ref inputs "python") 5) 3)
1869 "/site-packages"))
1870 (zero? (system* "nosetests" "-P" "tests")))))))
1f94bff2
TGR		 1871 (inputs
1872 `(("python-xopen" ,python-xopen)))
8e913213
RW		 1873 (native-inputs
1874 `(("python-cython" ,python-cython)
f3b98f4f 1875 ("python-nose" ,python-nose)))
0c6c9c00 1876 (home-page "https://cutadapt.readthedocs.io/en/stable/")
8e913213
RW		 1877 (synopsis "Remove adapter sequences from nucleotide sequencing reads")
1878 (description
1879 "Cutadapt finds and removes adapter sequences, primers, poly-A tails and
1880other types of unwanted sequence from high-throughput sequencing reads.")
1881 (license license:expat)))
1882
1baee943
RW		 1883(define-public libbigwig
1884 (package
1885 (name "libbigwig")
1886 (version "0.1.4")
1887 (source (origin
1888 (method url-fetch)
1889 (uri (string-append "https://github.com/dpryan79/libBigWig/"
1890 "archive/" version ".tar.gz"))
1891 (file-name (string-append name "-" version ".tar.gz"))
1892 (sha256
1893 (base32
1894 "098rjh35pi4a9q83n8wiwvyzykjqj6l8q189p1xgfw4ghywdlvw1"))))
1895 (build-system gnu-build-system)
1896 (arguments
1897 `(#:test-target "test"
1898 #:make-flags
1899 (list "CC=gcc"
1900 (string-append "prefix=" (assoc-ref %outputs "out")))
1901 #:phases
1902 (modify-phases %standard-phases
1903 (delete 'configure)
1904 (add-before 'check 'disable-curl-test
1905 (lambda _
1906 (substitute* "Makefile"
1907 (("./test/testRemote.*") ""))
1908 #t))
1909 ;; This has been fixed with the upstream commit 4ff6959cd8a0, but
1910 ;; there has not yet been a release containing this change.
1911 (add-before 'install 'create-target-dirs
1912 (lambda* (#:key outputs #:allow-other-keys)
1913 (let ((out (assoc-ref outputs "out")))
1914 (mkdir-p (string-append out "/lib"))
1915 (mkdir-p (string-append out "/include"))
1916 #t))))))
1917 (inputs
1918 `(("zlib" ,zlib)
1919 ("curl" ,curl)))
1920 (native-inputs
1921 `(("doxygen" ,doxygen)))
1922 (home-page "https://github.com/dpryan79/libBigWig")
1923 (synopsis "C library for handling bigWig files")
1924 (description
1925 "This package provides a C library for parsing local and remote BigWig
1926files.")
1927 (license license:expat)))
1928
69e0e03c
RW		 1929(define-public python-pybigwig
1930 (package
1931 (name "python-pybigwig")
1932 (version "0.2.5")
1933 (source (origin
1934 (method url-fetch)
1935 (uri (pypi-uri "pyBigWig" version))
1936 (sha256
1937 (base32
1938 "0yrpdxg3y0sny25x4w22lv1k47jzccqjmg7j4bp0hywklvp0hg7d"))
1939 (modules '((guix build utils)))
1940 (snippet
1941 '(begin
1942 ;; Delete bundled libBigWig sources
1943 (delete-file-recursively "libBigWig")))))
1944 (build-system python-build-system)
1945 (arguments
1946 `(#:phases
1947 (modify-phases %standard-phases
1948 (add-after 'unpack 'link-with-libBigWig
1949 (lambda* (#:key inputs #:allow-other-keys)
1950 (substitute* "setup.py"
1951 (("libs=\\[") "libs=[\"BigWig\", "))
1952 #t)))))
1953 (inputs
1954 `(("libbigwig" ,libbigwig)
1955 ("zlib" ,zlib)
1956 ("curl" ,curl)))
1957 (home-page "https://github.com/dpryan79/pyBigWig")
1958 (synopsis "Access bigWig files in Python using libBigWig")
1959 (description
1960 "This package provides Python bindings to the libBigWig library for
1961accessing bigWig files.")
1962 (license license:expat)))
1963
1964(define-public python2-pybigwig
5c31f4aa 1965 (package-with-python2 python-pybigwig))
69e0e03c 1966
ec2a67de
BW		 1967(define-public python-dendropy
1968 (package
1969 (name "python-dendropy")
25d84d31 1970 (version "4.2.0")
ec2a67de
BW		 1971 (source
1972 (origin
1973 (method url-fetch)
1974 (uri (pypi-uri "DendroPy" version))
1975 (sha256
1976 (base32
1885bb0c
RW		 1977 "15c7s3d5gf19ljsxvq5advaa752wfi7pwrdjyhzmg85hccyvp47p"))
1978 (patches (search-patches "python-dendropy-fix-tests.patch"))))
ec2a67de
BW		 1979 (build-system python-build-system)
1980 (home-page "http://packages.python.org/DendroPy/")
1981 (synopsis "Library for phylogenetics and phylogenetic computing")
1982 (description
1983 "DendroPy is a library for phylogenetics and phylogenetic computing: reading,
1984writing, simulation, processing and manipulation of phylogenetic
1985trees (phylogenies) and characters.")
1986 (license license:bsd-3)
1987 (properties `((python2-variant . ,(delay python2-dendropy))))))
1988
1989(define-public python2-dendropy
1990 (let ((base (package-with-python2 (strip-python2-variant python-dendropy))))
1991 (package
1992 (inherit base)
9602e3cc
BW		 1993 (arguments
1994 `(#:python ,python-2
1995 #:phases
1996 (modify-phases %standard-phases
1997 (replace 'check
1998 ;; There is currently a test failure that only happens on some
1999 ;; systems, and only using "setup.py test"
2000 (lambda _ (zero? (system* "nosetests")))))))
f3b98f4f 2001 (native-inputs `(("python2-nose" ,python2-nose)
ec2a67de
BW		 2002 ,@(package-native-inputs base))))))
2003
eb2200f3
RW		 2004(define-public python-py2bit
2005 (package
2006 (name "python-py2bit")
2007 (version "0.2.1")
2008 (source
2009 (origin
2010 (method url-fetch)
2011 (uri (pypi-uri "py2bit" version))
2012 (sha256
2013 (base32
2014 "1cdf4qlmgwsh1f4k0wdv2sr8x9qn4366p0k3614vbd0fpqiarxrl"))))
2015 (build-system python-build-system)
2016 (home-page "https://github.com/dpryan79/py2bit")
2017 (synopsis "Access 2bit files using lib2bit")
2018 (description
2019 "This package provides Python bindings for lib2bit to access 2bit files
2020with Python.")
2021 (license license:expat)))
ec2a67de 2022
1921b1de
RW		 2023(define-public deeptools
2024 (package
2025 (name "deeptools")
fed72008 2026 (version "2.5.1")
1921b1de
RW		 2027 (source (origin
2028 (method url-fetch)
3acb8c85
RW		 2029 (uri (string-append "https://github.com/fidelram/deepTools/"
2030 "archive/" version ".tar.gz"))
1921b1de
RW		 2031 (file-name (string-append name "-" version ".tar.gz"))
2032 (sha256
2033 (base32
fed72008 2034 "1q8i12l2gvk4n2s8lhyzwhh9g4qbc8lrk5l7maz00yvd5g6z5540"))))
1921b1de 2035 (build-system python-build-system)
14bda1ff 2036 (inputs
fed72008
RW		 2037 `(("python-scipy" ,python-scipy)
2038 ("python-numpy" ,python-numpy)
2039 ("python-numpydoc" ,python-numpydoc)
2040 ("python-matplotlib" ,python-matplotlib)
2041 ("python-pysam" ,python-pysam)
2042 ("python-py2bit" ,python-py2bit)
2043 ("python-pybigwig" ,python-pybigwig)))
1921b1de 2044 (native-inputs
fed72008
RW		 2045 `(("python-mock" ,python-mock) ;for tests
2046 ("python-nose" ,python-nose) ;for tests
2047 ("python-pytz" ,python-pytz))) ;for tests
1921b1de
RW		 2048 (home-page "https://github.com/fidelram/deepTools")
2049 (synopsis "Tools for normalizing and visualizing deep-sequencing data")
2050 (description
2051 "DeepTools addresses the challenge of handling the large amounts of data
2052that are now routinely generated from DNA sequencing centers. To do so,
2053deepTools contains useful modules to process the mapped reads data to create
2054coverage files in standard bedGraph and bigWig file formats. By doing so,
2055deepTools allows the creation of normalized coverage files or the comparison
2056between two files (for example, treatment and control). Finally, using such
2057normalized and standardized files, multiple visualizations can be created to
2058identify enrichments with functional annotations of the genome.")
2059 (license license:gpl3+)))
2060
684bf7c7
BW		 2061(define-public diamond
2062 (package
2063 (name "diamond")
171ba248 2064 (version "0.9.12")
684bf7c7
BW		 2065 (source (origin
2066 (method url-fetch)
2067 (uri (string-append
2068 "https://github.com/bbuchfink/diamond/archive/v"
2069 version ".tar.gz"))
2070 (file-name (string-append name "-" version ".tar.gz"))
2071 (sha256
2072 (base32
171ba248 2073 "1zrnr59pqdg56wxfdx3xg4h4d8fa14cxq5kfmc3q88hxrj42rqgs"))))
122395f9 2074 (build-system cmake-build-system)
684bf7c7 2075 (arguments
7c544991
BW		 2076 '(#:tests? #f ; no "check" target
2077 #:phases
2078 (modify-phases %standard-phases
2079 (add-after 'unpack 'remove-native-compilation
2080 (lambda _
2081 (substitute* "CMakeLists.txt" (("-march=native") ""))
2082 #t)))))
684bf7c7 2083 (inputs
122395f9 2084 `(("zlib" ,zlib)))
684bf7c7
BW		 2085 (home-page "https://github.com/bbuchfink/diamond")
2086 (synopsis "Accelerated BLAST compatible local sequence aligner")
2087 (description
2088 "DIAMOND is a BLAST-compatible local aligner for mapping protein and
2089translated DNA query sequences against a protein reference database (BLASTP
2090and BLASTX alignment mode). The speedup over BLAST is up to 20,000 on short
2091reads at a typical sensitivity of 90-99% relative to BLAST depending on the
2092data and settings.")
ef81341f 2093 (license license:agpl3+)))
684bf7c7 2094
97b9da68
RW		 2095(define-public discrover
2096 (package
2097 (name "discrover")
2098 (version "1.6.0")
2099 (source
2100 (origin
2101 (method url-fetch)
2102 (uri (string-append "https://github.com/maaskola/discrover/archive/"
2103 version ".tar.gz"))
2104 (file-name (string-append name "-" version ".tar.gz"))
2105 (sha256
2106 (base32
2107 "0rah9ja4m0rl5mldd6vag9rwrivw1zrqxssfq8qx64m7961fp68k"))))
2108 (build-system cmake-build-system)
fa702e1a
RW		 2109 (arguments
2110 `(#:tests? #f ; there are no tests
2111 #:phases
2112 (modify-phases %standard-phases
2113 (add-after 'unpack 'add-missing-includes
2114 (lambda _
2115 (substitute* "src/executioninformation.hpp"
2116 (("#define EXECUTIONINFORMATION_HPP" line)
2117 (string-append line "\n#include <random>")))
2118 (substitute* "src/plasma/fasta.hpp"
2119 (("#define FASTA_HPP" line)
2120 (string-append line "\n#include <random>")))
2121 #t)))))
97b9da68
RW		 2122 (inputs
2123 `(("boost" ,boost)
2124 ("cairo" ,cairo)))
2125 (native-inputs
2126 `(("texlive" ,texlive)
2127 ("imagemagick" ,imagemagick)))
2128 (home-page "http://dorina.mdc-berlin.de/public/rajewsky/discrover/")
2129 (synopsis "Discover discriminative nucleotide sequence motifs")
2130 (description "Discrover is a motif discovery method to find binding sites
2131of nucleic acid binding proteins.")
2132 (license license:gpl3+)))
2133
6619f9c7
RW		 2134(define-public eigensoft
2135 (let ((revision "1")
2136 (commit "b14d1e202e21e532536ff8004f0419cd5e259dc7"))
2137 (package
2138 (name "eigensoft")
2139 (version (string-append "6.1.2-"
2140 revision "."
2141 (string-take commit 9)))
2142 (source
2143 (origin
2144 (method git-fetch)
2145 (uri (git-reference
2146 (url "https://github.com/DReichLab/EIG.git")
2147 (commit commit)))
2148 (file-name (string-append "eigensoft-" commit "-checkout"))
2149 (sha256
2150 (base32
2151 "0f5m6k2j5c16xc3xbywcs989xyc26ncy1zfzp9j9n55n9r4xcaiq"))
2152 (modules '((guix build utils)))
2153 ;; Remove pre-built binaries.
2154 (snippet '(begin
2155 (delete-file-recursively "bin")
2156 (mkdir "bin")
2157 #t))))
2158 (build-system gnu-build-system)
2159 (arguments
2160 `(#:tests? #f ; There are no tests.
2161 #:make-flags '("CC=gcc")
2162 #:phases
2163 (modify-phases %standard-phases
2164 ;; There is no configure phase, but the Makefile is in a
2165 ;; sub-directory.
2166 (replace 'configure
2167 (lambda _
2168 (chdir "src")
2169 ;; The link flags are incomplete.
2170 (substitute* "Makefile"
2171 (("-lgsl") "-lgsl -lm -llapack -llapacke -lpthread"))
2172 #t))
2173 ;; The provided install target only copies executables to
2174 ;; the "bin" directory in the build root.
2175 (add-after 'install 'actually-install
2176 (lambda* (#:key outputs #:allow-other-keys)
2177 (let* ((out (assoc-ref outputs "out"))
2178 (bin (string-append out "/bin")))
6619f9c7
RW		 2179 (for-each (lambda (file)
2180 (install-file file bin))
2181 (find-files "../bin" ".*"))
2182 #t))))))
2183 (inputs
2184 `(("gsl" ,gsl)
2185 ("lapack" ,lapack)
6619f9c7
RW		 2186 ("openblas" ,openblas)
2187 ("perl" ,perl)
2188 ("gfortran" ,gfortran "lib")))
2189 (home-page "https://github.com/DReichLab/EIG")
2190 (synopsis "Tools for population genetics")
2191 (description "The EIGENSOFT package provides tools for population
2192genetics and stratification correction. EIGENSOFT implements methods commonly
2193used in population genetics analyses such as PCA, computation of Tracy-Widom
2194statistics, and finding related individuals in structured populations. It
2195comes with a built-in plotting script and supports multiple file formats and
2196quantitative phenotypes.")
2197 ;; The license of the eigensoft tools is Expat, but since it's
2198 ;; linking with the GNU Scientific Library (GSL) the effective
2199 ;; license is the GPL.
2200 (license license:gpl3+))))
2201
365c8153
RW		 2202(define-public edirect
2203 (package
2204 (name "edirect")
83b84fa8 2205 (version "4.10")
365c8153
RW		 2206 (source (origin
2207 (method url-fetch)
83b84fa8
RW		 2208 (uri (string-append "ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/"
2209 "versions/2016-05-03/edirect.tar.gz"))
365c8153
RW		 2210 (sha256
2211 (base32
83b84fa8 2212 "15zsprak5yh8c1yrz4r1knmb5s8qcmdid4xdhkh3lqcv64l60hli"))))
365c8153
RW		 2213 (build-system perl-build-system)
2214 (arguments
2215 `(#:tests? #f ;no "check" target
2216 #:phases
2217 (modify-phases %standard-phases
2218 (delete 'configure)
2219 (delete 'build)
2220 (replace 'install
2221 (lambda* (#:key outputs #:allow-other-keys)
2222 (let ((target (string-append (assoc-ref outputs "out")
2223 "/bin")))
2224 (mkdir-p target)
f3860753 2225 (install-file "edirect.pl" target)
365c8153
RW		 2226 #t)))
2227 (add-after
2228 'install 'wrap-program
2229 (lambda* (#:key inputs outputs #:allow-other-keys)
2230 ;; Make sure 'edirect.pl' finds all perl inputs at runtime.
2231 (let* ((out (assoc-ref outputs "out"))
2232 (path (getenv "PERL5LIB")))
2233 (wrap-program (string-append out "/bin/edirect.pl")
2234 `("PERL5LIB" ":" prefix (,path)))))))))
2235 (inputs
2236 `(("perl-html-parser" ,perl-html-parser)
2237 ("perl-encode-locale" ,perl-encode-locale)
2238 ("perl-file-listing" ,perl-file-listing)
2239 ("perl-html-tagset" ,perl-html-tagset)
2240 ("perl-html-tree" ,perl-html-tree)
2241 ("perl-http-cookies" ,perl-http-cookies)
2242 ("perl-http-date" ,perl-http-date)
2243 ("perl-http-message" ,perl-http-message)
2244 ("perl-http-negotiate" ,perl-http-negotiate)
2245 ("perl-lwp-mediatypes" ,perl-lwp-mediatypes)
2246 ("perl-lwp-protocol-https" ,perl-lwp-protocol-https)
2247 ("perl-net-http" ,perl-net-http)
2248 ("perl-uri" ,perl-uri)
2249 ("perl-www-robotrules" ,perl-www-robotrules)
2250 ("perl" ,perl)))
3d51ec91 2251 (home-page "http://www.ncbi.nlm.nih.gov/books/NBK179288/")
365c8153
RW		 2252 (synopsis "Tools for accessing the NCBI's set of databases")
2253 (description
2254 "Entrez Direct (EDirect) is a method for accessing the National Center
2255for Biotechnology Information's (NCBI) set of interconnected
2256databases (publication, sequence, structure, gene, variation, expression,
2257etc.) from a terminal. Functions take search terms from command-line
2258arguments. Individual operations are combined to build multi-step queries.
2259Record retrieval and formatting normally complete the process.
2260
2261EDirect also provides an argument-driven function that simplifies the
2262extraction of data from document summaries or other results that are returned
2263in structured XML format. This can eliminate the need for writing custom
2264software to answer ad hoc questions.")
2265 (license license:public-domain)))
2266
b16728b0
BW		 2267(define-public exonerate
2268 (package
2269 (name "exonerate")
2270 (version "2.4.0")
2271 (source
2272 (origin
2273 (method url-fetch)
2274 (uri
2275 (string-append
2276 "http://ftp.ebi.ac.uk/pub/software/vertebrategenomics/exonerate/"
2277 "exonerate-" version ".tar.gz"))
2278 (sha256
2279 (base32
2280 "0hj0m9xygiqsdxvbg79wq579kbrx1mdrabi2bzqz2zn9qwfjcjgq"))))
2281 (build-system gnu-build-system)
2282 (arguments
2283 `(#:parallel-build? #f)) ; Building in parallel fails on some machines.
2284 (native-inputs
2285 `(("pkg-config" ,pkg-config)))
2286 (inputs
2287 `(("glib" ,glib)))
2288 (home-page
2289 "https://www.ebi.ac.uk/about/vertebrate-genomics/software/exonerate")
2290 (synopsis "Generic tool for biological sequence alignment")
2291 (description
2292 "Exonerate is a generic tool for pairwise sequence comparison. It allows
2293the alignment of sequences using a many alignment models, either exhaustive
2294dynamic programming or a variety of heuristics.")
2295 (license license:gpl3)))
2296
e4e5a4d8
RW		 2297(define-public express
2298 (package
2299 (name "express")
2300 (version "1.5.1")
2301 (source (origin
2302 (method url-fetch)
2303 (uri
2304 (string-append
2305 "http://bio.math.berkeley.edu/eXpress/downloads/express-"
2306 version "/express-" version "-src.tgz"))
2307 (sha256
2308 (base32
2309 "03rczxd0gjp2l1jxcmjfmf5j94j77zqyxa6x063zsc585nj40n0c"))))
2310 (build-system cmake-build-system)
2311 (arguments
2312 `(#:tests? #f ;no "check" target
2313 #:phases
dc1d3cde
KK		 2314 (modify-phases %standard-phases
2315 (add-after 'unpack 'use-shared-boost-libs-and-set-bamtools-paths
2316 (lambda* (#:key inputs #:allow-other-keys)
2317 (substitute* "CMakeLists.txt"
2318 (("set\\(Boost_USE_STATIC_LIBS ON\\)")
2319 "set(Boost_USE_STATIC_LIBS OFF)")
2320 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/bamtools/include")
2321 (string-append (assoc-ref inputs "bamtools") "/include/bamtools")))
2322 (substitute* "src/CMakeLists.txt"
2323 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/\\.\\./bamtools/lib")
2324 (string-append (assoc-ref inputs "bamtools") "/lib/bamtools")))
2325 #t)))))
e4e5a4d8
RW		 2326 (inputs
2327 `(("boost" ,boost)
2328 ("bamtools" ,bamtools)
2329 ("protobuf" ,protobuf)
2330 ("zlib" ,zlib)))
2331 (home-page "http://bio.math.berkeley.edu/eXpress")
2332 (synopsis "Streaming quantification for high-throughput genomic sequencing")
2333 (description
2334 "eXpress is a streaming tool for quantifying the abundances of a set of
2335target sequences from sampled subsequences. Example applications include
2336transcript-level RNA-Seq quantification, allele-specific/haplotype expression
2337analysis (from RNA-Seq), transcription factor binding quantification in
2338ChIP-Seq, and analysis of metagenomic data.")
2339 (license license:artistic2.0)))
2340
f3674b1c
BW		 2341(define-public express-beta-diversity
2342 (package
2343 (name "express-beta-diversity")
2344 (version "1.0.7")
2345 (source (origin
2346 (method url-fetch)
2347 (uri
2348 (string-append
2349 "https://github.com/dparks1134/ExpressBetaDiversity/archive/v"
2350 version ".tar.gz"))
2351 (file-name (string-append name "-" version ".tar.gz"))
2352 (sha256
2353 (base32
2354 "1djvdlmqvjf6h0zq7w36y8cl5cli6rgj86x65znl48agnwmzxfxr"))))
2355 (build-system gnu-build-system)
2356 (arguments
2357 `(#:phases
2358 (modify-phases %standard-phases
2359 (delete 'configure)
2360 (add-before 'build 'enter-source (lambda _ (chdir "source") #t))
2361 (replace 'check
2362 (lambda _ (zero? (system* "../bin/ExpressBetaDiversity"
2363 "-u"))))
2364 (add-after 'check 'exit-source (lambda _ (chdir "..") #t))
2365 (replace 'install
2366 (lambda* (#:key outputs #:allow-other-keys)
2367 (let ((bin (string-append (assoc-ref outputs "out")
2368 "/bin")))
2369 (mkdir-p bin)
f3860753
TGR		 2370 (install-file "scripts/convertToEBD.py" bin)
2371 (install-file "bin/ExpressBetaDiversity" bin)
f3674b1c
BW		 2372 #t))))))
2373 (inputs
2374 `(("python" ,python-2)))
2375 (home-page "http://kiwi.cs.dal.ca/Software/ExpressBetaDiversity")
2376 (synopsis "Taxon- and phylogenetic-based beta diversity measures")
2377 (description
2378 "Express Beta Diversity (EBD) calculates ecological beta diversity
2379(dissimilarity) measures between biological communities. EBD implements a
2380variety of diversity measures including those that make use of phylogenetic
2381similarity of community members.")
2382 (license license:gpl3+)))
2383
12b04cbe
BW		 2384(define-public fasttree
2385 (package
2386 (name "fasttree")
88682c9a 2387 (version "2.1.10")
12b04cbe
BW		 2388 (source (origin
2389 (method url-fetch)
2390 (uri (string-append
2391 "http://www.microbesonline.org/fasttree/FastTree-"
2392 version ".c"))
2393 (sha256
2394 (base32
88682c9a 2395 "0vcjdvy1j4m702vmak4svbfkrpcw63k7wymfksjp9a982zy8kjsl"))))
12b04cbe
BW		 2396 (build-system gnu-build-system)
2397 (arguments
2398 `(#:tests? #f ; no "check" target
2399 #:phases
2400 (modify-phases %standard-phases
2401 (delete 'unpack)
2402 (delete 'configure)
2403 (replace 'build
e03a5153
BW		 2404 (lambda* (#:key source #:allow-other-keys)
2405 (and (zero? (system* "gcc"
2406 "-O3"
2407 "-finline-functions"
2408 "-funroll-loops"
2409 "-Wall"
2410 "-o"
2411 "FastTree"
2412 source
2413 "-lm"))
2414 (zero? (system* "gcc"
2415 "-DOPENMP"
2416 "-fopenmp"
2417 "-O3"
2418 "-finline-functions"
2419 "-funroll-loops"
2420 "-Wall"
2421 "-o"
2422 "FastTreeMP"
2423 source
2424 "-lm")))))
12b04cbe 2425 (replace 'install
e03a5153
BW		 2426 (lambda* (#:key outputs #:allow-other-keys)
2427 (let ((bin (string-append (assoc-ref outputs "out")
2428 "/bin")))
2429 (mkdir-p bin)
f3860753
TGR		 2430 (install-file "FastTree" bin)
2431 (install-file "FastTreeMP" bin)
e03a5153 2432 #t))))))
12b04cbe
BW		 2433 (home-page "http://www.microbesonline.org/fasttree")
2434 (synopsis "Infers approximately-maximum-likelihood phylogenetic trees")
2435 (description
2436 "FastTree can handle alignments with up to a million of sequences in a
2437reasonable amount of time and memory. For large alignments, FastTree is
2438100-1,000 times faster than PhyML 3.0 or RAxML 7.")
2439 (license license:gpl2+)))
2440
2127cedb
RW		 2441(define-public fastx-toolkit
2442 (package
2443 (name "fastx-toolkit")
2444 (version "0.0.14")
2445 (source (origin
2446 (method url-fetch)
2447 (uri
2448 (string-append
2449 "https://github.com/agordon/fastx_toolkit/releases/download/"
2450 version "/fastx_toolkit-" version ".tar.bz2"))
2451 (sha256
2452 (base32
2453 "01jqzw386873sr0pjp1wr4rn8fsga2vxs1qfmicvx1pjr72007wy"))))
2454 (build-system gnu-build-system)
2455 (inputs
2456 `(("libgtextutils" ,libgtextutils)))
2457 (native-inputs
2458 `(("pkg-config" ,pkg-config)))
2459 (home-page "http://hannonlab.cshl.edu/fastx_toolkit/")
2460 (synopsis "Tools for FASTA/FASTQ file preprocessing")
2461 (description
2462 "The FASTX-Toolkit is a collection of command line tools for Short-Reads
2463FASTA/FASTQ files preprocessing.
2464
2465Next-Generation sequencing machines usually produce FASTA or FASTQ files,
2466containing multiple short-reads sequences. The main processing of such
2467FASTA/FASTQ files is mapping the sequences to reference genomes. However, it
2468is sometimes more productive to preprocess the files before mapping the
2469sequences to the genome---manipulating the sequences to produce better mapping
2470results. The FASTX-Toolkit tools perform some of these preprocessing tasks.")
2471 (license license:agpl3+)))
2472
d7678942
RW		 2473(define-public flexbar
2474 (package
2475 (name "flexbar")
2476 (version "2.5")
2477 (source (origin
2478 (method url-fetch)
2479 (uri
2480 (string-append "mirror://sourceforge/flexbar/"
2481 version "/flexbar_v" version "_src.tgz"))
2482 (sha256
2483 (base32
2484 "13jaykc3y1x8y5nn9j8ljnb79s5y51kyxz46hdmvvjj6qhyympmf"))))
2485 (build-system cmake-build-system)
2486 (arguments
4ca009c0 2487 `(#:configure-flags (list
d7678942
RW		 2488 (string-append "-DFLEXBAR_BINARY_DIR="
2489 (assoc-ref %outputs "out")
2490 "/bin/"))
2491 #:phases
dc1d3cde
KK		 2492 (modify-phases %standard-phases
2493 (replace 'check
2494 (lambda* (#:key outputs #:allow-other-keys)
2495 (setenv "PATH" (string-append
2496 (assoc-ref outputs "out") "/bin:"
2497 (getenv "PATH")))
2498 (chdir "../flexbar_v2.5_src/test")
2499 (zero? (system* "bash" "flexbar_validate.sh"))))
2500 (delete 'install))))
d7678942
RW		 2501 (inputs
2502 `(("tbb" ,tbb)
2503 ("zlib" ,zlib)))
2504 (native-inputs
2505 `(("pkg-config" ,pkg-config)
2506 ("seqan" ,seqan)))
2507 (home-page "http://flexbar.sourceforge.net")
2508 (synopsis "Barcode and adapter removal tool for sequencing platforms")
2509 (description
2510 "Flexbar preprocesses high-throughput nucleotide sequencing data
2511efficiently. It demultiplexes barcoded runs and removes adapter sequences.
2512Moreover, trimming and filtering features are provided. Flexbar increases
2513read mapping rates and improves genome and transcriptome assemblies. It
2514supports next-generation sequencing data in fasta/q and csfasta/q format from
2515Illumina, Roche 454, and the SOLiD platform.")
2516 (license license:gpl3)))
2517
19f4554c
BW		 2518(define-public fraggenescan
2519 (package
2520 (name "fraggenescan")
74297231 2521 (version "1.30")
19f4554c
BW		 2522 (source
2523 (origin
2524 (method url-fetch)
2525 (uri
2526 (string-append "mirror://sourceforge/fraggenescan/"
2527 "FragGeneScan" version ".tar.gz"))
2528 (sha256
74297231 2529 (base32 "158dcnwczgcyhwm4qlx19sanrwgdpzf6bn2y57mbpx55lkgz1mzj"))))
19f4554c
BW		 2530 (build-system gnu-build-system)
2531 (arguments
2532 `(#:phases
2533 (modify-phases %standard-phases
2534 (delete 'configure)
2535 (add-before 'build 'patch-paths
2536 (lambda* (#:key outputs #:allow-other-keys)
2537 (let* ((out (string-append (assoc-ref outputs "out")))
2538 (share (string-append out "/share/fraggenescan/")))
2539 (substitute* "run_FragGeneScan.pl"
2540 (("system\\(\"rm")
2541 (string-append "system(\"" (which "rm")))
2542 (("system\\(\"mv")
2543 (string-append "system(\"" (which "mv")))
74297231 2544 (("\\\"awk") (string-append "\"" (which "awk")))
19f4554c
BW		 2545 ;; This script and other programs expect the training files
2546 ;; to be in the non-standard location bin/train/XXX. Change
2547 ;; this to be share/fraggenescan/train/XXX instead.
2548 (("^\\$train.file = \\$dir.*")
2549 (string-append "$train_file = \""
2550 share
2551 "train/\".$FGS_train_file;")))
2552 (substitute* "run_hmm.c"
2553 (("^ strcat\\(train_dir, \\\"train/\\\"\\);")
74297231 2554 (string-append " strcpy(train_dir, \"" share "/train/\");"))))
19f4554c
BW		 2555 #t))
2556 (replace 'build
2557 (lambda _ (and (zero? (system* "make" "clean"))
2558 (zero? (system* "make" "fgs")))))
2559 (replace 'install
2560 (lambda* (#:key outputs #:allow-other-keys)
2561 (let* ((out (string-append (assoc-ref outputs "out")))
2562 (bin (string-append out "/bin/"))
2563 (share (string-append out "/share/fraggenescan/train")))
2564 (install-file "run_FragGeneScan.pl" bin)
2565 (install-file "FragGeneScan" bin)
19f4554c
BW		 2566 (copy-recursively "train" share))))
2567 (delete 'check)
2568 (add-after 'install 'post-install-check
2569 ;; In lieu of 'make check', run one of the examples and check the
2570 ;; output files gets created.
2571 (lambda* (#:key outputs #:allow-other-keys)
2572 (let* ((out (string-append (assoc-ref outputs "out")))
74297231
BW		 2573 (bin (string-append out "/bin/"))
2574 (frag (string-append bin "run_FragGeneScan.pl")))
2575 (and (zero? (system* frag ; Test complete genome.
19f4554c
BW		 2576 "-genome=./example/NC_000913.fna"
2577 "-out=./test2"
2578 "-complete=1"
2579 "-train=complete"))
2580 (file-exists? "test2.faa")
2581 (file-exists? "test2.ffn")
2582 (file-exists? "test2.gff")
74297231
BW		 2583 (file-exists? "test2.out")
2584 (zero? (system* ; Test incomplete sequences.
2585 frag
2586 "-genome=./example/NC_000913-fgs.ffn"
2587 "-out=out"
2588 "-complete=0"
2589 "-train=454_30")))))))))
19f4554c
BW		 2590 (inputs
2591 `(("perl" ,perl)
2592 ("python" ,python-2))) ;not compatible with python 3.
2593 (home-page "https://sourceforge.net/projects/fraggenescan/")
2594 (synopsis "Finds potentially fragmented genes in short reads")
2595 (description
2596 "FragGeneScan is a program for predicting bacterial and archaeal genes in
2597short and error-prone DNA sequencing reads. It can also be applied to predict
2598genes in incomplete assemblies or complete genomes.")
2599 ;; GPL3+ according to private correspondense with the authors.
2600 (license license:gpl3+)))
2601
81f3e0c1
BW		 2602(define-public fxtract
2603 (let ((util-commit "776ca85a18a47492af3794745efcb4a905113115"))
2604 (package
2605 (name "fxtract")
2606 (version "2.3")
2607 (source
2608 (origin
2609 (method url-fetch)
2610 (uri (string-append
2611 "https://github.com/ctSkennerton/fxtract/archive/"
2612 version ".tar.gz"))
2613 (file-name (string-append "ctstennerton-util-"
2614 (string-take util-commit 7)
2615 "-checkout"))
2616 (sha256
2617 (base32
2618 "0275cfdhis8517hm01is62062swmi06fxzifq7mr3knbbxjlaiwj"))))
2619 (build-system gnu-build-system)
2620 (arguments
2621 `(#:make-flags (list
2622 (string-append "PREFIX=" (assoc-ref %outputs "out"))
2623 "CC=gcc")
2624 #:test-target "fxtract_test"
2625 #:phases
2626 (modify-phases %standard-phases
2627 (delete 'configure)
2628 (add-before 'build 'copy-util
2629 (lambda* (#:key inputs #:allow-other-keys)
2630 (rmdir "util")
2631 (copy-recursively (assoc-ref inputs "ctskennerton-util") "util")
2632 #t))
2633 ;; Do not use make install as this requires additional dependencies.
2634 (replace 'install
2635 (lambda* (#:key outputs #:allow-other-keys)
2636 (let* ((out (assoc-ref outputs "out"))
2637 (bin (string-append out"/bin")))
2638 (install-file "fxtract" bin)
2639 #t))))))
2640 (inputs
2641 `(("pcre" ,pcre)
2642 ("zlib" ,zlib)))
2643 (native-inputs
2644 ;; ctskennerton-util is licensed under GPL2.
2645 `(("ctskennerton-util"
2646 ,(origin
2647 (method git-fetch)
2648 (uri (git-reference
2649 (url "https://github.com/ctSkennerton/util.git")
2650 (commit util-commit)))
2651 (file-name (string-append
2652 "ctstennerton-util-" util-commit "-checkout"))
2653 (sha256
2654 (base32
2655 "0cls1hd4vgj3f36fpzzg4xc77d6f3hpc60cbpfmn2gdr7ykzzad7"))))))
2656 (home-page "https://github.com/ctSkennerton/fxtract")
2657 (synopsis "Extract sequences from FASTA and FASTQ files")
2658 (description
2659 "Fxtract extracts sequences from a protein or nucleotide fastx (FASTA
2660or FASTQ) file given a subsequence. It uses a simple substring search for
2661basic tasks but can change to using POSIX regular expressions, PCRE, hash
2662lookups or multi-pattern searching as required. By default fxtract looks in
2663the sequence of each record but can also be told to look in the header,
2664comment or quality sections.")
afde1a26
BW		 2665 ;; 'util' requires SSE instructions.
2666 (supported-systems '("x86_64-linux"))
81f3e0c1
BW		 2667 (license license:expat))))
2668
2b18ad05
PP		 2669(define-public gemma
2670 (package
2671 (name "gemma")
2672 (version "0.96")
2673 (source (origin
2674 (method url-fetch)
2675 (uri (string-append "https://github.com/xiangzhou/GEMMA/archive/v"
2676 version ".tar.gz"))
2677 (file-name (string-append name "-" version ".tar.gz"))
2678 (sha256
2679 (base32
ce7e361f
EF		 2680 "055ynn16gd12pf78n4vr2a9jlwsbwzajpdnf2y2yilg1krfff222"))
2681 (patches (search-patches "gemma-intel-compat.patch"))))
2b18ad05
PP		 2682 (inputs
2683 `(("gsl" ,gsl)
2684 ("lapack" ,lapack)
2685 ("zlib" ,zlib)))
2686 (build-system gnu-build-system)
2687 (arguments
2c9232ae 2688 `(#:make-flags
ce7e361f
EF		 2689 '(,@(match (%current-system)
2690 ("x86_64-linux"
2691 '("FORCE_DYNAMIC=1"))
2692 ("i686-linux"
2693 '("FORCE_DYNAMIC=1" "FORCE_32BIT=1"))
2694 (_
2695 '("FORCE_DYNAMIC=1" "NO_INTEL_COMPAT=1"))))
2b18ad05
PP		 2696 #:phases
2697 (modify-phases %standard-phases
2698 (delete 'configure)
2699 (add-before 'build 'bin-mkdir
07bf6929
EF		 2700 (lambda _
2701 (mkdir-p "bin")
2702 #t))
2b18ad05 2703 (replace 'install
07bf6929
EF		 2704 (lambda* (#:key outputs #:allow-other-keys)
2705 (let ((out (assoc-ref outputs "out")))
2706 (install-file "bin/gemma"
2707 (string-append
2708 out "/bin")))
2709 #t)))
2b18ad05
PP		 2710 #:tests? #f)) ; no tests included yet
2711 (home-page "https://github.com/xiangzhou/GEMMA")
2712 (synopsis "Tool for genome-wide efficient mixed model association")
2713 (description
2714 "Genome-wide Efficient Mixed Model Association (GEMMA) provides a
2715standard linear mixed model resolver with application in genome-wide
2716association studies (GWAS).")
2717 (license license:gpl3)))
2718
5854f685
RW		 2719(define-public grit
2720 (package
2721 (name "grit")
2722 (version "2.0.2")
2723 (source (origin
2724 (method url-fetch)
2725 (uri (string-append
2726 "https://github.com/nboley/grit/archive/"
2727 version ".tar.gz"))
2728 (file-name (string-append name "-" version ".tar.gz"))
2729 (sha256
2730 (base32
2731 "157in84dj70wimbind3x7sy1whs3h57qfgcnj2s6lrd38fbrb7mj"))))
2732 (build-system python-build-system)
2733 (arguments
2734 `(#:python ,python-2
2735 #:phases
dc1d3cde
KK		 2736 (modify-phases %standard-phases
2737 (add-after 'unpack 'generate-from-cython-sources
2738 (lambda* (#:key inputs outputs #:allow-other-keys)
2739 ;; Delete these C files to force fresh generation from pyx sources.
2740 (delete-file "grit/sparsify_support_fns.c")
2741 (delete-file "grit/call_peaks_support_fns.c")
2742 (substitute* "setup.py"
2743 (("Cython.Setup") "Cython.Build")
2744 ;; Add numpy include path to fix compilation
2745 (("pyx\", \\]")
2746 (string-append "pyx\", ], include_dirs = ['"
2747 (assoc-ref inputs "python-numpy")
2748 "/lib/python2.7/site-packages/numpy/core/include/"
2749 "']")))
2750 #t)))))
5854f685
RW		 2751 (inputs
2752 `(("python-scipy" ,python2-scipy)
2753 ("python-numpy" ,python2-numpy)
2754 ("python-pysam" ,python2-pysam)
2755 ("python-networkx" ,python2-networkx)))
2756 (native-inputs
f3b98f4f 2757 `(("python-cython" ,python2-cython)))
5854f685
RW		 2758 (home-page "http://grit-bio.org")
2759 (synopsis "Tool for integrative analysis of RNA-seq type assays")
2760 (description
2761 "GRIT is designed to use RNA-seq, TES, and TSS data to build and quantify
2762full length transcript models. When none of these data sources are available,
2763GRIT can be run by providing a candidate set of TES or TSS sites. In
2764addition, GRIT can merge in reference junctions and gene boundaries. GRIT can
2765also be run in quantification mode, where it uses a provided GTF file and just
2766estimates transcript expression.")
2767 (license license:gpl3+)))
2768
346a829a
RW		 2769(define-public hisat
2770 (package
2771 (name "hisat")
2772 (version "0.1.4")
2773 (source (origin
2774 (method url-fetch)
2775 (uri (string-append
2776 "http://ccb.jhu.edu/software/hisat/downloads/hisat-"
2777 version "-beta-source.zip"))
2778 (sha256
2779 (base32
2780 "1k381ydranqxp09yf2y7w1d0chz5d59vb6jchi89hbb0prq19lk5"))))
2781 (build-system gnu-build-system)
2782 (arguments
e58d01fa
RW		 2783 `(#:tests? #f ;no check target
2784 #:make-flags '("allall"
2785 ;; Disable unsupported `popcnt' instructions on
2786 ;; architectures other than x86_64
2787 ,@(if (string-prefix? "x86_64"
2788 (or (%current-target-system)
2789 (%current-system)))
2790 '()
2791 '("POPCNT_CAPABILITY=0")))
346a829a 2792 #:phases
dc1d3cde
KK		 2793 (modify-phases %standard-phases
2794 (add-after 'unpack 'patch-sources
2795 (lambda _
2796 ;; XXX Cannot use snippet because zip files are not supported
2797 (substitute* "Makefile"
2798 (("^CC = .*$") "CC = gcc")
2799 (("^CPP = .*$") "CPP = g++")
2800 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
2801 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
2802 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\""))
2803 (substitute* '("hisat-build" "hisat-inspect")
2804 (("/usr/bin/env") (which "env")))
2805 #t))
2806 (replace 'install
2807 (lambda* (#:key outputs #:allow-other-keys)
2808 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
2809 (for-each (lambda (file)
2810 (install-file file bin))
2811 (find-files
2812 "."
2813 "hisat(-(build|align|inspect)(-(s|l)(-debug)*)*)*$")))
2814 #t))
2815 (delete 'configure))))
346a829a
RW		 2816 (native-inputs
2817 `(("unzip" ,unzip)))
2818 (inputs
2819 `(("perl" ,perl)
2820 ("python" ,python)
2821 ("zlib" ,zlib)))
60af3d82
RW		 2822 ;; Non-portable SSE instructions are used so building fails on platforms
2823 ;; other than x86_64.
2824 (supported-systems '("x86_64-linux"))
346a829a
RW		 2825 (home-page "http://ccb.jhu.edu/software/hisat/index.shtml")
2826 (synopsis "Hierarchical indexing for spliced alignment of transcripts")
2827 (description
2828 "HISAT is a fast and sensitive spliced alignment program for mapping
2829RNA-seq reads. In addition to one global FM index that represents a whole
2830genome, HISAT uses a large set of small FM indexes that collectively cover the
2831whole genome. These small indexes (called local indexes) combined with
2832several alignment strategies enable effective alignment of RNA-seq reads, in
2833particular, reads spanning multiple exons.")
2834 (license license:gpl3+)))
2835
e84efc50
RW		 2836(define-public hisat2
2837 (package
2838 (name "hisat2")
2839 (version "2.0.5")
2840 (source
2841 (origin
2842 (method url-fetch)
2843 ;; FIXME: a better source URL is
2844 ;; (string-append "ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2"
2845 ;; "/downloads/hisat2-" version "-source.zip")
2846 ;; with hash "0lywnr8kijwsc2aw10dwxic0n0yvip6fl3rjlvc8zzwahamy4x7g"
2847 ;; but it is currently unavailable.
2848 (uri "https://github.com/infphilo/hisat2/archive/cba6e8cb.tar.gz")
2849 (file-name (string-append name "-" version ".tar.gz"))
2850 (sha256
2851 (base32
2852 "1mf2hdsyv7cd97xm9mp9a4qws02yrj95y6w6f6cdwnq0klp81r50"))))
2853 (build-system gnu-build-system)
2854 (arguments
2855 `(#:tests? #f ; no check target
2856 #:make-flags (list "CC=gcc" "CXX=g++" "allall")
2857 #:modules ((guix build gnu-build-system)
2858 (guix build utils)
2859 (srfi srfi-26))
2860 #:phases
2861 (modify-phases %standard-phases
2862 (add-after 'unpack 'make-deterministic
2863 (lambda _
2864 (substitute* "Makefile"
2865 (("`date`") "0"))
2866 #t))
2867 (delete 'configure)
2868 (replace 'install
2869 (lambda* (#:key outputs #:allow-other-keys)
2870 (let* ((out (assoc-ref outputs "out"))
2871 (bin (string-append out "/bin/"))
2872 (doc (string-append out "/share/doc/hisat2/")))
2873 (for-each
2874 (cut install-file <> bin)
2875 (find-files "."
2876 "hisat2(-(build|align|inspect)(-(s|l)(-debug)*)*)*$"))
2877 (mkdir-p doc)
2878 (install-file "doc/manual.inc.html" doc))
2879 #t)))))
2880 (native-inputs
2881 `(("unzip" ,unzip) ; needed for archive from ftp
2882 ("perl" ,perl)
2883 ("pandoc" ,ghc-pandoc))) ; for documentation
2884 (home-page "http://ccb.jhu.edu/software/hisat2/index.shtml")
2885 (synopsis "Graph-based alignment of genomic sequencing reads")
2886 (description "HISAT2 is a fast and sensitive alignment program for mapping
2887next-generation sequencing reads (both DNA and RNA) to a population of human
2888genomes (as well as to a single reference genome). In addition to using one
2889global @dfn{graph FM} (GFM) index that represents a population of human
2890genomes, HISAT2 uses a large set of small GFM indexes that collectively cover
2891the whole genome. These small indexes, combined with several alignment
2892strategies, enable rapid and accurate alignment of sequencing reads. This new
2893indexing scheme is called a @dfn{Hierarchical Graph FM index} (HGFM).")
2894 ;; HISAT2 contains files from Bowtie2, which is released under
2895 ;; GPLv2 or later. The HISAT2 source files are released under
2896 ;; GPLv3 or later.
2897 (license license:gpl3+)))
2898
c684629f
BW		 2899(define-public hmmer
2900 (package
2901 (name "hmmer")
2902 (version "3.1b2")
79f09fa2
BW		 2903 (source
2904 (origin
2905 (method url-fetch)
2906 (uri (string-append
2907 "http://eddylab.org/software/hmmer"
2908 (version-prefix version 1) "/"
2909 version "/hmmer-" version ".tar.gz"))
2910 (sha256
2911 (base32
2912 "0djmgc0pfli0jilfx8hql1axhwhqxqb8rxg2r5rg07aw73sfs5nx"))
2913 (patches (search-patches "hmmer-remove-cpu-specificity.patch"))))
c684629f 2914 (build-system gnu-build-system)
b3546174 2915 (native-inputs `(("perl" ,perl)))
a83e6046 2916 (home-page "http://hmmer.org/")
c684629f
BW		 2917 (synopsis "Biosequence analysis using profile hidden Markov models")
2918 (description
2919 "HMMER is used for searching sequence databases for homologs of protein
2920sequences, and for making protein sequence alignments. It implements methods
2921using probabilistic models called profile hidden Markov models (profile
2922HMMs).")
2923 (license (list license:gpl3+
2924 ;; The bundled library 'easel' is distributed
2925 ;; under The Janelia Farm Software License.
2926 (license:non-copyleft
2927 "file://easel/LICENSE"
2928 "See easel/LICENSE in the distribution.")))))
2929
85652f59
RW		 2930(define-public htseq
2931 (package
2932 (name "htseq")
92971d68 2933 (version "0.9.1")
85652f59
RW		 2934 (source (origin
2935 (method url-fetch)
75e6639f 2936 (uri (pypi-uri "HTSeq" version))
85652f59
RW		 2937 (sha256
2938 (base32
92971d68 2939 "11flgb1381xdhk43bzbfm3vhnszkpqg6jk76rpa5xd1zbrvvlnxg"))))
85652f59 2940 (build-system python-build-system)
92971d68
BW		 2941 (native-inputs
2942 `(("python-cython" ,python-cython)))
0536727e
RW		 2943 ;; Numpy needs to be propagated when htseq is used as a Python library.
2944 (propagated-inputs
92971d68 2945 `(("python-numpy" ,python-numpy)))
578b05d9 2946 (inputs
92971d68
BW		 2947 `(("python-pysam" ,python-pysam)
2948 ("python-matplotlib" ,python-matplotlib)))
85652f59
RW		 2949 (home-page "http://www-huber.embl.de/users/anders/HTSeq/")
2950 (synopsis "Analysing high-throughput sequencing data with Python")
2951 (description
2952 "HTSeq is a Python package that provides infrastructure to process data
2953from high-throughput sequencing assays.")
2954 (license license:gpl3+)))
2955
92971d68
BW		 2956(define-public python2-htseq
2957 (package-with-python2 htseq))
2958
1ad15c16 2959(define-public java-htsjdk
15a3c3d4 2960 (package
1ad15c16 2961 (name "java-htsjdk")
15a3c3d4
RW		 2962 (version "1.129")
2963 (source (origin
2964 (method url-fetch)
2965 (uri (string-append
2966 "https://github.com/samtools/htsjdk/archive/"
2967 version ".tar.gz"))
2968 (file-name (string-append name "-" version ".tar.gz"))
2969 (sha256
2970 (base32
2971 "0asdk9b8jx2ij7yd6apg9qx03li8q7z3ml0qy2r2qczkra79y6fw"))
2972 (modules '((guix build utils)))
2973 ;; remove build dependency on git
2974 (snippet '(substitute* "build.xml"
2975 (("failifexecutionfails=\"true\"")
2976 "failifexecutionfails=\"false\"")))))
10b4a969 2977 (build-system ant-build-system)
15a3c3d4 2978 (arguments
10b4a969
RW		 2979 `(#:tests? #f ; test require Internet access
2980 #:make-flags
2981 (list (string-append "-Ddist=" (assoc-ref %outputs "out")
2982 "/share/java/htsjdk/"))
2983 #:build-target "all"
2984 #:phases
2985 (modify-phases %standard-phases
2986 ;; The build phase also installs the jars
2987 (delete 'install))))
15a3c3d4
RW		 2988 (home-page "http://samtools.github.io/htsjdk/")
2989 (synopsis "Java API for high-throughput sequencing data (HTS) formats")
2990 (description
2991 "HTSJDK is an implementation of a unified Java library for accessing
2992common file formats, such as SAM and VCF, used for high-throughput
2993sequencing (HTS) data. There are also an number of useful utilities for
2994manipulating HTS data.")
2995 (license license:expat)))
2996
e7c09730
RW		 2997(define-public htslib
2998 (package
2999 (name "htslib")
4fb6d128 3000 (version "1.5")
e7c09730
RW		 3001 (source (origin
3002 (method url-fetch)
3003 (uri (string-append
3004 "https://github.com/samtools/htslib/releases/download/"
3005 version "/htslib-" version ".tar.bz2"))
3006 (sha256
3007 (base32
4fb6d128 3008 "0bcjmnbwp2bib1z1bkrp95w9v2syzdwdfqww10mkb1hxlmg52ax0"))))
e7c09730
RW		 3009 (build-system gnu-build-system)
3010 (arguments
3011 `(#:phases
3012 (modify-phases %standard-phases
3013 (add-after
3014 'unpack 'patch-tests
3015 (lambda _
3016 (substitute* "test/test.pl"
3017 (("/bin/bash") (which "bash")))
3018 #t)))))
3019 (inputs
a9e4a1e6
BW		 3020 `(("openssl" ,openssl)
3021 ("curl" ,curl)
3022 ("zlib" ,zlib)))
e7c09730
RW		 3023 (native-inputs
3024 `(("perl" ,perl)))
3025 (home-page "http://www.htslib.org")
3026 (synopsis "C library for reading/writing high-throughput sequencing data")
3027 (description
3028 "HTSlib is a C library for reading/writing high-throughput sequencing
3029data. It also provides the bgzip, htsfile, and tabix utilities.")
3030 ;; Files under cram/ are released under the modified BSD license;
3031 ;; the rest is released under the Expat license
3032 (license (list license:expat license:bsd-3))))
3033
bca2c576
BW		 3034;; This package should be removed once no packages rely upon it.
3035(define htslib-1.3
3036 (package
3037 (inherit htslib)
3038 (version "1.3.1")
3039 (source (origin
3040 (method url-fetch)
3041 (uri (string-append
3042 "https://github.com/samtools/htslib/releases/download/"
3043 version "/htslib-" version ".tar.bz2"))
3044 (sha256
3045 (base32
3046 "1rja282fwdc25ql6izkhdyh8ppw8x2fs0w0js78zgkmqjlikmma9"))))))
3047
c4325f62
RW		 3048(define-public idr
3049 (package
3050 (name "idr")
3051 (version "2.0.0")
3052 (source (origin
3053 (method url-fetch)
3054 (uri (string-append
3055 "https://github.com/nboley/idr/archive/"
3056 version ".tar.gz"))
3057 (file-name (string-append name "-" version ".tar.gz"))
3058 (sha256
3059 (base32
3060 "1k3x44biak00aiv3hpm1yd6nn4hhp7n0qnbs3zh2q9sw7qr1qj5r"))))
3061 (build-system python-build-system)
3062 (arguments
14386fc7 3063 `(#:tests? #f)) ; FIXME: "ImportError: No module named 'utility'"
b7a820fc 3064 (propagated-inputs
c4325f62 3065 `(("python-scipy" ,python-scipy)
b7a820fc 3066 ("python-sympy" ,python-sympy)
c4325f62
RW		 3067 ("python-numpy" ,python-numpy)
3068 ("python-matplotlib" ,python-matplotlib)))
3069 (native-inputs
f3b98f4f 3070 `(("python-cython" ,python-cython)))
c4325f62
RW		 3071 (home-page "https://github.com/nboley/idr")
3072 (synopsis "Tool to measure the irreproducible discovery rate (IDR)")
3073 (description
3074 "The IDR (Irreproducible Discovery Rate) framework is a unified approach
3075to measure the reproducibility of findings identified from replicate
3076experiments and provide highly stable thresholds based on reproducibility.")
3077 (license license:gpl3+)))
3078
43c565d2
RW		 3079(define-public jellyfish
3080 (package
3081 (name "jellyfish")
3082 (version "2.2.4")
3083 (source (origin
3084 (method url-fetch)
3085 (uri (string-append "https://github.com/gmarcais/Jellyfish/"
3086 "releases/download/v" version
3087 "/jellyfish-" version ".tar.gz"))
3088 (sha256
3089 (base32
3090 "0a6xnynqy2ibfbfz86b9g2m2dgm7f1469pmymkpam333gi3p26nk"))))
3091 (build-system gnu-build-system)
3092 (outputs '("out" ;for library
3093 "ruby" ;for Ruby bindings
3094 "python")) ;for Python bindings
3095 (arguments
3096 `(#:configure-flags
3097 (list (string-append "--enable-ruby-binding="
3098 (assoc-ref %outputs "ruby"))
3099 (string-append "--enable-python-binding="
3100 (assoc-ref %outputs "python")))
3101 #:phases
3102 (modify-phases %standard-phases
3103 (add-before 'check 'set-SHELL-variable
3104 (lambda _
3105 ;; generator_manager.hpp either uses /bin/sh or $SHELL
3106 ;; to run tests.
3107 (setenv "SHELL" (which "bash"))
3108 #t)))))
3109 (native-inputs
3110 `(("bc" ,bc)
3111 ("time" ,time)
3112 ("ruby" ,ruby)
3113 ("python" ,python-2)))
3114 (synopsis "Tool for fast counting of k-mers in DNA")
3115 (description
3116 "Jellyfish is a tool for fast, memory-efficient counting of k-mers in
3117DNA. A k-mer is a substring of length k, and counting the occurrences of all
3118such substrings is a central step in many analyses of DNA sequence. Jellyfish
3119is a command-line program that reads FASTA and multi-FASTA files containing
3120DNA sequences. It outputs its k-mer counts in a binary format, which can be
3121translated into a human-readable text format using the @code{jellyfish dump}
3122command, or queried for specific k-mers with @code{jellyfish query}.")
3123 (home-page "http://www.genome.umd.edu/jellyfish.html")
6e8faf77
EF		 3124 ;; From their website: JELLYFISH runs on 64-bit Intel-compatible processors
3125 (supported-systems '("x86_64-linux"))
43c565d2
RW		 3126 ;; The combined work is published under the GPLv3 or later. Individual
3127 ;; files such as lib/jsoncpp.cpp are released under the Expat license.
3128 (license (list license:gpl3+ license:expat))))
3129
94ff3157
BW		 3130(define-public khmer
3131 (package
3132 (name "khmer")
3133 (version "2.0")
3134 (source
3135 (origin
3136 (method url-fetch)
3137 (uri (pypi-uri "khmer" version))
3138 (sha256
3139 (base32
3140 "0wb05shqh77v00256qlm68vbbx3kl76fyzihszbz5nhanl4ni33a"))
3141 (patches (search-patches "khmer-use-libraries.patch"))))
3142 (build-system python-build-system)
3143 (arguments
3144 `(#:phases
3145 (modify-phases %standard-phases
3146 (add-after 'unpack 'set-paths
3147 (lambda* (#:key inputs outputs #:allow-other-keys)
3148 ;; Delete bundled libraries.
3149 (delete-file-recursively "third-party/zlib")
3150 (delete-file-recursively "third-party/bzip2")
3151 ;; Replace bundled seqan.
3152 (let* ((seqan-all "third-party/seqan")
3153 (seqan-include (string-append
3154 seqan-all "/core/include")))
3155 (delete-file-recursively seqan-all)
3156 (copy-recursively (string-append (assoc-ref inputs "seqan")
3157 "/include/seqan")
3158 (string-append seqan-include "/seqan")))
3159 ;; We do not replace the bundled MurmurHash as the canonical
3160 ;; repository for this code 'SMHasher' is unsuitable for
3161 ;; providing a library. See
3162 ;; https://lists.gnu.org/archive/html/guix-devel/2016-06/msg00977.html
3163 #t))
3164 (add-after 'unpack 'set-cc
3165 (lambda _
3166 (setenv "CC" "gcc")
3167 #t))
3168 ;; It is simpler to test after installation.
3169 (delete 'check)
3170 (add-after 'install 'post-install-check
3171 (lambda* (#:key inputs outputs #:allow-other-keys)
3172 (let ((out (assoc-ref outputs "out")))
3173 (setenv "PATH"
3174 (string-append
3175 (getenv "PATH")
3176 ":"
3177 (assoc-ref outputs "out")
3178 "/bin"))
3179 (setenv "PYTHONPATH"
3180 (string-append
3181 (getenv "PYTHONPATH")
3182 ":"
3183 out
3184 "/lib/python"
3185 (string-take (string-take-right
3186 (assoc-ref inputs "python") 5) 3)
3187 "/site-packages"))
3188 (with-directory-excursion "build"
3189 (zero? (system* "nosetests" "khmer" "--attr"
3190 "!known_failing")))))))))
3191 (native-inputs
3192 `(("seqan" ,seqan)
3193 ("python-nose" ,python-nose)))
3194 (inputs
3195 `(("zlib" ,zlib)
3196 ("bzip2" ,bzip2)
3197 ("python-screed" ,python-screed)
cf1d5814
BW		 3198 ("python-bz2file" ,python-bz2file)
3199 ;; Tests fail when gcc-5 is used for compilation. Use gcc-4.9 at least
3200 ;; until the next version of khmer (likely 2.1) is released.
3201 ("gcc" ,gcc-4.9)))
94ff3157
BW		 3202 (home-page "https://khmer.readthedocs.org/")
3203 (synopsis "K-mer counting, filtering and graph traversal library")
3204 (description "The khmer software is a set of command-line tools for
3205working with DNA shotgun sequencing data from genomes, transcriptomes,
3206metagenomes and single cells. Khmer can make de novo assemblies faster, and
3207sometimes better. Khmer can also identify and fix problems with shotgun
3208data.")
8157af2e
EF		 3209 ;; When building on i686, armhf and mips64el, we get the following error:
3210 ;; error: ['khmer', 'khmer.tests', 'oxli'] require 64-bit operating system
3211 (supported-systems '("x86_64-linux"))
94ff3157
BW		 3212 (license license:bsd-3)))
3213
b9a601d9
RJ		 3214(define-public kaiju
3215 (package
3216 (name "kaiju")
3217 (version "1.5.0")
3218 (source (origin
3219 (method url-fetch)
3220 (uri (string-append
3221 "https://github.com/bioinformatics-centre/kaiju/archive/v"
3222 version ".tar.gz"))
3223 (file-name (string-append name "-" version ".tar.gz"))
3224 (sha256
3225 (base32
3226 "0afbfalfw9y39bkwnqjrh9bghs118ws1pzj5h8l0nblgn3mbjdks"))))
3227 (build-system gnu-build-system)
3228 (arguments
3229 `(#:tests? #f ; There are no tests.
3230 #:phases
3231 (modify-phases %standard-phases
3232 (delete 'configure)
3233 (add-before 'build 'move-to-src-dir
3234 (lambda _ (chdir "src") #t))
3235 (replace 'install
3236 (lambda* (#:key inputs outputs #:allow-other-keys)
3237 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
3238 (mkdir-p bin)
3239 (chdir "..")
3240 (copy-recursively "bin" bin)
3241 (copy-recursively "util" bin))
3242 #t)))))
3243 (inputs
3244 `(("perl" ,perl)))
3245 (home-page "http://kaiju.binf.ku.dk/")
3246 (synopsis "Fast and sensitive taxonomic classification for metagenomics")
3247 (description "Kaiju is a program for sensitive taxonomic classification
3248of high-throughput sequencing reads from metagenomic whole genome sequencing
3249experiments.")
3250 (license license:gpl3+)))
3251
d57e6d0f
RW		 3252(define-public macs
3253 (package
3254 (name "macs")
0d0bcaa0 3255 (version "2.1.0.20151222")
d57e6d0f
RW		 3256 (source (origin
3257 (method url-fetch)
43ec07f1 3258 (uri (pypi-uri "MACS2" version))
d57e6d0f
RW		 3259 (sha256
3260 (base32
0d0bcaa0 3261 "1r2hcz6irhcq7lwbafjks98jbn34hv05avgbdjnp6w6mlfjkf8x5"))))
d57e6d0f
RW		 3262 (build-system python-build-system)
3263 (arguments
3264 `(#:python ,python-2 ; only compatible with Python 2.7
3265 #:tests? #f)) ; no test target
3266 (inputs
3267 `(("python-numpy" ,python2-numpy)))
7bf837fd 3268 (home-page "https://github.com/taoliu/MACS/")
d57e6d0f
RW		 3269 (synopsis "Model based analysis for ChIP-Seq data")
3270 (description
3271 "MACS is an implementation of a ChIP-Seq analysis algorithm for
3272identifying transcript factor binding sites named Model-based Analysis of
3273ChIP-Seq (MACS). MACS captures the influence of genome complexity to evaluate
3274the significance of enriched ChIP regions and it improves the spatial
3275resolution of binding sites through combining the information of both
3276sequencing tag position and orientation.")
3277 (license license:bsd-3)))
3278
41ddebdd
BW		 3279(define-public mafft
3280 (package
3281 (name "mafft")
88eb119f 3282 (version "7.310")
41ddebdd
BW		 3283 (source (origin
3284 (method url-fetch)
3285 (uri (string-append
3286 "http://mafft.cbrc.jp/alignment/software/mafft-" version
3287 "-without-extensions-src.tgz"))
3288 (file-name (string-append name "-" version ".tgz"))
3289 (sha256
3290 (base32
88eb119f 3291 "0gbsaz6z2qa307kd7wfb06c3y4ikmv1hsdvlns11f6zq4w1z9pwc"))))
41ddebdd
BW		 3292 (build-system gnu-build-system)
3293 (arguments
3294 `(#:tests? #f ; no automated tests, though there are tests in the read me
3295 #:make-flags (let ((out (assoc-ref %outputs "out")))
3296 (list (string-append "PREFIX=" out)
3297 (string-append "BINDIR="
3298 (string-append out "/bin"))))
3299 #:phases
3300 (modify-phases %standard-phases
3301 (add-after 'unpack 'enter-dir
101e8f71 3302 (lambda _ (chdir "core") #t))
41ddebdd 3303 (add-after 'enter-dir 'patch-makefile
101e8f71
BW		 3304 (lambda _
3305 ;; on advice from the MAFFT authors, there is no need to
3306 ;; distribute mafft-profile, mafft-distance, or
3307 ;; mafft-homologs.rb as they are too "specialised".
3308 (substitute* "Makefile"
3309 ;; remove mafft-homologs.rb from SCRIPTS
3310 (("^SCRIPTS = mafft mafft-homologs.rb")
3311 "SCRIPTS = mafft")
3312 ;; remove mafft-homologs from MANPAGES
3313 (("^MANPAGES = mafft.1 mafft-homologs.1")
3314 "MANPAGES = mafft.1")
3315 ;; remove mafft-distance from PROGS
3316 (("^PROGS = dvtditr dndfast7 dndblast sextet5 mafft-distance")
3317 "PROGS = dvtditr dndfast7 dndblast sextet5")
3318 ;; remove mafft-profile from PROGS
3319 (("splittbfast disttbfast tbfast mafft-profile 2cl mccaskillwrap")
3320 "splittbfast disttbfast tbfast f2cl mccaskillwrap")
3321 (("^rm -f mafft-profile mafft-profile.exe") "#")
3322 (("^rm -f mafft-distance mafft-distance.exe") ")#")
3323 ;; do not install MAN pages in libexec folder
3324 (("^\t\\$\\(INSTALL\\) -m 644 \\$\\(MANPAGES\\) \
41ddebdd 3325\\$\\(DESTDIR\\)\\$\\(LIBDIR\\)") "#"))
101e8f71 3326 #t))
02f35bb5
BW		 3327 (add-after 'enter-dir 'patch-paths
3328 (lambda* (#:key inputs #:allow-other-keys)
3329 (substitute* '("pairash.c"
3330 "mafft.tmpl")
3331 (("perl") (which "perl"))
3332 (("([\"`| ])awk" _ prefix)
3333 (string-append prefix (which "awk")))
3334 (("grep") (which "grep")))
3335 #t))
101e8f71
BW		 3336 (delete 'configure)
3337 (add-after 'install 'wrap-programs
3338 (lambda* (#:key outputs #:allow-other-keys)
3339 (let* ((out (assoc-ref outputs "out"))
3340 (bin (string-append out "/bin"))
3341 (path (string-append
3342 (assoc-ref %build-inputs "coreutils") "/bin:")))
3343 (for-each (lambda (file)
3344 (wrap-program file
3345 `("PATH" ":" prefix (,path))))
3346 (find-files bin)))
3347 #t)))))
41ddebdd 3348 (inputs
02f35bb5 3349 `(("perl" ,perl)
71461f88 3350 ("ruby" ,ruby)
02f35bb5 3351 ("gawk" ,gawk)
101e8f71
BW		 3352 ("grep" ,grep)
3353 ("coreutils" ,coreutils)))
41ddebdd
BW		 3354 (home-page "http://mafft.cbrc.jp/alignment/software/")
3355 (synopsis "Multiple sequence alignment program")
3356 (description
3357 "MAFFT offers a range of multiple alignment methods for nucleotide and
3358protein sequences. For instance, it offers L-INS-i (accurate; for alignment
3359of <~200 sequences) and FFT-NS-2 (fast; for alignment of <~30,000
3360sequences).")
3361 (license (license:non-copyleft
3362 "http://mafft.cbrc.jp/alignment/software/license.txt"
3363 "BSD-3 with different formatting"))))
8fd790eb 3364
84be3b99
MB		 3365(define-public mash
3366 (package
3367 (name "mash")
3368 (version "1.1.1")
3369 (source (origin
3370 (method url-fetch)
3371 (uri (string-append
3372 "https://github.com/marbl/mash/archive/v"
3373 version ".tar.gz"))
3374 (file-name (string-append name "-" version ".tar.gz"))
3375 (sha256
3376 (base32
3377 "08znbvqq5xknfhmpp3wcj574zvi4p7i8zifi67c9qw9a6ikp42fj"))
3378 (modules '((guix build utils)))
3379 (snippet
3380 ;; Delete bundled kseq.
3381 ;; TODO: Also delete bundled murmurhash and open bloom filter.
3382 '(delete-file "src/mash/kseq.h"))))
3383 (build-system gnu-build-system)
3384 (arguments
3385 `(#:tests? #f ; No tests.
3386 #:configure-flags
3387 (list
3388 (string-append "--with-capnp=" (assoc-ref %build-inputs "capnproto"))
3389 (string-append "--with-gsl=" (assoc-ref %build-inputs "gsl")))
3390 #:make-flags (list "CC=gcc")
3391 #:phases
3392 (modify-phases %standard-phases
3393 (add-after 'unpack 'fix-includes
3394 (lambda _
3395 (substitute* '("src/mash/Sketch.cpp" "src/mash/CommandFind.cpp")
3396 (("^#include \"kseq\\.h\"")
3397 "#include \"htslib/kseq.h\""))
3398 #t))
d10092b8 3399 (add-after 'fix-includes 'autoconf
84be3b99
MB		 3400 (lambda _ (zero? (system* "autoconf")))))))
3401 (native-inputs
3402 `(("autoconf" ,autoconf)
3403 ;; Capnproto and htslib are statically embedded in the final
3404 ;; application. Therefore we also list their licenses, below.
3405 ("capnproto" ,capnproto)
3406 ("htslib" ,htslib)))
3407 (inputs
3408 `(("gsl" ,gsl)
3409 ("zlib" ,zlib)))
3410 (supported-systems '("x86_64-linux"))
3411 (home-page "https://mash.readthedocs.io")
3412 (synopsis "Fast genome and metagenome distance estimation using MinHash")
3413 (description "Mash is a fast sequence distance estimator that uses the
3414MinHash algorithm and is designed to work with genomes and metagenomes in the
3415form of assemblies or reads.")
3416 (license (list license:bsd-3 ; Mash
3417 license:expat ; HTSlib and capnproto
3418 license:public-domain ; MurmurHash 3
3419 license:cpl1.0)))) ; Open Bloom Filter
3420
8fd790eb 3421(define-public metabat
f3f68a44
BW		 3422 (package
3423 (name "metabat")
25bd1fc1 3424 (version "2.12.1")
f3f68a44
BW		 3425 (source
3426 (origin
25bd1fc1
BW		 3427 (method url-fetch)
3428 (uri (string-append "https://bitbucket.org/berkeleylab/metabat/get/v"
3429 version ".tar.gz"))
3430 (file-name (string-append name "-" version ".tar.gz"))
f3f68a44
BW		 3431 (sha256
3432 (base32
25bd1fc1
BW		 3433 "1hmvdalz3zj5sqqklg0l4npjdv37cv2hsdi1al9iby2ndxjs1b73"))
3434 (patches (search-patches "metabat-fix-compilation.patch"))))
8fd790eb
BW		 3435 (build-system gnu-build-system)
3436 (arguments
3437 `(#:phases
3438 (modify-phases %standard-phases
3439 (add-after 'unpack 'fix-includes
45469ebe
BW		 3440 (lambda _
3441 (substitute* "src/BamUtils.h"
3442 (("^#include \"bam/bam\\.h\"")
3443 "#include \"samtools/bam.h\"")
3444 (("^#include \"bam/sam\\.h\"")
3445 "#include \"samtools/sam.h\""))
3446 (substitute* "src/KseqReader.h"
3447 (("^#include \"bam/kseq\\.h\"")
3448 "#include \"htslib/kseq.h\""))
3449 #t))
8fd790eb 3450 (add-after 'unpack 'fix-scons
f3f68a44
BW		 3451 (lambda* (#:key inputs #:allow-other-keys)
3452 (substitute* "SConstruct"
3453 (("^htslib_dir += 'samtools'")
3454 (string-append "htslib_dir = '"
3455 (assoc-ref inputs "htslib")
3456 "'"))
3457 (("^samtools_dir = 'samtools'")
3458 (string-append "samtools_dir = '"
3459 (assoc-ref inputs "samtools")
3460 "'"))
3461 (("^findStaticOrShared\\('bam', hts_lib")
3462 (string-append "findStaticOrShared('bam', '"
3463 (assoc-ref inputs "samtools")
3464 "/lib'"))
3465 ;; Do not distribute README.
3466 (("^env\\.Install\\(idir_prefix, 'README\\.md'\\)") ""))
3467 #t))
8fd790eb
BW		 3468 (delete 'configure)
3469 (replace 'build
f3f68a44
BW		 3470 (lambda* (#:key inputs outputs #:allow-other-keys)
3471 (mkdir (assoc-ref outputs "out"))
3472 (zero? (system* "scons"
3473 (string-append
3474 "PREFIX="
3475 (assoc-ref outputs "out"))
3476 (string-append
3477 "BOOST_ROOT="
3478 (assoc-ref inputs "boost"))
3479 "install"))))
45469ebe 3480 ;; Check and install are carried out during build phase.
8fd790eb
BW		 3481 (delete 'check)
3482 (delete 'install))))
3483 (inputs
3484 `(("zlib" ,zlib)
3485 ("perl" ,perl)
3486 ("samtools" ,samtools)
3487 ("htslib" ,htslib)
3488 ("boost" ,boost)))
3489 (native-inputs
3490 `(("scons" ,scons)))
3491 (home-page "https://bitbucket.org/berkeleylab/metabat")
3492 (synopsis
3493 "Reconstruction of single genomes from complex microbial communities")
3494 (description
3495 "Grouping large genomic fragments assembled from shotgun metagenomic
3496sequences to deconvolute complex microbial communities, or metagenome binning,
3497enables the study of individual organisms and their interactions. MetaBAT is
3498an automated metagenome binning software, which integrates empirical
3499probabilistic distances of genome abundance and tetranucleotide frequency.")
d931a4bb
EF		 3500 ;; The source code contains inline assembly.
3501 (supported-systems '("x86_64-linux" "i686-linux"))
f3f68a44
BW		 3502 (license (license:non-copyleft "file://license.txt"
3503 "See license.txt in the distribution."))))
8fd790eb 3504
318c0aee
MB		 3505(define-public minced
3506 (package
3507 (name "minced")
3508 (version "0.2.0")
3509 (source (origin
3510 (method url-fetch)
3511 (uri (string-append
3512 "https://github.com/ctSkennerton/minced/archive/"
3513 version ".tar.gz"))
3514 (file-name (string-append name "-" version ".tar.gz"))
3515 (sha256
3516 (base32
3517 "0wxmlsapxfpxfd3ps9636h7i2xy6la8i42mwh0j2lsky63h63jp1"))))
3518 (build-system gnu-build-system)
3519 (arguments
3520 `(#:test-target "test"
3521 #:phases
3522 (modify-phases %standard-phases
3523 (delete 'configure)
3524 (add-before 'check 'fix-test
3525 (lambda _
3526 ;; Fix test for latest version.
3527 (substitute* "t/Aquifex_aeolicus_VF5.expected"
3528 (("minced:0.1.6") "minced:0.2.0"))
3529 #t))
3530 (replace 'install ; No install target.
3531 (lambda* (#:key inputs outputs #:allow-other-keys)
3532 (let* ((out (assoc-ref outputs "out"))
3533 (bin (string-append out "/bin"))
3534 (wrapper (string-append bin "/minced")))
3535 ;; Minced comes with a wrapper script that tries to figure out where
3536 ;; it is located before running the JAR. Since these paths are known
3537 ;; to us, we build our own wrapper to avoid coreutils dependency.
3538 (install-file "minced.jar" bin)
3539 (with-output-to-file wrapper
3540 (lambda _
3541 (display
3542 (string-append
3543 "#!" (assoc-ref inputs "bash") "/bin/sh\n\n"
3544 (assoc-ref inputs "jre") "/bin/java -jar "
3545 bin "/minced.jar \"$@\"\n"))))
3546 (chmod wrapper #o555)))))))
3547 (native-inputs
3548 `(("jdk" ,icedtea "jdk")))
3549 (inputs
3550 `(("bash" ,bash)
3551 ("jre" ,icedtea "out")))
3552 (home-page "https://github.com/ctSkennerton/minced")
3553 (synopsis "Mining CRISPRs in Environmental Datasets")
3554 (description
3555 "MinCED is a program to find Clustered Regularly Interspaced Short
3556Palindromic Repeats (CRISPRs) in DNA sequences. It can be used for
3557unassembled metagenomic reads, but is mainly designed for full genomes and
3558assembled metagenomic sequence.")
3559 (license license:gpl3+)))
3560
ddd82e0e
RW		 3561(define-public miso
3562 (package
3563 (name "miso")
ce4dfde0 3564 (version "0.5.4")
ddd82e0e
RW		 3565 (source (origin
3566 (method url-fetch)
34260a10 3567 (uri (pypi-uri "misopy" version))
ddd82e0e
RW		 3568 (sha256
3569 (base32
ce4dfde0 3570 "1z3x0vd8ma7pdrnywj7i3kgwl89sdkwrrn62zl7r5calqaq2hyip"))
ddd82e0e 3571 (modules '((guix build utils)))
ddd82e0e
RW		 3572 (snippet
3573 '(substitute* "setup.py"
0ec8b206
RW		 3574 ;; Use setuptools, or else the executables are not
3575 ;; installed.
3576 (("distutils.core") "setuptools")
3577 ;; use "gcc" instead of "cc" for compilation
ddd82e0e
RW		 3578 (("^defines")
3579 "cc.set_executables(
3580compiler='gcc',
3581compiler_so='gcc',
3582linker_exe='gcc',
3583linker_so='gcc -shared'); defines")))))
3584 (build-system python-build-system)
3585 (arguments
3586 `(#:python ,python-2 ; only Python 2 is supported
3587 #:tests? #f)) ; no "test" target
3588 (inputs
3589 `(("samtools" ,samtools)
3590 ("python-numpy" ,python2-numpy)
3591 ("python-pysam" ,python2-pysam)
3592 ("python-scipy" ,python2-scipy)
3593 ("python-matplotlib" ,python2-matplotlib)))
3594 (native-inputs
f3b98f4f
HG		 3595 `(("python-mock" ,python2-mock) ;for tests
3596 ("python-pytz" ,python2-pytz))) ;for tests
ddd82e0e
RW		 3597 (home-page "http://genes.mit.edu/burgelab/miso/index.html")
3598 (synopsis "Mixture of Isoforms model for RNA-Seq isoform quantitation")
3599 (description
3600 "MISO (Mixture-of-Isoforms) is a probabilistic framework that quantitates
3601the expression level of alternatively spliced genes from RNA-Seq data, and
3602identifies differentially regulated isoforms or exons across samples. By
3603modeling the generative process by which reads are produced from isoforms in
3604RNA-Seq, the MISO model uses Bayesian inference to compute the probability
3605that a read originated from a particular isoform.")
3606 (license license:gpl2)))
3607
324efb88
BW		 3608(define-public muscle
3609 (package
3610 (name "muscle")
3611 (version "3.8.1551")
3612 (source (origin
3613 (method url-fetch/tarbomb)
324efb88
BW		 3614 (uri (string-append
3615 "http://www.drive5.com/muscle/muscle_src_"
3616 version ".tar.gz"))
3617 (sha256
3618 (base32
3619 "0bj8kj7sdizy3987zx6w7axihk40fk8rn76mpbqqjcnd64i5a367"))))
3620 (build-system gnu-build-system)
3621 (arguments
3622 `(#:make-flags (list "LDLIBS = -lm")
3623 #:phases
3624 (modify-phases %standard-phases
3625 (delete 'configure)
3626 (replace 'check
3627 ;; There are no tests, so just test if it runs.
3628 (lambda _ (zero? (system* "./muscle" "-version"))))
3629 (replace 'install
3630 (lambda* (#:key outputs #:allow-other-keys)
3631 (let* ((out (assoc-ref outputs "out"))
3632 (bin (string-append out "/bin")))
3633 (install-file "muscle" bin)))))))
3634 (home-page "http://www.drive5.com/muscle")
3635 (synopsis "Multiple sequence alignment program")
3636 (description
3637 "MUSCLE aims to be a fast and accurate multiple sequence alignment
3638program for nucleotide and protein sequences.")
3639 ;; License information found in 'muscle -h' and usage.cpp.
3640 (license license:public-domain)))
3641
99268755
BW		 3642(define-public newick-utils
3643 ;; There are no recent releases so we package from git.
3644 (let ((commit "da121155a977197cab9fbb15953ca1b40b11eb87"))
3645 (package
3646 (name "newick-utils")
3647 (version (string-append "1.6-1." (string-take commit 8)))
3648 (source (origin
3649 (method git-fetch)
3650 (uri (git-reference
3651 (url "https://github.com/tjunier/newick_utils.git")
3652 (commit commit)))
3653 (file-name (string-append name "-" version "-checkout"))
3654 (sha256
3655 (base32
3656 "1hkw21rq1mwf7xp0rmbb2gqc0i6p11108m69i7mr7xcjl268pxnb"))))
3657 (build-system gnu-build-system)
3658 (arguments
3659 `(#:phases
3660 (modify-phases %standard-phases
3661 (add-after 'unpack 'autoconf
3662 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
3663 (inputs
3664 ;; XXX: TODO: Enable Lua and Guile bindings.
3665 ;; https://github.com/tjunier/newick_utils/issues/13
3666 `(("libxml2" ,libxml2)
3667 ("flex" ,flex)
3668 ("bison" ,bison)))
3669 (native-inputs
3670 `(("autoconf" ,autoconf)
3671 ("automake" ,automake)
3672 ("libtool" ,libtool)))
3673 (synopsis "Programs for working with newick format phylogenetic trees")
3674 (description
3675 "Newick-utils is a suite of utilities for processing phylogenetic trees
3676in Newick format. Functions include re-rooting, extracting subtrees,
3677trimming, pruning, condensing, drawing (ASCII graphics or SVG).")
3678 (home-page "https://github.com/tjunier/newick_utils")
3679 (license license:bsd-3))))
3680
1e44cf8b
BW		 3681(define-public orfm
3682 (package
3683 (name "orfm")
dfc83ead 3684 (version "0.7.1")
1e44cf8b
BW		 3685 (source (origin
3686 (method url-fetch)
3687 (uri (string-append
3688 "https://github.com/wwood/OrfM/releases/download/v"
3689 version "/orfm-" version ".tar.gz"))
3690 (sha256
3691 (base32
dfc83ead 3692 "16iigyr2gd8x0imzkk1dr3k5xsds9bpmwg31ayvjg0f4pir9rwqr"))))
1e44cf8b
BW		 3693 (build-system gnu-build-system)
3694 (inputs `(("zlib" ,zlib)))
6b6f7d6a
BW		 3695 (native-inputs
3696 `(("ruby-bio-commandeer" ,ruby-bio-commandeer)
3697 ("ruby-rspec" ,ruby-rspec)
3698 ("ruby" ,ruby)))
1e44cf8b
BW		 3699 (synopsis "Simple and not slow open reading frame (ORF) caller")
3700 (description
6b6f7d6a 3701 "An ORF caller finds stretches of DNA that, when translated, are not
1e44cf8b
BW		 3702interrupted by stop codons. OrfM finds and prints these ORFs.")
3703 (home-page "https://github.com/wwood/OrfM")
3704 (license license:lgpl3+)))
3705
c033f5d6
BW		 3706(define-public pplacer
3707 (let ((commit "g807f6f3"))
3708 (package
3709 (name "pplacer")
3710 ;; The commit should be updated with each version change.
3711 (version "1.1.alpha19")
3712 (source
3713 (origin
3714 (method url-fetch)
3715 (uri (string-append "https://github.com/matsen/pplacer/archive/v"
3716 version ".tar.gz"))
3717 (file-name (string-append name "-" version ".tar.gz"))
3718 (sha256
3719 (base32 "0z1lnd2s8sh6kpzg106wzbh2szw7h0hvq8syd5a6wv4rmyyz6x0f"))))
3720 (build-system ocaml-build-system)
3721 (arguments
3722 `(#:ocaml ,ocaml-4.01
3723 #:findlib ,ocaml4.01-findlib
3724 #:modules ((guix build ocaml-build-system)
3725 (guix build utils)
3726 (ice-9 ftw))
3727 #:phases
3728 (modify-phases %standard-phases
3729 (delete 'configure)
3730 (add-after 'unpack 'replace-bundled-cddlib
3731 (lambda* (#:key inputs #:allow-other-keys)
3732 (let* ((cddlib-src (assoc-ref inputs "cddlib-src"))
3733 (local-dir "cddlib_guix"))
3734 (mkdir local-dir)
3735 (with-directory-excursion local-dir
3736 (system* "tar" "xvf" cddlib-src))
3737 (let ((cddlib-src-folder
3738 (string-append local-dir "/"
3739 (list-ref (scandir local-dir) 2)
3740 "/lib-src")))
3741 (for-each
3742 (lambda (file)
3743 (copy-file file
3744 (string-append "cdd_src/" (basename file))))
3745 (find-files cddlib-src-folder ".*[ch]$")))
3746 #t)))
3747 (add-after 'unpack 'fix-makefile
3748 (lambda _
3749 ;; Remove system calls to 'git'.
3750 (substitute* "Makefile"
3751 (("^DESCRIPT:=pplacer-.*")
3752 (string-append
3753 "DESCRIPT:=pplacer-$(shell uname)-v" ,version "\n")))
3754 (substitute* "myocamlbuild.ml"
3755 (("git describe --tags --long .*\\\" with")
3756 (string-append
3757 "echo -n v" ,version "-" ,commit "\" with")))
3758 #t))
3759 (replace 'install
3760 (lambda* (#:key outputs #:allow-other-keys)
3761 (let* ((out (assoc-ref outputs "out"))
3762 (bin (string-append out "/bin")))
3763 (copy-recursively "bin" bin))
3764 #t)))))
3765 (native-inputs
3766 `(("zlib" ,zlib)
3767 ("gsl" ,gsl)
3768 ("ocaml-ounit" ,ocaml4.01-ounit)
3769 ("ocaml-batteries" ,ocaml4.01-batteries)
3770 ("ocaml-camlzip" ,ocaml4.01-camlzip)
3771 ("ocaml-csv" ,ocaml4.01-csv)
3772 ("ocaml-sqlite3" ,ocaml4.01-sqlite3)
3773 ("ocaml-xmlm" ,ocaml4.01-xmlm)
3774 ("ocaml-mcl" ,ocaml4.01-mcl)
3775 ("ocaml-gsl" ,ocaml4.01-gsl)
3776 ("cddlib-src" ,(package-source cddlib))))
3777 (propagated-inputs
3778 `(("pplacer-scripts" ,pplacer-scripts)))
3779 (synopsis "Phylogenetic placement of biological sequences")
3780 (description
3781 "Pplacer places query sequences on a fixed reference phylogenetic tree
3782to maximize phylogenetic likelihood or posterior probability according to a
3783reference alignment. Pplacer is designed to be fast, to give useful
3784information about uncertainty, and to offer advanced visualization and
3785downstream analysis.")
3786 (home-page "http://matsen.fhcrc.org/pplacer")
3787 (license license:gpl3))))
3788
3789;; This package is installed alongside 'pplacer'. It is a separate package so
3790;; that it can use the python-build-system for the scripts that are
3791;; distributed alongside the main OCaml binaries.
3792(define pplacer-scripts
3793 (package
3794 (inherit pplacer)
3795 (name "pplacer-scripts")
3796 (build-system python-build-system)
3797 (arguments
3798 `(#:python ,python-2
3799 #:phases
3800 (modify-phases %standard-phases
3801 (add-after 'unpack 'enter-scripts-dir
3802 (lambda _ (chdir "scripts")))
3803 (replace 'check
3804 (lambda _
3805 (zero? (system* "python" "-m" "unittest" "discover" "-v"))))
3806 (add-after 'install 'wrap-executables
3807 (lambda* (#:key inputs outputs #:allow-other-keys)
3808 (let* ((out (assoc-ref outputs "out"))
3809 (bin (string-append out "/bin")))
3810 (let ((path (string-append
3811 (assoc-ref inputs "hmmer") "/bin:"
3812 (assoc-ref inputs "infernal") "/bin")))
3813 (display path)
3814 (wrap-program (string-append bin "/refpkg_align.py")
3815 `("PATH" ":" prefix (,path))))
3816 (let ((path (string-append
3817 (assoc-ref inputs "hmmer") "/bin")))
3818 (wrap-program (string-append bin "/hrefpkg_query.py")
3819 `("PATH" ":" prefix (,path)))))
3820 #t)))))
3821 (inputs
3822 `(("infernal" ,infernal)
3823 ("hmmer" ,hmmer)))
3824 (propagated-inputs
3825 `(("python-biopython" ,python2-biopython)
3826 ("taxtastic" ,taxtastic)))
3827 (synopsis "Pplacer Python scripts")))
3828
19ee9201
RW		 3829(define-public python2-pbcore
3830 (package
3831 (name "python2-pbcore")
e301bfc8 3832 (version "1.2.10")
19ee9201
RW		 3833 (source (origin
3834 (method url-fetch)
ddb83129 3835 (uri (pypi-uri "pbcore" version))
19ee9201
RW		 3836 (sha256
3837 (base32
e301bfc8 3838 "1kjmv891d6qbpp4shhhvkl02ff4q5xlpnls2513sm2cjcrs52f1i"))))
19ee9201
RW		 3839 (build-system python-build-system)
3840 (arguments `(#:python ,python-2)) ; pbcore requires Python 2.7
de5bc890 3841 (propagated-inputs
19ee9201
RW		 3842 `(("python-cython" ,python2-cython)
3843 ("python-numpy" ,python2-numpy)
3844 ("python-pysam" ,python2-pysam)
3845 ("python-h5py" ,python2-h5py)))
3846 (native-inputs
de5bc890
HG		 3847 `(("python-nose" ,python2-nose)
3848 ("python-sphinx" ,python2-sphinx)
3849 ("python-pyxb" ,python2-pyxb)))
19ee9201
RW		 3850 (home-page "http://pacificbiosciences.github.io/pbcore/")
3851 (synopsis "Library for reading and writing PacBio data files")
3852 (description
3853 "The pbcore package provides Python APIs for interacting with PacBio data
3854files and writing bioinformatics applications.")
3855 (license license:bsd-3)))
3856
c61fe02c
RW		 3857(define-public python2-warpedlmm
3858 (package
3859 (name "python2-warpedlmm")
3860 (version "0.21")
3861 (source
3862 (origin
3863 (method url-fetch)
3864 (uri (string-append
3865 "https://pypi.python.org/packages/source/W/WarpedLMM/WarpedLMM-"
3866 version ".zip"))
3867 (sha256
3868 (base32
3869 "1agfz6zqa8nc6cw47yh0s3y14gkpa9wqazwcj7mwwj3ffnw39p3j"))))
3870 (build-system python-build-system)
3871 (arguments
54c85e12 3872 `(#:python ,python-2)) ; requires Python 2.7
c61fe02c
RW		 3873 (propagated-inputs
3874 `(("python-scipy" ,python2-scipy)
3875 ("python-numpy" ,python2-numpy)
3876 ("python-matplotlib" ,python2-matplotlib)
3877 ("python-fastlmm" ,python2-fastlmm)
3878 ("python-pandas" ,python2-pandas)
3879 ("python-pysnptools" ,python2-pysnptools)))
3880 (native-inputs
f3b98f4f 3881 `(("python-mock" ,python2-mock)
c61fe02c
RW		 3882 ("python-nose" ,python2-nose)
3883 ("unzip" ,unzip)))
3884 (home-page "https://github.com/PMBio/warpedLMM")
3885 (synopsis "Implementation of warped linear mixed models")
3886 (description
3887 "WarpedLMM is a Python implementation of the warped linear mixed model,
3888which automatically learns an optimal warping function (or transformation) for
3889the phenotype as it models the data.")
3890 (license license:asl2.0)))
3891
2c16316e 3892(define-public pbtranscript-tofu
698bd297 3893 (let ((commit "8f5467fe6a4472bcfb4226c8720993c8507adfe4"))
2c16316e
RW		 3894 (package
3895 (name "pbtranscript-tofu")
698bd297 3896 (version (string-append "2.2.3." (string-take commit 7)))
2c16316e
RW		 3897 (source (origin
3898 (method git-fetch)
3899 (uri (git-reference
3900 (url "https://github.com/PacificBiosciences/cDNA_primer.git")
3901 (commit commit)))
9a067efd 3902 (file-name (string-append name "-" version "-checkout"))
2c16316e
RW		 3903 (sha256
3904 (base32
9a067efd
RW		 3905 "1lgnpi35ihay42qx0b6yl3kkgra723i413j33kvs0kvs61h82w0f"))
3906 (modules '((guix build utils)))
3907 (snippet
3908 '(begin
3909 ;; remove bundled Cython sources
3910 (delete-file "pbtranscript-tofu/pbtranscript/Cython-0.20.1.tar.gz")
3911 #t))))
2c16316e
RW		 3912 (build-system python-build-system)
3913 (arguments
3914 `(#:python ,python-2
cdc2bb50
MB		 3915 ;; FIXME: Tests fail with "No such file or directory:
3916 ;; pbtools/pbtranscript/modified_bx_intervals/intersection_unique.so"
3917 #:tests? #f
2c16316e 3918 #:phases
9a067efd
RW		 3919 (modify-phases %standard-phases
3920 (add-after 'unpack 'enter-directory
3921 (lambda _
3922 (chdir "pbtranscript-tofu/pbtranscript/")
3923 #t))
3924 ;; With setuptools version 18.0 and later this setup.py hack causes
3925 ;; a build error, so we disable it.
3926 (add-after 'enter-directory 'patch-setuppy
3927 (lambda _
3928 (substitute* "setup.py"
3929 (("if 'setuptools.extension' in sys.modules:")
3930 "if False:"))
3931 #t)))))
2c16316e 3932 (inputs
9a067efd 3933 `(("python-numpy" ,python2-numpy)
2c16316e 3934 ("python-bx-python" ,python2-bx-python)
c5372108
RW		 3935 ("python-networkx" ,python2-networkx)
3936 ("python-scipy" ,python2-scipy)
9a067efd
RW		 3937 ("python-pbcore" ,python2-pbcore)
3938 ("python-h5py" ,python2-h5py)))
2c16316e 3939 (native-inputs
9a067efd 3940 `(("python-cython" ,python2-cython)
f3b98f4f 3941 ("python-nose" ,python2-nose)))
2c16316e
RW		 3942 (home-page "https://github.com/PacificBiosciences/cDNA_primer")
3943 (synopsis "Analyze transcriptome data generated with the Iso-Seq protocol")
3944 (description
3945 "pbtranscript-tofu contains scripts to analyze transcriptome data
3946generated using the PacBio Iso-Seq protocol.")
3947 (license license:bsd-3))))
3948
024130d2
BW		 3949(define-public prank
3950 (package
3951 (name "prank")
3952 (version "150803")
3953 (source (origin
3954 (method url-fetch)
3955 (uri (string-append
3956 "http://wasabiapp.org/download/prank/prank.source."
3957 version ".tgz"))
3958 (sha256
3959 (base32
3960 "0am4z94fs3w2n5xpfls9zda61vq7qqz4q2i7b9hlsxz5q4j3kfm4"))))
3961 (build-system gnu-build-system)
3962 (arguments
3963 `(#:phases
3964 (modify-phases %standard-phases
3965 (add-after 'unpack 'enter-src-dir
3966 (lambda _
3967 (chdir "src")
3968 #t))
62d00095
EF		 3969 (add-after 'unpack 'remove-m64-flag
3970 ;; Prank will build with the correct 'bit-ness' without this flag
3971 ;; and this allows building on 32-bit machines.
3972 (lambda _ (substitute* "src/Makefile"
3973 (("-m64") ""))
3974 #t))
024130d2
BW		 3975 (delete 'configure)
3976 (replace 'install
3977 (lambda* (#:key outputs #:allow-other-keys)
3978 (let* ((out (assoc-ref outputs "out"))
3979 (bin (string-append out "/bin"))
3980 (man (string-append out "/share/man/man1"))
3981 (path (string-append
3982 (assoc-ref %build-inputs "mafft") "/bin:"
3983 (assoc-ref %build-inputs "exonerate") "/bin:"
3984 (assoc-ref %build-inputs "bppsuite") "/bin")))
3985 (install-file "prank" bin)
3986 (wrap-program (string-append bin "/prank")
3987 `("PATH" ":" prefix (,path)))
3988 (install-file "prank.1" man))
3989 #t)))))
3990 (inputs
3991 `(("mafft" ,mafft)
3992 ("exonerate" ,exonerate)
3993 ("bppsuite" ,bppsuite)))
3994 (home-page "http://wasabiapp.org/software/prank/")
3995 (synopsis "Probabilistic multiple sequence alignment program")
3996 (description
3997 "PRANK is a probabilistic multiple sequence alignment program for DNA,
3998codon and amino-acid sequences. It is based on a novel algorithm that treats
3999insertions correctly and avoids over-estimation of the number of deletion
4000events. In addition, PRANK borrows ideas from maximum likelihood methods used
4001in phylogenetics and correctly takes into account the evolutionary distances
4002between sequences. Lastly, PRANK allows for defining a potential structure
4003for sequences to be aligned and then, simultaneously with the alignment,
4004predicts the locations of structural units in the sequences.")
4005 (license license:gpl2+)))
4006
31a9d653
BW		 4007(define-public proteinortho
4008 (package
4009 (name "proteinortho")
8c864901 4010 (version "5.16b")
31a9d653
BW		 4011 (source
4012 (origin
4013 (method url-fetch)
4014 (uri
4015 (string-append
4016 "http://www.bioinf.uni-leipzig.de/Software/proteinortho/proteinortho_v"
4017 version "_src.tar.gz"))
4018 (sha256
4019 (base32
8c864901 4020 "1wl0dawpssqwfjvr651r4wlww8hhjin8nba6xh71ks7sbypx886j"))))
31a9d653
BW		 4021 (build-system gnu-build-system)
4022 (arguments
4023 `(#:test-target "test"
4024 #:phases
4025 (modify-phases %standard-phases
4026 (replace 'configure
4027 ;; There is no configure script, so we modify the Makefile directly.
4028 (lambda* (#:key outputs #:allow-other-keys)
4029 (substitute* "Makefile"
4030 (("INSTALLDIR=.*")
4031 (string-append
4032 "INSTALLDIR=" (assoc-ref outputs "out") "/bin\n")))
4033 #t))
4034 (add-before 'install 'make-install-directory
4035 ;; The install directory is not created during 'make install'.
4036 (lambda* (#:key outputs #:allow-other-keys)
4037 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
4038 #t))
4039 (add-after 'install 'wrap-programs
4040 (lambda* (#:key inputs outputs #:allow-other-keys)
4041 (let* ((path (getenv "PATH"))
4042 (out (assoc-ref outputs "out"))
4043 (binary (string-append out "/bin/proteinortho5.pl")))
4044 (wrap-program binary `("PATH" ":" prefix (,path))))
4045 #t)))))
4046 (inputs
4047 `(("perl" ,perl)
4048 ("python" ,python-2)
4049 ("blast+" ,blast+)))
4050 (home-page "http://www.bioinf.uni-leipzig.de/Software/proteinortho")
4051 (synopsis "Detect orthologous genes across species")
4052 (description
4053 "Proteinortho is a tool to detect orthologous genes across different
4054species. For doing so, it compares similarities of given gene sequences and
4055clusters them to find significant groups. The algorithm was designed to handle
4056large-scale data and can be applied to hundreds of species at once.")
4057 (license license:gpl2+)))
4058
846e3409
RW		 4059(define-public pyicoteo
4060 (package
4061 (name "pyicoteo")
4062 (version "2.0.7")
4063 (source
4064 (origin
4065 (method url-fetch)
4066 (uri (string-append "https://bitbucket.org/regulatorygenomicsupf/"
4067 "pyicoteo/get/v" version ".tar.bz2"))
4068 (file-name (string-append name "-" version ".tar.bz2"))
4069 (sha256
4070 (base32
4071 "0d6087f29xp8wxwlj111c3sylli98n0l8ry58c51ixzq0zfm50wa"))))
4072 (build-system python-build-system)
4073 (arguments
4074 `(#:python ,python-2 ; does not work with Python 3
4075 #:tests? #f)) ; there are no tests
4076 (inputs
4077 `(("python2-matplotlib" ,python2-matplotlib)))
4078 (home-page "https://bitbucket.org/regulatorygenomicsupf/pyicoteo")
4079 (synopsis "Analyze high-throughput genetic sequencing data")
4080 (description
4081 "Pyicoteo is a suite of tools for the analysis of high-throughput genetic
4082sequencing data. It works with genomic coordinates. There are currently six
4083different command-line tools:
4084
4085@enumerate
4086@item pyicoregion: for generating exploratory regions automatically;
4087@item pyicoenrich: for differential enrichment between two conditions;
4088@item pyicoclip: for calling CLIP-Seq peaks without a control;
4089@item pyicos: for genomic coordinates manipulation;
4090@item pyicoller: for peak calling on punctuated ChIP-Seq;
4091@item pyicount: to count how many reads from N experiment files overlap in a
4092 region file;
4093@item pyicotrocol: to combine operations from pyicoteo.
4094@end enumerate\n")
4095 (license license:gpl3+)))
4096
af860475
BW		 4097(define-public prodigal
4098 (package
4099 (name "prodigal")
e70f7a23 4100 (version "2.6.3")
af860475
BW		 4101 (source (origin
4102 (method url-fetch)
4103 (uri (string-append
4104 "https://github.com/hyattpd/Prodigal/archive/v"
4105 version ".tar.gz"))
4106 (file-name (string-append name "-" version ".tar.gz"))
4107 (sha256
4108 (base32
e70f7a23 4109 "17srxkqd3jc77xk15pfbgg1a9xahqg7337w95mrsia7mpza4l2c9"))))
af860475
BW		 4110 (build-system gnu-build-system)
4111 (arguments
4112 `(#:tests? #f ;no check target
4113 #:make-flags (list (string-append "INSTALLDIR="
4114 (assoc-ref %outputs "out")
4115 "/bin"))
4116 #:phases
4117 (modify-phases %standard-phases
4118 (delete 'configure))))
4119 (home-page "http://prodigal.ornl.gov")
4120 (synopsis "Protein-coding gene prediction for Archaea and Bacteria")
4121 (description
4122 "Prodigal runs smoothly on finished genomes, draft genomes, and
4123metagenomes, providing gene predictions in GFF3, Genbank, or Sequin table
4124format. It runs quickly, in an unsupervised fashion, handles gaps, handles
4125partial genes, and identifies translation initiation sites.")
4126 (license license:gpl3+)))
608dd932 4127
ceb62d54
BW		 4128(define-public roary
4129 (package
4130 (name "roary")
da4a707c 4131 (version "3.11.0")
ceb62d54
BW		 4132 (source
4133 (origin
4134 (method url-fetch)
4135 (uri (string-append
4136 "mirror://cpan/authors/id/A/AJ/AJPAGE/Bio-Roary-"
4137 version ".tar.gz"))
4138 (sha256
4139 (base32
da4a707c 4140 "10lw78x1xzvn7xzvnmh4bm3cak3ah5cssapl0yidvhaj1f44h29i"))))
ceb62d54
BW		 4141 (build-system perl-build-system)
4142 (arguments
4143 `(#:phases
4144 (modify-phases %standard-phases
4145 (delete 'configure)
4146 (delete 'build)
4147 (replace 'check
4148 (lambda _
4149 ;; The tests are not run by default, so we run each test file
4150 ;; directly.
4151 (setenv "PATH" (string-append (getcwd) "/bin" ":"
4152 (getenv "PATH")))
4153 (setenv "PERL5LIB" (string-append (getcwd) "/lib" ":"
4154 (getenv "PERL5LIB")))
4155 (zero? (length (filter (lambda (file)
4156 (display file)(display "\n")
4157 (not (zero? (system* "perl" file))))
4158 (find-files "t" ".*\\.t$"))))))
4159 (replace 'install
4160 ;; There is no 'install' target in the Makefile.
4161 (lambda* (#:key outputs #:allow-other-keys)
4162 (let* ((out (assoc-ref outputs "out"))
4163 (bin (string-append out "/bin"))
4164 (perl (string-append out "/lib/perl5/site_perl"))
4165 (roary-plots "contrib/roary_plots"))
4166 (mkdir-p bin)
4167 (mkdir-p perl)
4168 (copy-recursively "bin" bin)
4169 (copy-recursively "lib" perl)
4170 #t)))
4171 (add-after 'install 'wrap-programs
4172 (lambda* (#:key inputs outputs #:allow-other-keys)
4173 (let* ((out (assoc-ref outputs "out"))
4174 (perl5lib (getenv "PERL5LIB"))
4175 (path (getenv "PATH")))
4176 (for-each (lambda (prog)
4177 (let ((binary (string-append out "/" prog)))
4178 (wrap-program binary
4179 `("PERL5LIB" ":" prefix
4180 (,(string-append perl5lib ":" out
4181 "/lib/perl5/site_perl"))))
4182 (wrap-program binary
4183 `("PATH" ":" prefix
4184 (,(string-append path ":" out "/bin"))))))
4185 (find-files "bin" ".*[^R]$"))
4186 (let ((file
4187 (string-append out "/bin/roary-create_pan_genome_plots.R"))
4188 (r-site-lib (getenv "R_LIBS_SITE"))
4189 (coreutils-path
4190 (string-append (assoc-ref inputs "coreutils") "/bin")))
4191 (wrap-program file
4192 `("R_LIBS_SITE" ":" prefix
4193 (,(string-append r-site-lib ":" out "/site-library/"))))
4194 (wrap-program file
4195 `("PATH" ":" prefix
4196 (,(string-append coreutils-path ":" out "/bin"))))))
4197 #t)))))
4198 (native-inputs
4199 `(("perl-env-path" ,perl-env-path)
4200 ("perl-test-files" ,perl-test-files)
4201 ("perl-test-most" ,perl-test-most)
4202 ("perl-test-output" ,perl-test-output)))
4203 (inputs
4204 `(("perl-array-utils" ,perl-array-utils)
4205 ("bioperl" ,bioperl-minimal)
da4a707c 4206 ("perl-digest-md5-file" ,perl-digest-md5-file)
ceb62d54
BW		 4207 ("perl-exception-class" ,perl-exception-class)
4208 ("perl-file-find-rule" ,perl-file-find-rule)
4209 ("perl-file-grep" ,perl-file-grep)
4210 ("perl-file-slurper" ,perl-file-slurper)
4211 ("perl-file-which" ,perl-file-which)
4212 ("perl-graph" ,perl-graph)
4213 ("perl-graph-readwrite" ,perl-graph-readwrite)
4214 ("perl-log-log4perl" ,perl-log-log4perl)
4215 ("perl-moose" ,perl-moose)
4216 ("perl-perlio-utf8_strict" ,perl-perlio-utf8_strict)
4217 ("perl-text-csv" ,perl-text-csv)
4218 ("bedtools" ,bedtools)
4219 ("cd-hit" ,cd-hit)
4220 ("blast+" ,blast+)
4221 ("mcl" ,mcl)
4222 ("parallel" ,parallel)
4223 ("prank" ,prank)
4224 ("mafft" ,mafft)
4225 ("fasttree" ,fasttree)
4226 ("grep" ,grep)
4227 ("sed" ,sed)
4228 ("gawk" ,gawk)
2d7c4ae3 4229 ("r-minimal" ,r-minimal)
ceb62d54
BW		 4230 ("r-ggplot2" ,r-ggplot2)
4231 ("coreutils" ,coreutils)))
4232 (home-page "http://sanger-pathogens.github.io/Roary")
4233 (synopsis "High speed stand-alone pan genome pipeline")
4234 (description
4235 "Roary is a high speed stand alone pan genome pipeline, which takes
4236annotated assemblies in GFF3 format (produced by the Prokka program) and
4237calculates the pan genome. Using a standard desktop PC, it can analyse
4238datasets with thousands of samples, without compromising the quality of the
4239results. 128 samples can be analysed in under 1 hour using 1 GB of RAM and a
4240single processor. Roary is not intended for metagenomics or for comparing
4241extremely diverse sets of genomes.")
4242 (license license:gpl3)))
4243
608dd932
BW		 4244(define-public raxml
4245 (package
4246 (name "raxml")
e9e0fab0 4247 (version "8.2.10")
608dd932
BW		 4248 (source
4249 (origin
4250 (method url-fetch)
4251 (uri
4252 (string-append
4253 "https://github.com/stamatak/standard-RAxML/archive/v"
4254 version ".tar.gz"))
4255 (file-name (string-append name "-" version ".tar.gz"))
4256 (sha256
4257 (base32
e9e0fab0 4258 "13s7aspfdcfr6asynwdg1x6vznys6pzap5f8wsffbnnwpkkg9ya8"))))
608dd932
BW		 4259 (build-system gnu-build-system)
4260 (arguments
4261 `(#:tests? #f ; There are no tests.
4262 ;; Use 'standard' Makefile rather than SSE or AVX ones.
4263 #:make-flags (list "-f" "Makefile.HYBRID.gcc")
4264 #:phases
4265 (modify-phases %standard-phases
4266 (delete 'configure)
4267 (replace 'install
4268 (lambda* (#:key outputs #:allow-other-keys)
4269 (let* ((out (assoc-ref outputs "out"))
4270 (bin (string-append out "/bin"))
4271 (executable "raxmlHPC-HYBRID"))
4272 (install-file executable bin)
4273 (symlink (string-append bin "/" executable) "raxml"))
4274 #t)))))
4275 (inputs
4276 `(("openmpi" ,openmpi)))
4277 (home-page "http://sco.h-its.org/exelixis/web/software/raxml/index.html")
4278 (synopsis "Randomized Axelerated Maximum Likelihood phylogenetic trees")
4279 (description
4280 "RAxML is a tool for phylogenetic analysis and post-analysis of large
4281phylogenies.")
2805f9b2
EF		 4282 ;; The source includes x86 specific code
4283 (supported-systems '("x86_64-linux" "i686-linux"))
608dd932 4284 (license license:gpl2+)))
af860475 4285
66e3eff1
RW		 4286(define-public rsem
4287 (package
4288 (name "rsem")
4289 (version "1.2.20")
4290 (source
4291 (origin
4292 (method url-fetch)
4293 (uri
4294 (string-append "http://deweylab.biostat.wisc.edu/rsem/src/rsem-"
4295 version ".tar.gz"))
4296 (sha256
4297 (base32 "0nzdc0j0hjllhsd5f2xli95dafm3nawskigs140xzvjk67xh0r9q"))
fc1adab1 4298 (patches (search-patches "rsem-makefile.patch"))
66e3eff1
RW		 4299 (modules '((guix build utils)))
4300 (snippet
4301 '(begin
4302 ;; remove bundled copy of boost
4303 (delete-file-recursively "boost")
4304 #t))))
4305 (build-system gnu-build-system)
4306 (arguments
4307 `(#:tests? #f ;no "check" target
4308 #:phases
4309 (modify-phases %standard-phases
4310 ;; No "configure" script.
4311 ;; Do not build bundled samtools library.
4312 (replace 'configure
4313 (lambda _
4314 (substitute* "Makefile"
4315 (("^all : sam/libbam.a") "all : "))
4316 #t))
4317 (replace 'install
4318 (lambda* (#:key outputs #:allow-other-keys)
4319 (let* ((out (string-append (assoc-ref outputs "out")))
4320 (bin (string-append out "/bin/"))
4321 (perl (string-append out "/lib/perl5/site_perl")))
4322 (mkdir-p bin)
4323 (mkdir-p perl)
4324 (for-each (lambda (file)
f3860753 4325 (install-file file bin))
66e3eff1 4326 (find-files "." "rsem-.*"))
f3860753 4327 (install-file "rsem_perl_utils.pm" perl))
66e3eff1
RW		 4328 #t))
4329 (add-after
4330 'install 'wrap-program
4331 (lambda* (#:key outputs #:allow-other-keys)
4332 (let ((out (assoc-ref outputs "out")))
4333 (for-each (lambda (prog)
4334 (wrap-program (string-append out "/bin/" prog)
4335 `("PERL5LIB" ":" prefix
4336 (,(string-append out "/lib/perl5/site_perl")))))
4337 '("rsem-plot-transcript-wiggles"
4338 "rsem-calculate-expression"
4339 "rsem-generate-ngvector"
4340 "rsem-run-ebseq"
4341 "rsem-prepare-reference")))
4342 #t)))))
4343 (inputs
4344 `(("boost" ,boost)
4345 ("ncurses" ,ncurses)
2d7c4ae3 4346 ("r-minimal" ,r-minimal)
66e3eff1
RW		 4347 ("perl" ,perl)
4348 ("samtools" ,samtools-0.1)
4349 ("zlib" ,zlib)))
4350 (home-page "http://deweylab.biostat.wisc.edu/rsem/")
4351 (synopsis "Estimate gene expression levels from RNA-Seq data")
4352 (description
4353 "RSEM is a software package for estimating gene and isoform expression
4354levels from RNA-Seq data. The RSEM package provides a user-friendly
4355interface, supports threads for parallel computation of the EM algorithm,
4356single-end and paired-end read data, quality scores, variable-length reads and
4357RSPD estimation. In addition, it provides posterior mean and 95% credibility
4358interval estimates for expression levels. For visualization, it can generate
4359BAM and Wiggle files in both transcript-coordinate and genomic-coordinate.")
4360 (license license:gpl3+)))
4361
8622a072
RW		 4362(define-public rseqc
4363 (package
4364 (name "rseqc")
4365 (version "2.6.1")
4366 (source
4367 (origin
4368 (method url-fetch)
4369 (uri
4370 (string-append "mirror://sourceforge/rseqc/"
de67e922 4371 "RSeQC-" version ".tar.gz"))
8622a072 4372 (sha256
8214b7fb 4373 (base32 "15ly0254yi032qzkdplg00q144qfdsd986gh62829rl5bkxhj330"))
8622a072
RW		 4374 (modules '((guix build utils)))
4375 (snippet
4376 '(begin
4377 ;; remove bundled copy of pysam
4378 (delete-file-recursively "lib/pysam")
4379 (substitute* "setup.py"
4380 ;; remove dependency on outdated "distribute" module
4381 (("^from distribute_setup import use_setuptools") "")
4382 (("^use_setuptools\\(\\)") "")
4383 ;; do not use bundled copy of pysam
4384 (("^have_pysam = False") "have_pysam = True"))))))
4385 (build-system python-build-system)
4386 (arguments `(#:python ,python-2))
4387 (inputs
4388 `(("python-cython" ,python2-cython)
4389 ("python-pysam" ,python2-pysam)
4390 ("python-numpy" ,python2-numpy)
8622a072
RW		 4391 ("zlib" ,zlib)))
4392 (native-inputs
4393 `(("python-nose" ,python2-nose)))
4394 (home-page "http://rseqc.sourceforge.net/")
4395 (synopsis "RNA-seq quality control package")
4396 (description
4397 "RSeQC provides a number of modules that can comprehensively evaluate
4398high throughput sequence data, especially RNA-seq data. Some basic modules
4399inspect sequence quality, nucleotide composition bias, PCR bias and GC bias,
4400while RNA-seq specific modules evaluate sequencing saturation, mapped reads
4401distribution, coverage uniformity, strand specificity, etc.")
4402 (license license:gpl3+)))
4403
ec946638
RW		 4404(define-public seek
4405 ;; There are no release tarballs. According to the installation
4406 ;; instructions at http://seek.princeton.edu/installation.jsp, the latest
4407 ;; stable release is identified by this changeset ID.
4408 (let ((changeset "2329130")
4409 (revision "1"))
4410 (package
4411 (name "seek")
4412 (version (string-append "0-" revision "." changeset))
4413 (source (origin
4414 (method hg-fetch)
4415 (uri (hg-reference
4416 (url "https://bitbucket.org/libsleipnir/sleipnir")
4417 (changeset changeset)))
4418 (sha256
4419 (base32
4420 "0qrvilwh18dpbhkf92qvxbmay0j75ra3jg2wrhz67gf538zzphsx"))))
4421 (build-system gnu-build-system)
4422 (arguments
4423 `(#:modules ((srfi srfi-1)
4424 (guix build gnu-build-system)
4425 (guix build utils))
4426 #:phases
4427 (let ((dirs '("SeekMiner"
4428 "SeekEvaluator"
4429 "SeekPrep"
4430 "Distancer"
4431 "Data2DB"
4432 "PCL2Bin")))
4433 (modify-phases %standard-phases
4561498e 4434 (add-before 'configure 'bootstrap
ec946638
RW		 4435 (lambda _
4436 (zero? (system* "bash" "gen_auto"))))
4437 (add-after 'build 'build-additional-tools
4438 (lambda* (#:key make-flags #:allow-other-keys)
4439 (every (lambda (dir)
4440 (with-directory-excursion (string-append "tools/" dir)
4441 (zero? (apply system* "make" make-flags))))
4442 dirs)))
4443 (add-after 'install 'install-additional-tools
4444 (lambda* (#:key make-flags #:allow-other-keys)
4445 (fold (lambda (dir result)
4446 (with-directory-excursion (string-append "tools/" dir)
4447 (and result
4448 (zero? (apply system*
4449 `("make" ,@make-flags "install"))))))
4450 #t dirs)))))))
4451 (inputs
4452 `(("gsl" ,gsl)
4453 ("boost" ,boost)
4454 ("libsvm" ,libsvm)
4455 ("readline" ,readline)
4456 ("gengetopt" ,gengetopt)
4457 ("log4cpp" ,log4cpp)))
4458 (native-inputs
4459 `(("autoconf" ,autoconf)
4460 ("automake" ,automake)
4461 ("perl" ,perl)))
4462 (home-page "http://seek.princeton.edu")
4463 (synopsis "Gene co-expression search engine")
4464 (description
4465 "SEEK is a computational gene co-expression search engine. SEEK provides
4466biologists with a way to navigate the massive human expression compendium that
4467now contains thousands of expression datasets. SEEK returns a robust ranking
4468of co-expressed genes in the biological area of interest defined by the user's
4469query genes. It also prioritizes thousands of expression datasets according
4470to the user's query of interest.")
4471 (license license:cc-by3.0))))
4472
4e10a221
RW		 4473(define-public samtools
4474 (package
4475 (name "samtools")
63fa0ffc 4476 (version "1.5")
4e10a221
RW		 4477 (source
4478 (origin
4479 (method url-fetch)
4480 (uri
de67e922 4481 (string-append "mirror://sourceforge/samtools/samtools/"
4e10a221
RW		 4482 version "/samtools-" version ".tar.bz2"))
4483 (sha256
4484 (base32
63fa0ffc 4485 "1xidmv0jmfy7l0kb32hdnlshcxgzi1hmygvig0cqrq1fhckdlhl5"))))
4e10a221
RW		 4486 (build-system gnu-build-system)
4487 (arguments
c4473411 4488 `(#:modules ((ice-9 ftw)
5bdda30b
RW		 4489 (ice-9 regex)
4490 (guix build gnu-build-system)
4491 (guix build utils))
c4473411 4492 #:make-flags (list (string-append "prefix=" (assoc-ref %outputs "out")))
4ab16440 4493 #:configure-flags (list "--with-ncurses" "--with-htslib=system")
4e10a221 4494 #:phases
c842d425
BW		 4495 (modify-phases %standard-phases
4496 (add-after 'unpack 'patch-tests
4497 (lambda _
4498 (substitute* "test/test.pl"
4499 ;; The test script calls out to /bin/bash
4500 (("/bin/bash") (which "bash")))
b2d748ed 4501 #t))
c842d425
BW		 4502 (add-after 'install 'install-library
4503 (lambda* (#:key outputs #:allow-other-keys)
4504 (let ((lib (string-append (assoc-ref outputs "out") "/lib")))
4505 (install-file "libbam.a" lib)
4506 #t)))
4507 (add-after 'install 'install-headers
4508 (lambda* (#:key outputs #:allow-other-keys)
4509 (let ((include (string-append (assoc-ref outputs "out")
4510 "/include/samtools/")))
4511 (for-each (lambda (file)
4512 (install-file file include))
4513 (scandir "." (lambda (name) (string-match "\\.h$" name))))
4514 #t))))))
4e10a221 4515 (native-inputs `(("pkg-config" ,pkg-config)))
4ab16440
BW		 4516 (inputs
4517 `(("htslib" ,htslib)
4518 ("ncurses" ,ncurses)
4519 ("perl" ,perl)
4520 ("python" ,python)
4521 ("zlib" ,zlib)))
4e10a221
RW		 4522 (home-page "http://samtools.sourceforge.net")
4523 (synopsis "Utilities to efficiently manipulate nucleotide sequence alignments")
4524 (description
4525 "Samtools implements various utilities for post-processing nucleotide
4526sequence alignments in the SAM, BAM, and CRAM formats, including indexing,
4527variant calling (in conjunction with bcftools), and a simple alignment
4528viewer.")
4529 (license license:expat)))
d3517eda 4530
0b84a0aa
RW		 4531(define-public samtools-0.1
4532 ;; This is the most recent version of the 0.1 line of samtools. The input
4533 ;; and output formats differ greatly from that used and produced by samtools
4534 ;; 1.x and is still used in many bioinformatics pipelines.
4535 (package (inherit samtools)
4536 (version "0.1.19")
4537 (source
4538 (origin
4539 (method url-fetch)
4540 (uri
de67e922 4541 (string-append "mirror://sourceforge/samtools/samtools/"
0b84a0aa
RW		 4542 version "/samtools-" version ".tar.bz2"))
4543 (sha256
4544 (base32 "1m33xsfwz0s8qi45lylagfllqg7fphf4dr0780rsvw75av9wk06h"))))
4545 (arguments
2309ed68
RW		 4546 `(#:tests? #f ;no "check" target
4547 ,@(substitute-keyword-arguments (package-arguments samtools)
4548 ((#:make-flags flags)
4549 `(cons "LIBCURSES=-lncurses" ,flags))
4550 ((#:phases phases)
4551 `(modify-phases ,phases
4552 (replace 'install
4553 (lambda* (#:key outputs #:allow-other-keys)
4554 (let ((bin (string-append
4555 (assoc-ref outputs "out") "/bin")))
4556 (mkdir-p bin)
f3860753 4557 (install-file "samtools" bin)
b2d748ed 4558 #t)))
2309ed68
RW		 4559 (delete 'patch-tests)
4560 (delete 'configure))))))))
0b84a0aa 4561
fe4c37c2 4562(define-public mosaik
698bd297 4563 (let ((commit "5c25216d3522d6a33e53875cd76a6d65001e4e67"))
fe4c37c2
RW		 4564 (package
4565 (name "mosaik")
4566 (version "2.2.30")
4567 (source (origin
4568 ;; There are no release tarballs nor tags.
4569 (method git-fetch)
4570 (uri (git-reference
4571 (url "https://github.com/wanpinglee/MOSAIK.git")
4572 (commit commit)))
4573 (file-name (string-append name "-" version))
4574 (sha256
4575 (base32
4576 "17gj3s07cm77r41z92awh0bim7w7q7fbn0sf5nkqmcm1vw052qgw"))))
4577 (build-system gnu-build-system)
4578 (arguments
4579 `(#:tests? #f ; no tests
4580 #:make-flags (list "CC=gcc")
4581 #:phases
4582 (modify-phases %standard-phases
4583 (replace 'configure
4584 (lambda _ (chdir "src") #t))
4585 (replace 'install
4586 (lambda* (#:key outputs #:allow-other-keys)
4587 (let ((bin (string-append (assoc-ref outputs "out")
4588 "/bin")))
4589 (mkdir-p bin)
4590 (copy-recursively "../bin" bin)
4591 #t))))))
4592 (inputs
4593 `(("perl" ,perl)
4594 ("zlib" ,zlib)))
029d9f77 4595 (supported-systems '("x86_64-linux"))
0c6c9c00 4596 (home-page "https://github.com/wanpinglee/MOSAIK")
fe4c37c2
RW		 4597 (synopsis "Map nucleotide sequence reads to reference genomes")
4598 (description
4599 "MOSAIK is a program for mapping second and third-generation sequencing
4600reads to a reference genome. MOSAIK can align reads generated by all the
4601major sequencing technologies, including Illumina, Applied Biosystems SOLiD,
4602Roche 454, Ion Torrent and Pacific BioSciences SMRT.")
4603 ;; MOSAIK is released under the GPLv2+ with the exception of third-party
4604 ;; code released into the public domain:
4605 ;; 1. fastlz by Ariya Hidayat - http://www.fastlz.org/
4606 ;; 2. MD5 implementation - RSA Data Security, RFC 1321
4607 (license (list license:gpl2+ license:public-domain)))))
4608
282c5087
RW		 4609(define-public ngs-sdk
4610 (package
4611 (name "ngs-sdk")
6c4ccf32 4612 (version "1.3.0")
282c5087
RW		 4613 (source
4614 (origin
4615 (method url-fetch)
4616 (uri
4617 (string-append "https://github.com/ncbi/ngs/archive/"
4618 version ".tar.gz"))
4619 (file-name (string-append name "-" version ".tar.gz"))
4620 (sha256
4621 (base32
6c4ccf32 4622 "1wiyf4c6nm2j87pv015cbi0qny5byf3pbvcw3likifz5dl56ag40"))))
282c5087
RW		 4623 (build-system gnu-build-system)
4624 (arguments
4625 `(#:parallel-build? #f ; not supported
4626 #:tests? #f ; no "check" target
4627 #:phases
dc1d3cde
KK		 4628 (modify-phases %standard-phases
4629 (replace 'configure
4630 (lambda* (#:key outputs #:allow-other-keys)
4631 (let ((out (assoc-ref outputs "out")))
4632 ;; Allow 'konfigure.perl' to find 'package.prl'.
4633 (setenv "PERL5LIB"
4634 (string-append ".:" (getenv "PERL5LIB")))
4635
4636 ;; The 'configure' script doesn't recognize things like
4637 ;; '--enable-fast-install'.
4638 (zero? (system* "./configure"
4639 (string-append "--build-prefix=" (getcwd) "/build")
4640 (string-append "--prefix=" out))))))
4641 (add-after 'unpack 'enter-dir
4642 (lambda _ (chdir "ngs-sdk") #t)))))
282c5087 4643 (native-inputs `(("perl" ,perl)))
a0dadf0c
AE		 4644 ;; According to the test
4645 ;; unless ($MARCH =~ /x86_64/i || $MARCH =~ /i?86/i)
4646 ;; in ngs-sdk/setup/konfigure.perl
ab29be81 4647 (supported-systems '("i686-linux" "x86_64-linux"))
282c5087
RW		 4648 (home-page "https://github.com/ncbi/ngs")
4649 (synopsis "API for accessing Next Generation Sequencing data")
4650 (description
4651 "NGS is a domain-specific API for accessing reads, alignments and pileups
4652produced from Next Generation Sequencing. The API itself is independent from
4653any particular back-end implementation, and supports use of multiple back-ends
4654simultaneously.")
4655 (license license:public-domain)))
4656
1ad15c16 4657(define-public java-ngs
2651a5e6 4658 (package (inherit ngs-sdk)
1ad15c16 4659 (name "java-ngs")
2651a5e6
RW		 4660 (arguments
4661 `(,@(substitute-keyword-arguments
4662 `(#:modules ((guix build gnu-build-system)
4663 (guix build utils)
4664 (srfi srfi-1)
4665 (srfi srfi-26))
4666 ,@(package-arguments ngs-sdk))
4667 ((#:phases phases)
614a8977
RW		 4668 `(modify-phases ,phases
4669 (replace 'enter-dir (lambda _ (chdir "ngs-java") #t)))))))
2651a5e6 4670 (inputs
d2540f80 4671 `(("jdk" ,icedtea "jdk")
2651a5e6
RW		 4672 ("ngs-sdk" ,ngs-sdk)))
4673 (synopsis "Java bindings for NGS SDK")))
4674
75dd2424
RW		 4675(define-public ncbi-vdb
4676 (package
4677 (name "ncbi-vdb")
5021f547 4678 (version "2.8.2")
75dd2424
RW		 4679 (source
4680 (origin
4681 (method url-fetch)
4682 (uri
4683 (string-append "https://github.com/ncbi/ncbi-vdb/archive/"
4684 version ".tar.gz"))
4685 (file-name (string-append name "-" version ".tar.gz"))
4686 (sha256
4687 (base32
5021f547 4688 "1acn4bv81mfl137qnbn9995mjjhwd36pm0b7qli1iw5skrxa9j8m"))))
75dd2424
RW		 4689 (build-system gnu-build-system)
4690 (arguments
4691 `(#:parallel-build? #f ; not supported
4692 #:tests? #f ; no "check" target
4693 #:phases
70111198 4694 (modify-phases %standard-phases
0691851a
BW		 4695 (add-before 'configure 'set-perl-search-path
4696 (lambda _
4697 ;; Work around "dotless @INC" build failure.
4698 (setenv "PERL5LIB"
4699 (string-append (getcwd) "/setup:"
4700 (getenv "PERL5LIB")))
4701 #t))
70111198
RW		 4702 (replace 'configure
4703 (lambda* (#:key inputs outputs #:allow-other-keys)
4704 (let ((out (assoc-ref outputs "out")))
4705 ;; Override include path for libmagic
4706 (substitute* "setup/package.prl"
4707 (("name => 'magic', Include => '/usr/include'")
4708 (string-append "name=> 'magic', Include => '"
4709 (assoc-ref inputs "libmagic")
4710 "/include" "'")))
75dd2424 4711
70111198
RW		 4712 ;; Install kdf5 library (needed by sra-tools)
4713 (substitute* "build/Makefile.install"
4714 (("LIBRARIES_TO_INSTALL =")
4715 "LIBRARIES_TO_INSTALL = kdf5.$(VERSION_LIBX) kdf5.$(VERSION_SHLX)"))
75dd2424 4716
70111198
RW		 4717 (substitute* "build/Makefile.env"
4718 (("CFLAGS =" prefix)
4719 (string-append prefix "-msse2 ")))
675d7ae2 4720
558e2307
RW		 4721 ;; Override search path for ngs-java
4722 (substitute* "setup/package.prl"
4723 (("/usr/local/ngs/ngs-java")
4724 (assoc-ref inputs "java-ngs")))
4725
70111198
RW		 4726 ;; The 'configure' script doesn't recognize things like
4727 ;; '--enable-fast-install'.
4728 (zero? (system*
4729 "./configure"
4730 (string-append "--build-prefix=" (getcwd) "/build")
4731 (string-append "--prefix=" (assoc-ref outputs "out"))
4732 (string-append "--debug")
4733 (string-append "--with-xml2-prefix="
4734 (assoc-ref inputs "libxml2"))
4735 (string-append "--with-ngs-sdk-prefix="
4736 (assoc-ref inputs "ngs-sdk"))
70111198
RW		 4737 (string-append "--with-hdf5-prefix="
4738 (assoc-ref inputs "hdf5")))))))
4739 (add-after 'install 'install-interfaces
4740 (lambda* (#:key outputs #:allow-other-keys)
4741 ;; Install interface libraries. On i686 the interface libraries
4742 ;; are installed to "linux/gcc/i386", so we need to use the Linux
4743 ;; architecture name ("i386") instead of the target system prefix
4744 ;; ("i686").
4745 (mkdir (string-append (assoc-ref outputs "out") "/ilib"))
4746 (copy-recursively (string-append "build/ncbi-vdb/linux/gcc/"
4747 ,(system->linux-architecture
4748 (or (%current-target-system)
4749 (%current-system)))
4750 "/rel/ilib")
4751 (string-append (assoc-ref outputs "out")
4752 "/ilib"))
4753 ;; Install interface headers
4754 (copy-recursively "interfaces"
4755 (string-append (assoc-ref outputs "out")
4756 "/include"))
d5e17162
RW		 4757 #t))
4758 ;; These files are needed by sra-tools.
4759 (add-after 'install 'install-configuration-files
4760 (lambda* (#:key outputs #:allow-other-keys)
4761 (let ((target (string-append (assoc-ref outputs "out") "/kfg")))
4762 (mkdir target)
4763 (install-file "libs/kfg/default.kfg" target)
4764 (install-file "libs/kfg/certs.kfg" target))
70111198 4765 #t)))))
75dd2424
RW		 4766 (inputs
4767 `(("libxml2" ,libxml2)
4768 ("ngs-sdk" ,ngs-sdk)
1ad15c16 4769 ("java-ngs" ,java-ngs)
75dd2424
RW		 4770 ("libmagic" ,file)
4771 ("hdf5" ,hdf5)))
4772 (native-inputs `(("perl" ,perl)))
675d7ae2
RW		 4773 ;; NCBI-VDB requires SSE capability.
4774 (supported-systems '("i686-linux" "x86_64-linux"))
75dd2424
RW		 4775 (home-page "https://github.com/ncbi/ncbi-vdb")
4776 (synopsis "Database engine for genetic information")
4777 (description
4778 "The NCBI-VDB library implements a highly compressed columnar data
4779warehousing engine that is most often used to store genetic information.
4780Databases are stored in a portable image within the file system, and can be
4781accessed/downloaded on demand across HTTP.")
4782 (license license:public-domain)))
4783
cc6ed477
RW		 4784(define-public plink
4785 (package
4786 (name "plink")
4787 (version "1.07")
4788 (source
4789 (origin
4790 (method url-fetch)
4791 (uri (string-append
4792 "http://pngu.mgh.harvard.edu/~purcell/plink/dist/plink-"
4793 version "-src.zip"))
4794 (sha256
4795 (base32 "0as8gxm4pjyc8dxmm1sl873rrd7wn5qs0l29nqfnl31x8i467xaa"))
0dbb7ac2
EF		 4796 (patches (search-patches "plink-1.07-unclobber-i.patch"
4797 "plink-endian-detection.patch"))))
cc6ed477
RW		 4798 (build-system gnu-build-system)
4799 (arguments
4800 '(#:tests? #f ;no "check" target
4801 #:make-flags (list (string-append "LIB_LAPACK="
4802 (assoc-ref %build-inputs "lapack")
4803 "/lib/liblapack.so")
4804 "WITH_LAPACK=1"
4805 "FORCE_DYNAMIC=1"
4806 ;; disable phoning home
4807 "WITH_WEBCHECK=")
4808 #:phases
4809 (modify-phases %standard-phases
4810 ;; no "configure" script
4811 (delete 'configure)
4812 (replace 'install
4813 (lambda* (#:key outputs #:allow-other-keys)
4814 (let ((bin (string-append (assoc-ref outputs "out")
4815 "/bin/")))
96c46210 4816 (install-file "plink" bin)
cc6ed477
RW		 4817 #t))))))
4818 (inputs
4819 `(("zlib" ,zlib)
4820 ("lapack" ,lapack)))
4821 (native-inputs
4822 `(("unzip" ,unzip)))
4823 (home-page "http://pngu.mgh.harvard.edu/~purcell/plink/")
4824 (synopsis "Whole genome association analysis toolset")
4825 (description
4826 "PLINK is a whole genome association analysis toolset, designed to
4827perform a range of basic, large-scale analyses in a computationally efficient
4828manner. The focus of PLINK is purely on analysis of genotype/phenotype data,
4829so there is no support for steps prior to this (e.g. study design and
4830planning, generating genotype or CNV calls from raw data). Through
4831integration with gPLINK and Haploview, there is some support for the
4832subsequent visualization, annotation and storage of results.")
4833 ;; Code is released under GPLv2, except for fisher.h, which is under
4834 ;; LGPLv2.1+
4835 (license (list license:gpl2 license:lgpl2.1+))))
4836
dad66da4
RW		 4837(define-public plink-ng
4838 (package (inherit plink)
4839 (name "plink-ng")
4840 (version "1.90b4")
4841 (source
4842 (origin
4843 (method url-fetch)
4844 (uri (string-append "https://github.com/chrchang/plink-ng/archive/v"
4845 version ".tar.gz"))
4846 (file-name (string-append name "-" version ".tar.gz"))
4847 (sha256
4848 (base32 "09ixrds009aczjswxr2alcb774mksq5g0v78dgjjn1h4dky0kf9a"))))
4849 (build-system gnu-build-system)
4850 (arguments
4851 '(#:tests? #f ;no "check" target
4852 #:make-flags (list "BLASFLAGS=-llapack -lopenblas"
4853 "CFLAGS=-Wall -O2 -DDYNAMIC_ZLIB=1"
4854 "ZLIB=-lz"
4855 "-f" "Makefile.std")
4856 #:phases
4857 (modify-phases %standard-phases
4858 (add-after 'unpack 'chdir
4859 (lambda _ (chdir "1.9") #t))
4860 (delete 'configure) ; no "configure" script
4861 (replace 'install
4862 (lambda* (#:key outputs #:allow-other-keys)
4863 (let ((bin (string-append (assoc-ref outputs "out")
4864 "/bin/")))
4865 (install-file "plink" bin)
4866 #t))))))
4867 (inputs
4868 `(("zlib" ,zlib)
4869 ("lapack" ,lapack)
4870 ("openblas" ,openblas)))
4871 (home-page "https://www.cog-genomics.org/plink/")
4872 (license license:gpl3+)))
4873
c6a24d6e
RW		 4874(define-public smithlab-cpp
4875 (let ((revision "1")
698bd297 4876 (commit "728a097bec88c6f4b8528b685932049e660eff2e"))
c6a24d6e
RW		 4877 (package
4878 (name "smithlab-cpp")
698bd297 4879 (version (string-append "0." revision "." (string-take commit 7)))
c6a24d6e
RW		 4880 (source (origin
4881 (method git-fetch)
4882 (uri (git-reference
4883 (url "https://github.com/smithlabcode/smithlab_cpp.git")
4884 (commit commit)))
4885 (file-name (string-append name "-" version "-checkout"))
4886 (sha256
4887 (base32
4888 "0d476lmj312xk77kr9fzrv7z1bv96yfyx0w7y62ycmnfbx32ll74"))))
4889 (build-system gnu-build-system)
4890 (arguments
4891 `(#:modules ((guix build gnu-build-system)
4892 (guix build utils)
4893 (srfi srfi-26))
4894 #:tests? #f ;no "check" target
4895 #:phases
4896 (modify-phases %standard-phases
4897 (add-after 'unpack 'use-samtools-headers
4898 (lambda _
4899 (substitute* '("SAM.cpp"
4900 "SAM.hpp")
4901 (("sam.h") "samtools/sam.h"))
4902 #t))
4903 (replace 'install
4904 (lambda* (#:key outputs #:allow-other-keys)
4905 (let* ((out (assoc-ref outputs "out"))
4906 (lib (string-append out "/lib"))
4907 (include (string-append out "/include/smithlab-cpp")))
4908 (mkdir-p lib)
4909 (mkdir-p include)
4910 (for-each (cut install-file <> lib)
4911 (find-files "." "\\.o$"))
4912 (for-each (cut install-file <> include)
4913 (find-files "." "\\.hpp$")))
4914 #t))
4915 (delete 'configure))))
4916 (inputs
4917 `(("samtools" ,samtools-0.1)
4918 ("zlib" ,zlib)))
4919 (home-page "https://github.com/smithlabcode/smithlab_cpp")
4920 (synopsis "C++ helper library for functions used in Smith lab projects")
4921 (description
4922 "Smithlab CPP is a C++ library that includes functions used in many of
4923the Smith lab bioinformatics projects, such as a wrapper around Samtools data
4924structures, classes for genomic regions, mapped sequencing reads, etc.")
4925 (license license:gpl3+))))
4926
56e373ef
RW		 4927(define-public preseq
4928 (package
4929 (name "preseq")
b49c5a58 4930 (version "2.0")
56e373ef
RW		 4931 (source (origin
4932 (method url-fetch)
b49c5a58
RW		 4933 (uri (string-append "https://github.com/smithlabcode/"
4934 "preseq/archive/v" version ".tar.gz"))
4935 (file-name (string-append name "-" version ".tar.gz"))
56e373ef 4936 (sha256
b49c5a58 4937 (base32 "08r684l50pnxjpvmhzjgqq56yv9rfw90k8vx0nsrnrzk8mf9hsdq"))
56e373ef
RW		 4938 (modules '((guix build utils)))
4939 (snippet
4940 ;; Remove bundled samtools.
b49c5a58 4941 '(delete-file-recursively "samtools"))))
56e373ef
RW		 4942 (build-system gnu-build-system)
4943 (arguments
4944 `(#:tests? #f ;no "check" target
4945 #:phases
4946 (modify-phases %standard-phases
56e373ef 4947 (delete 'configure))
b49c5a58
RW		 4948 #:make-flags
4949 (list (string-append "PREFIX="
4950 (assoc-ref %outputs "out"))
4951 (string-append "LIBBAM="
4952 (assoc-ref %build-inputs "samtools")
4953 "/lib/libbam.a")
4954 (string-append "SMITHLAB_CPP="
4955 (assoc-ref %build-inputs "smithlab-cpp")
4956 "/lib")
4957 "PROGS=preseq"
4958 "INCLUDEDIRS=$(SMITHLAB_CPP)/../include/smithlab-cpp $(SAMTOOLS_DIR)")))
56e373ef
RW		 4959 (inputs
4960 `(("gsl" ,gsl)
4961 ("samtools" ,samtools-0.1)
b49c5a58 4962 ("smithlab-cpp" ,smithlab-cpp)
56e373ef
RW		 4963 ("zlib" ,zlib)))
4964 (home-page "http://smithlabresearch.org/software/preseq/")
4965 (synopsis "Program for analyzing library complexity")
4966 (description
4967 "The preseq package is aimed at predicting and estimating the complexity
4968of a genomic sequencing library, equivalent to predicting and estimating the
4969number of redundant reads from a given sequencing depth and how many will be
4970expected from additional sequencing using an initial sequencing experiment.
4971The estimates can then be used to examine the utility of further sequencing,
4972optimize the sequencing depth, or to screen multiple libraries to avoid low
4973complexity samples.")
4974 (license license:gpl3+)))
4975
9ded1457
BW		 4976(define-public python-screed
4977 (package
4978 (name "python-screed")
4979 (version "0.9")
4980 (source
4981 (origin
4982 (method url-fetch)
4983 (uri (pypi-uri "screed" version))
4984 (sha256
4985 (base32
4986 "18czszp9fkx3j6jr7y5kp6dfialscgddk05mw1zkhh2zhn0jd8i0"))))
4987 (build-system python-build-system)
4988 (arguments
4989 `(#:phases
4990 (modify-phases %standard-phases
4991 (replace 'check
4992 (lambda _
4993 (setenv "PYTHONPATH"
4994 (string-append (getenv "PYTHONPATH") ":."))
4995 (zero? (system* "nosetests" "--attr" "!known_failing")))))))
4996 (native-inputs
4997 `(("python-nose" ,python-nose)))
4998 (inputs
4999 `(("python-bz2file" ,python-bz2file)))
7bf837fd 5000 (home-page "https://github.com/dib-lab/screed/")
9ded1457
BW		 5001 (synopsis "Short read sequence database utilities")
5002 (description "Screed parses FASTA and FASTQ files and generates databases.
5003Values such as sequence name, sequence description, sequence quality and the
5004sequence itself can be retrieved from these databases.")
5005 (license license:bsd-3)))
5006
5007(define-public python2-screed
5c31f4aa 5008 (package-with-python2 python-screed))
9ded1457 5009
51c64999
RW		 5010(define-public sra-tools
5011 (package
5012 (name "sra-tools")
79849358 5013 (version "2.8.2-1")
51c64999
RW		 5014 (source
5015 (origin
5016 (method url-fetch)
5017 (uri
5018 (string-append "https://github.com/ncbi/sra-tools/archive/"
5019 version ".tar.gz"))
5020 (file-name (string-append name "-" version ".tar.gz"))
5021 (sha256
5022 (base32
79849358 5023 "1camsijmvv2s45mb4iyf44ghl4gkd4rl0viphpcgl3ccchy32a0g"))))
51c64999
RW		 5024 (build-system gnu-build-system)
5025 (arguments
5026 `(#:parallel-build? #f ; not supported
5027 #:tests? #f ; no "check" target
2320e76b 5028 #:make-flags
79849358
RW		 5029 (list (string-append "DEFAULT_CRT="
5030 (assoc-ref %build-inputs "ncbi-vdb")
5031 "/kfg/certs.kfg")
5032 (string-append "DEFAULT_KFG="
5033 (assoc-ref %build-inputs "ncbi-vdb")
5034 "/kfg/default.kfg")
5035 (string-append "VDB_LIBDIR="
2320e76b
RW		 5036 (assoc-ref %build-inputs "ncbi-vdb")
5037 ,(if (string-prefix? "x86_64"
5038 (or (%current-target-system)
5039 (%current-system)))
5040 "/lib64"
5041 "/lib32")))
51c64999 5042 #:phases
beebe431 5043 (modify-phases %standard-phases
0691851a
BW		 5044 (add-before 'configure 'set-perl-search-path
5045 (lambda _
5046 ;; Work around "dotless @INC" build failure.
5047 (setenv "PERL5LIB"
5048 (string-append (getcwd) "/setup:"
5049 (getenv "PERL5LIB")))
5050 #t))
beebe431
RW		 5051 (replace 'configure
5052 (lambda* (#:key inputs outputs #:allow-other-keys)
5053 ;; The build system expects a directory containing the sources and
5054 ;; raw build output of ncbi-vdb, including files that are not
5055 ;; installed. Since we are building against an installed version of
5056 ;; ncbi-vdb, the following modifications are needed.
5057 (substitute* "setup/konfigure.perl"
5058 ;; Make the configure script look for the "ilib" directory of
5059 ;; "ncbi-vdb" without first checking for the existence of a
5060 ;; matching library in its "lib" directory.
5061 (("^ my \\$f = File::Spec->catdir\\(\\$libdir, \\$lib\\);")
5062 "my $f = File::Spec->catdir($ilibdir, $ilib);")
5063 ;; Look for interface libraries in ncbi-vdb's "ilib" directory.
5064 (("my \\$ilibdir = File::Spec->catdir\\(\\$builddir, 'ilib'\\);")
5065 "my $ilibdir = File::Spec->catdir($dir, 'ilib');"))
51c64999 5066
beebe431
RW		 5067 ;; Dynamic linking
5068 (substitute* "tools/copycat/Makefile"
5069 (("smagic-static") "lmagic"))
2320e76b 5070
beebe431
RW		 5071 ;; The 'configure' script doesn't recognize things like
5072 ;; '--enable-fast-install'.
5073 (zero? (system*
5074 "./configure"
5075 (string-append "--build-prefix=" (getcwd) "/build")
5076 (string-append "--prefix=" (assoc-ref outputs "out"))
5077 (string-append "--debug")
5078 (string-append "--with-fuse-prefix="
5079 (assoc-ref inputs "fuse"))
5080 (string-append "--with-magic-prefix="
5081 (assoc-ref inputs "libmagic"))
5082 ;; TODO: building with libxml2 fails with linker errors
5083 ;; (string-append "--with-xml2-prefix="
5084 ;; (assoc-ref inputs "libxml2"))
5085 (string-append "--with-ncbi-vdb-sources="
5086 (assoc-ref inputs "ncbi-vdb"))
5087 (string-append "--with-ncbi-vdb-build="
5088 (assoc-ref inputs "ncbi-vdb"))
5089 (string-append "--with-ngs-sdk-prefix="
5090 (assoc-ref inputs "ngs-sdk"))
5091 (string-append "--with-hdf5-prefix="
1108a920
RW		 5092 (assoc-ref inputs "hdf5"))))))
5093 ;; This version of sra-tools fails to build with glibc because of a
5094 ;; naming conflict. glibc-2.25/include/bits/mathcalls.h already
5095 ;; contains a definition of "canonicalize", so we rename it.
5096 ;;
5097 ;; See upstream bug report:
5098 ;; https://github.com/ncbi/sra-tools/issues/67
5099 (add-after 'unpack 'patch-away-glibc-conflict
5100 (lambda _
5101 (substitute* "tools/bam-loader/bam.c"
5102 (("canonicalize\\(" line)
5103 (string-append "sra_tools_" line)))
5104 #t)))))
51c64999
RW		 5105 (native-inputs `(("perl" ,perl)))
5106 (inputs
5107 `(("ngs-sdk" ,ngs-sdk)
5108 ("ncbi-vdb" ,ncbi-vdb)
5109 ("libmagic" ,file)
5110 ("fuse" ,fuse)
5111 ("hdf5" ,hdf5)
5112 ("zlib" ,zlib)))
5113 (home-page "http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software")
5114 (synopsis "Tools and libraries for reading and writing sequencing data")
5115 (description
5116 "The SRA Toolkit from NCBI is a collection of tools and libraries for
5117reading of sequencing files from the Sequence Read Archive (SRA) database and
5118writing files into the .sra format.")
5119 (license license:public-domain)))
5120
d3517eda
RW		 5121(define-public seqan
5122 (package
5123 (name "seqan")
5124 (version "1.4.2")
5125 (source (origin
5126 (method url-fetch)
5127 (uri (string-append "http://packages.seqan.de/seqan-library/"
5128 "seqan-library-" version ".tar.bz2"))
5129 (sha256
5130 (base32
5131 "05s3wrrwn50f81aklfm65i4a749zag1vr8z03k21xm0pdxy47yvp"))))
5132 ;; The documentation is 7.8MB and the includes are 3.6MB heavy, so it
5133 ;; makes sense to split the outputs.
5134 (outputs '("out" "doc"))
5135 (build-system trivial-build-system)
5136 (arguments
5137 `(#:modules ((guix build utils))
5138 #:builder
5139 (begin
5140 (use-modules (guix build utils))
5141 (let ((tar (assoc-ref %build-inputs "tar"))
5142 (bzip (assoc-ref %build-inputs "bzip2"))
5143 (out (assoc-ref %outputs "out"))
5144 (doc (assoc-ref %outputs "doc")))
5145 (setenv "PATH" (string-append tar "/bin:" bzip "/bin"))
5146 (system* "tar" "xvf" (assoc-ref %build-inputs "source"))
5147 (chdir (string-append "seqan-library-" ,version))
5148 (copy-recursively "include" (string-append out "/include"))
5149 (copy-recursively "share" (string-append doc "/share"))))))
5150 (native-inputs
5151 `(("source" ,source)
5152 ("tar" ,tar)
5153 ("bzip2" ,bzip2)))
5154 (home-page "http://www.seqan.de")
5155 (synopsis "Library for nucleotide sequence analysis")
5156 (description
5157 "SeqAn is a C++ library of efficient algorithms and data structures for
5158the analysis of sequences with the focus on biological data. It contains
5159algorithms and data structures for string representation and their
5160manipulation, online and indexed string search, efficient I/O of
5161bioinformatics file formats, sequence alignment, and more.")
5162 (license license:bsd-3)))
ce7155d5 5163
d708b7a9
BW		 5164(define-public seqmagick
5165 (package
5166 (name "seqmagick")
5167 (version "0.6.1")
5168 (source
5169 (origin
5170 (method url-fetch)
5171 (uri (string-append
5172 "https://pypi.python.org/packages/source/s/seqmagick/seqmagick-"
5173 version ".tar.gz"))
5174 (sha256
5175 (base32
5176 "0cgn477n74gsl4qdaakrrhi953kcsd4q3ivk2lr18x74s3g4ma1d"))))
5177 (build-system python-build-system)
5178 (arguments
5179 ;; python2 only, see https://github.com/fhcrc/seqmagick/issues/56
5180 `(#:python ,python-2
5181 #:phases
5182 (modify-phases %standard-phases
5183 ;; Current test in setup.py does not work as of 0.6.1,
5184 ;; so use nose to run tests instead for now. See
5185 ;; https://github.com/fhcrc/seqmagick/issues/55
5186 (replace 'check (lambda _ (zero? (system* "nosetests")))))))
5187 (inputs
7bba95b7
BW		 5188 ;; biopython-1.66 is required due to
5189 ;; https://github.com/fhcrc/seqmagick/issues/59
5190 ;; When that issue is resolved the 'python2-biopython-1.66' package
5191 ;; should be removed.
5192 `(("python-biopython" ,python2-biopython-1.66)))
d708b7a9 5193 (native-inputs
f3b98f4f 5194 `(("python-nose" ,python2-nose)))
7bf837fd 5195 (home-page "https://github.com/fhcrc/seqmagick")
d708b7a9
BW		 5196 (synopsis "Tools for converting and modifying sequence files")
5197 (description
5198 "Bioinformaticians often have to convert sequence files between formats
5199and do little manipulations on them, and it's not worth writing scripts for
5200that. Seqmagick is a utility to expose the file format conversion in
5201BioPython in a convenient way. Instead of having a big mess of scripts, there
5202is one that takes arguments.")
5203 (license license:gpl3)))
5204
66daf78c
BW		 5205(define-public seqtk
5206 (package
5207 (name "seqtk")
5208 (version "1.2")
5209 (source (origin
5210 (method url-fetch)
5211 (uri (string-append
5212 "https://github.com/lh3/seqtk/archive/v"
5213 version ".tar.gz"))
5214 (file-name (string-append name "-" version ".tar.gz"))
5215 (sha256
5216 (base32
5217 "0ywdyzpmfiz2wp6ampbzqg4y8bj450nfgqarpamg045b8mk32lxx"))
5218 (modules '((guix build utils)))
5219 (snippet
5220 '(begin
5221 ;; Remove extraneous header files, as is done in the seqtk
5222 ;; master branch.
5223 (for-each (lambda (file) (delete-file file))
5224 (list "ksort.h" "kstring.h" "kvec.h"))
5225 #t))))
5226 (build-system gnu-build-system)
5227 (arguments
5228 `(#:phases
5229 (modify-phases %standard-phases
5230 (delete 'configure)
5231 (replace 'check
5232 ;; There are no tests, so we just run a sanity check.
5233 (lambda _ (zero? (system* "./seqtk" "seq"))))
5234 (replace 'install
5235 (lambda* (#:key outputs #:allow-other-keys)
5236 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5237 (install-file "seqtk" bin)))))))
5238 (inputs
5239 `(("zlib" ,zlib)))
5240 (home-page "https://github.com/lh3/seqtk")
5241 (synopsis "Toolkit for processing biological sequences in FASTA/Q format")
5242 (description
5243 "Seqtk is a fast and lightweight tool for processing sequences in the
5244FASTA or FASTQ format. It parses both FASTA and FASTQ files which can be
5245optionally compressed by gzip.")
5246 (license license:expat)))
5247
5f7e17be
BW		 5248(define-public snap-aligner
5249 (package
5250 (name "snap-aligner")
5251 (version "1.0beta.18")
5252 (source (origin
5253 (method url-fetch)
5254 (uri (string-append
5255 "https://github.com/amplab/snap/archive/v"
5256 version ".tar.gz"))
5257 (file-name (string-append name "-" version ".tar.gz"))
5258 (sha256
5259 (base32
5260 "1vnsjwv007k1fl1q7d681kbwn6bc66cgw6h16hym6gvyy71qv2ly"))))
5261 (build-system gnu-build-system)
5262 (arguments
5263 '(#:phases
5264 (modify-phases %standard-phases
5265 (delete 'configure)
5266 (replace 'check (lambda _ (zero? (system* "./unit_tests"))))
5267 (replace 'install
5268 (lambda* (#:key outputs #:allow-other-keys)
5269 (let* ((out (assoc-ref outputs "out"))
5270 (bin (string-append out "/bin")))
5f7e17be
BW		 5271 (install-file "snap-aligner" bin)
5272 (install-file "SNAPCommand" bin)
5273 #t))))))
5274 (native-inputs
5275 `(("zlib" ,zlib)))
5276 (home-page "http://snap.cs.berkeley.edu/")
5277 (synopsis "Short read DNA sequence aligner")
5278 (description
5279 "SNAP is a fast and accurate aligner for short DNA reads. It is
5280optimized for modern read lengths of 100 bases or higher, and takes advantage
5281of these reads to align data quickly through a hash-based indexing scheme.")
3e6fdd5f
EF		 5282 ;; 32-bit systems are not supported by the unpatched code.
5283 ;; Following the bug reports https://github.com/amplab/snap/issues/68 and
5284 ;; https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=812378 we see that
5285 ;; systems without a lot of memory cannot make good use of this program.
5286 (supported-systems '("x86_64-linux"))
5f7e17be
BW		 5287 (license license:asl2.0)))
5288
bcadaf00
BW		 5289(define-public sortmerna
5290 (package
5291 (name "sortmerna")
849485f5 5292 (version "2.1b")
bcadaf00
BW		 5293 (source
5294 (origin
5295 (method url-fetch)
5296 (uri (string-append
5297 "https://github.com/biocore/sortmerna/archive/"
5298 version ".tar.gz"))
5299 (file-name (string-append name "-" version ".tar.gz"))
5300 (sha256
5301 (base32
849485f5 5302 "1ghaghvd82af9j5adavxh77g7hm247d1r69m3fbi6f1jdivj5ldk"))))
bcadaf00
BW		 5303 (build-system gnu-build-system)
5304 (outputs '("out" ;for binaries
5305 "db")) ;for sequence databases
5306 (arguments
5307 `(#:phases
5308 (modify-phases %standard-phases
5309 (replace 'install
5310 (lambda* (#:key outputs #:allow-other-keys)
5311 (let* ((out (assoc-ref outputs "out"))
5312 (bin (string-append out "/bin"))
5313 (db (assoc-ref outputs "db"))
5314 (share
5315 (string-append db "/share/sortmerna/rRNA_databases")))
5316 (install-file "sortmerna" bin)
5317 (install-file "indexdb_rna" bin)
5318 (for-each (lambda (file)
5319 (install-file file share))
5320 (find-files "rRNA_databases" ".*fasta"))
5321 #t))))))
849485f5
BW		 5322 (inputs
5323 `(("zlib" ,zlib)))
bcadaf00
BW		 5324 (home-page "http://bioinfo.lifl.fr/RNA/sortmerna")
5325 (synopsis "Biological sequence analysis tool for NGS reads")
5326 (description
5327 "SortMeRNA is a biological sequence analysis tool for filtering, mapping
5328and operational taxonomic unit (OTU) picking of next generation
5329sequencing (NGS) reads. The core algorithm is based on approximate seeds and
5330allows for fast and sensitive analyses of nucleotide sequences. The main
5331application of SortMeRNA is filtering rRNA from metatranscriptomic data.")
2324541b
EF		 5332 ;; The source includes x86 specific code
5333 (supported-systems '("x86_64-linux" "i686-linux"))
bcadaf00
BW		 5334 (license license:lgpl3)))
5335
ce7155d5
RW		 5336(define-public star
5337 (package
5338 (name "star")
939b6b11 5339 (version "2.5.3a")
ce7155d5
RW		 5340 (source (origin
5341 (method url-fetch)
3062d750
RW		 5342 (uri (string-append "https://github.com/alexdobin/STAR/archive/"
5343 version ".tar.gz"))
5344 (file-name (string-append name "-" version ".tar.gz"))
ce7155d5
RW		 5345 (sha256
5346 (base32
939b6b11 5347 "013wirlz8lllgjyagl48l75n1isxyabqb3sj7qlsl0x1rmvqw99a"))
ce7155d5
RW		 5348 (modules '((guix build utils)))
5349 (snippet
3062d750
RW		 5350 '(begin
5351 (substitute* "source/Makefile"
5352 (("/bin/rm") "rm"))
5353 ;; Remove pre-built binaries and bundled htslib sources.
5354 (delete-file-recursively "bin/MacOSX_x86_64")
5355 (delete-file-recursively "bin/Linux_x86_64")
c9e9154e 5356 (delete-file-recursively "bin/Linux_x86_64_static")
3062d750
RW		 5357 (delete-file-recursively "source/htslib")
5358 #t))))
ce7155d5
RW		 5359 (build-system gnu-build-system)
5360 (arguments
5361 '(#:tests? #f ;no check target
5362 #:make-flags '("STAR")
5363 #:phases
c0266e8d
RW		 5364 (modify-phases %standard-phases
5365 (add-after 'unpack 'enter-source-dir
5366 (lambda _ (chdir "source") #t))
3062d750
RW		 5367 (add-after 'enter-source-dir 'do-not-use-bundled-htslib
5368 (lambda _
5369 (substitute* "Makefile"
5370 (("(Depend.list: \\$\\(SOURCES\\) parametersDefault\\.xxd) htslib"
5371 _ prefix) prefix))
5372 (substitute* '("BAMfunctions.cpp"
5373 "signalFromBAM.h"
5374 "bam_cat.h"
5375 "bam_cat.c"
5376 "STAR.cpp"
5377 "bamRemoveDuplicates.cpp")
5378 (("#include \"htslib/([^\"]+\\.h)\"" _ header)
5379 (string-append "#include <" header ">")))
5380 (substitute* "IncludeDefine.h"
5381 (("\"htslib/(htslib/[^\"]+.h)\"" _ header)
5382 (string-append "<" header ">")))
5383 #t))
c0266e8d
RW		 5384 (replace 'install
5385 (lambda* (#:key outputs #:allow-other-keys)
5386 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5387 (install-file "STAR" bin))
5388 #t))
5389 (delete 'configure))))
ce7155d5 5390 (native-inputs
9fc513ad 5391 `(("xxd" ,xxd)))
ce7155d5 5392 (inputs
3062d750
RW		 5393 `(("htslib" ,htslib)
5394 ("zlib" ,zlib)))
ce7155d5
RW		 5395 (home-page "https://github.com/alexdobin/STAR")
5396 (synopsis "Universal RNA-seq aligner")
5397 (description
5398 "The Spliced Transcripts Alignment to a Reference (STAR) software is
5399based on a previously undescribed RNA-seq alignment algorithm that uses
5400sequential maximum mappable seed search in uncompressed suffix arrays followed
5401by seed clustering and stitching procedure. In addition to unbiased de novo
5402detection of canonical junctions, STAR can discover non-canonical splices and
5403chimeric (fusion) transcripts, and is also capable of mapping full-length RNA
5404sequences.")
c11f79a4
BW		 5405 ;; Only 64-bit systems are supported according to the README.
5406 (supported-systems '("x86_64-linux" "mips64el-linux"))
ce7155d5
RW		 5407 ;; STAR is licensed under GPLv3 or later; htslib is MIT-licensed.
5408 (license license:gpl3+)))
de07c0db 5409
dbf4ed7c
RW		 5410(define-public subread
5411 (package
5412 (name "subread")
c180533b 5413 (version "1.5.1")
dbf4ed7c
RW		 5414 (source (origin
5415 (method url-fetch)
de67e922
LF		 5416 (uri (string-append "mirror://sourceforge/subread/subread-"
5417 version "/subread-" version "-source.tar.gz"))
dbf4ed7c
RW		 5418 (sha256
5419 (base32
c180533b 5420 "0gn5zhbvllks0mmdg3qlmsbg91p2mpdc2wixwfqpi85yzfrh8hcy"))))
dbf4ed7c
RW		 5421 (build-system gnu-build-system)
5422 (arguments
5423 `(#:tests? #f ;no "check" target
104c1986
RW		 5424 ;; The CC and CCFLAGS variables are set to contain a lot of x86_64
5425 ;; optimizations by default, so we override these flags such that x86_64
5426 ;; flags are only added when the build target is an x86_64 system.
5427 #:make-flags
5428 (list (let ((system ,(or (%current-target-system)
5429 (%current-system)))
5430 (flags '("-ggdb" "-fomit-frame-pointer"
5431 "-ffast-math" "-funroll-loops"
5432 "-fmessage-length=0"
5433 "-O9" "-Wall" "-DMAKE_FOR_EXON"
5434 "-DMAKE_STANDALONE"
5435 "-DSUBREAD_VERSION=\\\"${SUBREAD_VERSION}\\\""))
5436 (flags64 '("-mmmx" "-msse" "-msse2" "-msse3")))
5437 (if (string-prefix? "x86_64" system)
5438 (string-append "CCFLAGS=" (string-join (append flags flags64)))
5439 (string-append "CCFLAGS=" (string-join flags))))
5440 "-f" "Makefile.Linux"
5441 "CC=gcc ${CCFLAGS}")
dbf4ed7c 5442 #:phases
dc1d3cde
KK		 5443 (modify-phases %standard-phases
5444 (add-after 'unpack 'enter-dir
5445 (lambda _ (chdir "src") #t))
5446 (replace 'install
5447 (lambda* (#:key outputs #:allow-other-keys)
5448 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5449 (mkdir-p bin)
5450 (copy-recursively "../bin" bin))))
dbf4ed7c 5451 ;; no "configure" script
dc1d3cde 5452 (delete 'configure))))
dbf4ed7c
RW		 5453 (inputs `(("zlib" ,zlib)))
5454 (home-page "http://bioinf.wehi.edu.au/subread-package/")
5455 (synopsis "Tool kit for processing next-gen sequencing data")
5456 (description
5457 "The subread package contains the following tools: subread aligner, a
5458general-purpose read aligner; subjunc aligner: detecting exon-exon junctions
5459and mapping RNA-seq reads; featureCounts: counting mapped reads for genomic
5460features; exactSNP: a SNP caller that discovers SNPs by testing signals
5461against local background noises.")
5462 (license license:gpl3+)))
5463
d15d981e
RW		 5464(define-public stringtie
5465 (package
5466 (name "stringtie")
5467 (version "1.2.1")
5468 (source (origin
5469 (method url-fetch)
5470 (uri (string-append "http://ccb.jhu.edu/software/stringtie/dl/"
5471 "stringtie-" version ".tar.gz"))
5472 (sha256
5473 (base32
5474 "1cqllsc1maq4kh92isi8yadgzbmnf042hlnalpk3y59aph1z3bfz"))
5475 (modules '((guix build utils)))
5476 (snippet
5477 '(begin
5478 (delete-file-recursively "samtools-0.1.18")
5479 #t))))
5480 (build-system gnu-build-system)
5481 (arguments
5482 `(#:tests? #f ;no test suite
5483 #:phases
5484 (modify-phases %standard-phases
5485 ;; no configure script
5486 (delete 'configure)
5487 (add-before 'build 'use-system-samtools
5488 (lambda _
5489 (substitute* "Makefile"
5490 (("stringtie: \\$\\{BAM\\}/libbam\\.a")
5491 "stringtie: "))
5492 (substitute* '("gclib/GBam.h"
5493 "gclib/GBam.cpp")
5494 (("#include \"(bam|sam|kstring).h\"" _ header)
5495 (string-append "#include <samtools/" header ".h>")))
5496 #t))
0d2c0562
RW		 5497 (add-after 'unpack 'remove-duplicate-typedef
5498 (lambda _
5499 ;; This typedef conflicts with the typedef in
5500 ;; glibc-2.25/include/bits/types.h
5501 (substitute* "gclib/GThreads.h"
5502 (("typedef long long __intmax_t;") ""))
5503 #t))
d15d981e
RW		 5504 (replace 'install
5505 (lambda* (#:key outputs #:allow-other-keys)
5506 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5507 (install-file "stringtie" bin)
5508 #t))))))
5509 (inputs
5510 `(("samtools" ,samtools-0.1)
5511 ("zlib" ,zlib)))
5512 (home-page "http://ccb.jhu.edu/software/stringtie/")
5513 (synopsis "Transcript assembly and quantification for RNA-Seq data")
5514 (description
5515 "StringTie is a fast and efficient assembler of RNA-Seq sequence
5516alignments into potential transcripts. It uses a novel network flow algorithm
5517as well as an optional de novo assembly step to assemble and quantitate
5518full-length transcripts representing multiple splice variants for each gene
5519locus. Its input can include not only the alignments of raw reads used by
5520other transcript assemblers, but also alignments of longer sequences that have
5521been assembled from those reads. To identify differentially expressed genes
5522between experiments, StringTie's output can be processed either by the
5523Cuffdiff or Ballgown programs.")
5524 (license license:artistic2.0)))
5525
ad0ae297
BW		 5526(define-public taxtastic
5527 (package
5528 (name "taxtastic")
3cbfc149 5529 (version "0.6.4")
ad0ae297
BW		 5530 (source (origin
5531 (method url-fetch)
3cbfc149 5532 (uri (pypi-uri "taxtastic" version))
ad0ae297
BW		 5533 (sha256
5534 (base32
3cbfc149 5535 "0s79z8kfl853x7l4h8ms05k31q87aw62nrchlk20w9n227j35929"))))
ad0ae297
BW		 5536 (build-system python-build-system)
5537 (arguments
5538 `(#:python ,python-2
5539 #:phases
5540 (modify-phases %standard-phases
5541 (replace 'check
5542 (lambda _
5543 (zero? (system* "python" "-m" "unittest" "discover" "-v")))))))
5544 (propagated-inputs
5545 `(("python-sqlalchemy" ,python2-sqlalchemy)
5546 ("python-decorator" ,python2-decorator)
5547 ("python-biopython" ,python2-biopython)
5548 ("python-pandas" ,python2-pandas)))
5549 (home-page "https://github.com/fhcrc/taxtastic")
5550 (synopsis "Tools for taxonomic naming and annotation")
5551 (description
5552 "Taxtastic is software written in python used to build and maintain
5553reference packages i.e. collections of reference trees, reference alignments,
5554profiles, and associated taxonomic information.")
5555 (license license:gpl3+)))
5556
de07c0db
RW		 5557(define-public vcftools
5558 (package
5559 (name "vcftools")
f4322542 5560 (version "0.1.15")
de07c0db
RW		 5561 (source (origin
5562 (method url-fetch)
5563 (uri (string-append
9b36e256
RJ		 5564 "https://github.com/vcftools/vcftools/releases/download/v"
5565 version "/vcftools-" version ".tar.gz"))
de07c0db
RW		 5566 (sha256
5567 (base32
f4322542 5568 "1qw30c45wihgy632rbz4rh3njnwj4msj46l1rsgdhyg6bgypmr1i"))))
de07c0db
RW		 5569 (build-system gnu-build-system)
5570 (arguments
5571 `(#:tests? #f ; no "check" target
5572 #:make-flags (list
7c3958e1 5573 "CFLAGS=-O2" ; override "-m64" flag
de07c0db
RW		 5574 (string-append "PREFIX=" (assoc-ref %outputs "out"))
5575 (string-append "MANDIR=" (assoc-ref %outputs "out")
9b36e256
RJ		 5576 "/share/man/man1"))))
5577 (native-inputs
5578 `(("pkg-config" ,pkg-config)))
de07c0db
RW		 5579 (inputs
5580 `(("perl" ,perl)
5581 ("zlib" ,zlib)))
9b36e256 5582 (home-page "https://vcftools.github.io/")
de07c0db
RW		 5583 (synopsis "Tools for working with VCF files")
5584 (description
5585 "VCFtools is a program package designed for working with VCF files, such
5586as those generated by the 1000 Genomes Project. The aim of VCFtools is to
5587provide easily accessible methods for working with complex genetic variation
5588data in the form of VCF files.")
5589 ;; The license is declared as LGPLv3 in the README and
9b36e256 5590 ;; at https://vcftools.github.io/license.html
de07c0db 5591 (license license:lgpl3)))
9c38b540 5592
35aa90a1
RW		 5593(define-public infernal
5594 (package
5595 (name "infernal")
5596 (version "1.1.2")
5597 (source (origin
5598 (method url-fetch)
5599 (uri (string-append "http://eddylab.org/software/infernal/"
5600 "infernal-" version ".tar.gz"))
5601 (sha256
5602 (base32
5603 "0sr2hiz3qxfwqpz3whxr6n82p3x27336v3f34iqznp10hks2935c"))))
5604 (build-system gnu-build-system)
5605 (native-inputs
5606 `(("perl" ,perl))) ; for tests
5607 (home-page "http://eddylab.org/infernal/")
5608 (synopsis "Inference of RNA alignments")
5609 (description "Infernal (\"INFERence of RNA ALignment\") is a tool for
5610searching DNA sequence databases for RNA structure and sequence similarities.
5611It is an implementation of a special case of profile stochastic context-free
5612grammars called @dfn{covariance models} (CMs). A CM is like a sequence
5613profile, but it scores a combination of sequence consensus and RNA secondary
5614structure consensus, so in many cases, it is more capable of identifying RNA
5615homologs that conserve their secondary structure more than their primary
5616sequence.")
48409ef2
EF		 5617 ;; Infernal 1.1.2 requires VMX or SSE capability for parallel instructions.
5618 (supported-systems '("i686-linux" "x86_64-linux"))
35aa90a1
RW		 5619 (license license:bsd-3)))
5620
b91cfa22
RW		 5621(define-public r-centipede
5622 (package
5623 (name "r-centipede")
5624 (version "1.2")
5625 (source (origin
5626 (method url-fetch)
5627 (uri (string-append "http://download.r-forge.r-project.org/"
5628 "src/contrib/CENTIPEDE_" version ".tar.gz"))
5629 (sha256
5630 (base32
5631 "1hsx6qgwr0i67fhy9257zj7s0ppncph2hjgbia5nn6nfmj0ax6l9"))))
5632 (build-system r-build-system)
5633 (home-page "http://centipede.uchicago.edu/")
5634 (synopsis "Predict transcription factor binding sites")
5635 (description
5636 "CENTIPEDE applies a hierarchical Bayesian mixture model to infer regions
5637of the genome that are bound by particular transcription factors. It starts
5638by identifying a set of candidate binding sites, and then aims to classify the
5639sites according to whether each site is bound or not bound by a transcription
5640factor. CENTIPEDE is an unsupervised learning algorithm that discriminates
5641between two different types of motif instances using as much relevant
5642information as possible.")
5643 (license (list license:gpl2+ license:gpl3+))))
5644
7b3df1e5
BW		 5645(define-public r-vegan
5646 (package
5647 (name "r-vegan")
db2e4386 5648 (version "2.4-4")
7b3df1e5
BW		 5649 (source
5650 (origin
5651 (method url-fetch)
5652 (uri (cran-uri "vegan" version))
5653 (sha256
5654 (base32
db2e4386 5655 "1n57dzv2aid6iqd9fkqik401sidqanhzsawyak94qbiyh6dbd1x9"))))
7b3df1e5 5656 (build-system r-build-system)
7b3df1e5 5657 (native-inputs
db2e4386 5658 `(("gfortran" ,gfortran)))
7b3df1e5
BW		 5659 (propagated-inputs
5660 `(("r-cluster" ,r-cluster)
5661 ("r-lattice" ,r-lattice)
aeb64f3c 5662 ("r-mass" ,r-mass)
7b3df1e5
BW		 5663 ("r-mgcv" ,r-mgcv)
5664 ("r-permute" ,r-permute)))
5665 (home-page "https://cran.r-project.org/web/packages/vegan")
5666 (synopsis "Functions for community ecology")
5667 (description
5668 "The vegan package provides tools for descriptive community ecology. It
5669has most basic functions of diversity analysis, community ordination and
5670dissimilarity analysis. Most of its multivariate tools can be used for other
5671data types as well.")
5672 (license license:gpl2+)))
5673
8c6de588
RW		 5674(define-public r-annotate
5675 (package
5676 (name "r-annotate")
2d4e887d 5677 (version "1.56.0")
8c6de588
RW		 5678 (source
5679 (origin
5680 (method url-fetch)
5681 (uri (bioconductor-uri "annotate" version))
5682 (sha256
5683 (base32
2d4e887d 5684 "0wlrp3v2jxw9is98ap39dfi7z97kmw1wv1xi4h7yfh12zpj2r8l0"))))
8c6de588
RW		 5685 (build-system r-build-system)
5686 (propagated-inputs
5687 `(("r-annotationdbi" ,r-annotationdbi)
5688 ("r-biobase" ,r-biobase)
5689 ("r-biocgenerics" ,r-biocgenerics)
5690 ("r-dbi" ,r-dbi)
d0f0579e 5691 ("r-rcurl" ,r-rcurl)
8c6de588
RW		 5692 ("r-xml" ,r-xml)
5693 ("r-xtable" ,r-xtable)))
5694 (home-page
5713bbf1 5695 "https://bioconductor.org/packages/annotate")
8c6de588 5696 (synopsis "Annotation for microarrays")
d1e4ad1b 5697 (description "This package provides R environments for the annotation of
8c6de588
RW		 5698microarrays.")
5699 (license license:artistic2.0)))
5700
07a664cd
RW		 5701(define-public r-geneplotter
5702 (package
5703 (name "r-geneplotter")
cb4d3ff2 5704 (version "1.56.0")
07a664cd
RW		 5705 (source
5706 (origin
5707 (method url-fetch)
5708 (uri (bioconductor-uri "geneplotter" version))
5709 (sha256
5710 (base32
cb4d3ff2 5711 "1z3g7frc1iviwrsv2dlm4nqvkc0685h4va0388yfxn102ln8wwma"))))
07a664cd
RW		 5712 (build-system r-build-system)
5713 (propagated-inputs
5714 `(("r-annotate" ,r-annotate)
5715 ("r-annotationdbi" ,r-annotationdbi)
5716 ("r-biobase" ,r-biobase)
5717 ("r-biocgenerics" ,r-biocgenerics)
5718 ("r-lattice" ,r-lattice)
5719 ("r-rcolorbrewer" ,r-rcolorbrewer)))
5713bbf1 5720 (home-page "https://bioconductor.org/packages/geneplotter")
07a664cd
RW		 5721 (synopsis "Graphics functions for genomic data")
5722 (description
5723 "This package provides functions for plotting genomic data.")
5724 (license license:artistic2.0)))
5725
2301fd3e
RW		 5726(define-public r-genefilter
5727 (package
5728 (name "r-genefilter")
554fac74 5729 (version "1.60.0")
2301fd3e
RW		 5730 (source
5731 (origin
5732 (method url-fetch)
5733 (uri (bioconductor-uri "genefilter" version))
5734 (sha256
5735 (base32
554fac74 5736 "173swlg6gj4kdllbqvyiw5dggbcxiwlwpqmllsv4dxzn7h25i3g7"))))
2301fd3e
RW		 5737 (build-system r-build-system)
5738 (native-inputs
5739 `(("gfortran" ,gfortran)))
5740 (propagated-inputs
5741 `(("r-annotate" ,r-annotate)
5742 ("r-annotationdbi" ,r-annotationdbi)
5743 ("r-biobase" ,r-biobase)
aeb64f3c
RW		 5744 ("r-s4vectors" ,r-s4vectors)
5745 ("r-survival" ,r-survival)))
5713bbf1 5746 (home-page "https://bioconductor.org/packages/genefilter")
2301fd3e
RW		 5747 (synopsis "Filter genes from high-throughput experiments")
5748 (description
5749 "This package provides basic functions for filtering genes from
5750high-throughput sequencing experiments.")
5751 (license license:artistic2.0)))
5752
ad34f0ac
RW		 5753(define-public r-deseq2
5754 (package
5755 (name "r-deseq2")
0166dc07 5756 (version "1.18.0")
ad34f0ac
RW		 5757 (source
5758 (origin
5759 (method url-fetch)
5760 (uri (bioconductor-uri "DESeq2" version))
5761 (sha256
5762 (base32
0166dc07 5763 "1hcxnkkjfvz4hj8iqidshwsjq7jnl1z7wj63dvcwlx1zx5aichyh"))))
ad34f0ac
RW		 5764 (properties `((upstream-name . "DESeq2")))
5765 (build-system r-build-system)
ad34f0ac
RW		 5766 (propagated-inputs
5767 `(("r-biobase" ,r-biobase)
5768 ("r-biocgenerics" ,r-biocgenerics)
5769 ("r-biocparallel" ,r-biocparallel)
5770 ("r-genefilter" ,r-genefilter)
5771 ("r-geneplotter" ,r-geneplotter)
5772 ("r-genomicranges" ,r-genomicranges)
5773 ("r-ggplot2" ,r-ggplot2)
5774 ("r-hmisc" ,r-hmisc)
5775 ("r-iranges" ,r-iranges)
5776 ("r-locfit" ,r-locfit)
5777 ("r-rcpp" ,r-rcpp)
5778 ("r-rcpparmadillo" ,r-rcpparmadillo)
5779 ("r-s4vectors" ,r-s4vectors)
5780 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5781 (home-page "https://bioconductor.org/packages/DESeq2")
ad34f0ac
RW		 5782 (synopsis "Differential gene expression analysis")
5783 (description
5784 "This package provides functions to estimate variance-mean dependence in
5785count data from high-throughput nucleotide sequencing assays and test for
5786differential expression based on a model using the negative binomial
5787distribution.")
5788 (license license:lgpl3+)))
5789
86763fdd
RW		 5790(define-public r-dexseq
5791 (package
5792 (name "r-dexseq")
3fd010ff 5793 (version "1.24.0")
86763fdd
RW		 5794 (source
5795 (origin
5796 (method url-fetch)
5797 (uri (bioconductor-uri "DEXSeq" version))
5798 (sha256
5799 (base32
3fd010ff 5800 "0qxwnz2ffhav9slcn095k206cfza9i3i5l7w1154plf08gpy1d1d"))))
86763fdd
RW		 5801 (properties `((upstream-name . "DEXSeq")))
5802 (build-system r-build-system)
5803 (propagated-inputs
5804 `(("r-annotationdbi" ,r-annotationdbi)
5805 ("r-biobase" ,r-biobase)
5806 ("r-biocgenerics" ,r-biocgenerics)
5807 ("r-biocparallel" ,r-biocparallel)
5808 ("r-biomart" ,r-biomart)
5809 ("r-deseq2" ,r-deseq2)
5810 ("r-genefilter" ,r-genefilter)
5811 ("r-geneplotter" ,r-geneplotter)
5812 ("r-genomicranges" ,r-genomicranges)
5813 ("r-hwriter" ,r-hwriter)
5814 ("r-iranges" ,r-iranges)
5815 ("r-rcolorbrewer" ,r-rcolorbrewer)
5816 ("r-rsamtools" ,r-rsamtools)
5817 ("r-s4vectors" ,r-s4vectors)
5818 ("r-statmod" ,r-statmod)
5819 ("r-stringr" ,r-stringr)
5820 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5821 (home-page "https://bioconductor.org/packages/DEXSeq")
86763fdd
RW		 5822 (synopsis "Inference of differential exon usage in RNA-Seq")
5823 (description
5824 "This package is focused on finding differential exon usage using RNA-seq
5825exon counts between samples with different experimental designs. It provides
5826functions that allows the user to make the necessary statistical tests based
5827on a model that uses the negative binomial distribution to estimate the
5828variance between biological replicates and generalized linear models for
5829testing. The package also provides functions for the visualization and
5830exploration of the results.")
5831 (license license:gpl3+)))
5832
e8163773
RW		 5833(define-public r-annotationforge
5834 (package
5835 (name "r-annotationforge")
727706ad 5836 (version "1.20.0")
e8163773
RW		 5837 (source
5838 (origin
5839 (method url-fetch)
5840 (uri (bioconductor-uri "AnnotationForge" version))
5841 (sha256
5842 (base32
727706ad 5843 "01vbrf76vqfvxh6vpfxkjwccxggnha3byqzj333glqz2b6kwx5q1"))))
e8163773
RW		 5844 (properties
5845 `((upstream-name . "AnnotationForge")))
5846 (build-system r-build-system)
5847 (propagated-inputs
5848 `(("r-annotationdbi" ,r-annotationdbi)
5849 ("r-biobase" ,r-biobase)
5850 ("r-biocgenerics" ,r-biocgenerics)
5851 ("r-dbi" ,r-dbi)
55cd914c 5852 ("r-rcurl" ,r-rcurl)
e8163773
RW		 5853 ("r-rsqlite" ,r-rsqlite)
5854 ("r-s4vectors" ,r-s4vectors)
5855 ("r-xml" ,r-xml)))
5713bbf1 5856 (home-page "https://bioconductor.org/packages/AnnotationForge")
e8163773
RW		 5857 (synopsis "Code for building annotation database packages")
5858 (description
5859 "This package provides code for generating Annotation packages and their
5860databases. Packages produced are intended to be used with AnnotationDbi.")
5861 (license license:artistic2.0)))
5862
cd9e7dc7
RW		 5863(define-public r-rbgl
5864 (package
5865 (name "r-rbgl")
c25154e5 5866 (version "1.54.0")
cd9e7dc7
RW		 5867 (source
5868 (origin
5869 (method url-fetch)
5870 (uri (bioconductor-uri "RBGL" version))
5871 (sha256
5872 (base32
c25154e5 5873 "18jad23i3899ypv4bg3l47cvvs3qnj1pqis2p9x0135yv5y6wnv7"))))
cd9e7dc7
RW		 5874 (properties `((upstream-name . "RBGL")))
5875 (build-system r-build-system)
5876 (propagated-inputs `(("r-graph" ,r-graph)))
5713bbf1 5877 (home-page "https://www.bioconductor.org/packages/RBGL")
cd9e7dc7
RW		 5878 (synopsis "Interface to the Boost graph library")
5879 (description
5880 "This package provides a fairly extensive and comprehensive interface to
5881the graph algorithms contained in the Boost library.")
5882 (license license:artistic2.0)))
5883
ad740ff8
RW		 5884(define-public r-gseabase
5885 (package
5886 (name "r-gseabase")
9f52e53d 5887 (version "1.40.0")
ad740ff8
RW		 5888 (source
5889 (origin
5890 (method url-fetch)
5891 (uri (bioconductor-uri "GSEABase" version))
5892 (sha256
5893 (base32
9f52e53d 5894 "0kpkl6c5lrar6ip7wlhvd5axqlb9lb5l3lgbdb3dlih32c3nz0yq"))))
ad740ff8
RW		 5895 (properties `((upstream-name . "GSEABase")))
5896 (build-system r-build-system)
5897 (propagated-inputs
5898 `(("r-annotate" ,r-annotate)
5899 ("r-annotationdbi" ,r-annotationdbi)
5900 ("r-biobase" ,r-biobase)
5901 ("r-biocgenerics" ,r-biocgenerics)
5902 ("r-graph" ,r-graph)
5903 ("r-xml" ,r-xml)))
5713bbf1 5904 (home-page "https://bioconductor.org/packages/GSEABase")
ad740ff8
RW		 5905 (synopsis "Gene set enrichment data structures and methods")
5906 (description
5907 "This package provides classes and methods to support @dfn{Gene Set
5908Enrichment Analysis} (GSEA).")
5909 (license license:artistic2.0)))
5910
1a1931f7
RW		 5911(define-public r-category
5912 (package
5913 (name "r-category")
2404cc42 5914 (version "2.44.0")
1a1931f7
RW		 5915 (source
5916 (origin
5917 (method url-fetch)
5918 (uri (bioconductor-uri "Category" version))
5919 (sha256
5920 (base32
2404cc42 5921 "0mkav04vbla0xfa0dssxdd0rjs589sxi83xklf5iq5hj3dm8y0i8"))))
1a1931f7
RW		 5922 (properties `((upstream-name . "Category")))
5923 (build-system r-build-system)
5924 (propagated-inputs
5925 `(("r-annotate" ,r-annotate)
5926 ("r-annotationdbi" ,r-annotationdbi)
5927 ("r-biobase" ,r-biobase)
5928 ("r-biocgenerics" ,r-biocgenerics)
5929 ("r-genefilter" ,r-genefilter)
5930 ("r-graph" ,r-graph)
5931 ("r-gseabase" ,r-gseabase)
5932 ("r-matrix" ,r-matrix)
5933 ("r-rbgl" ,r-rbgl)
2404cc42 5934 ("r-dbi" ,r-dbi)))
5713bbf1 5935 (home-page "https://bioconductor.org/packages/Category")
1a1931f7
RW		 5936 (synopsis "Category analysis")
5937 (description
5938 "This package provides a collection of tools for performing category
5939analysis.")
5940 (license license:artistic2.0)))
5941
89f40c5e
RW		 5942(define-public r-gostats
5943 (package
5944 (name "r-gostats")
eb3f5cc7 5945 (version "2.44.0")
89f40c5e
RW		 5946 (source
5947 (origin
5948 (method url-fetch)
5949 (uri (bioconductor-uri "GOstats" version))
5950 (sha256
5951 (base32
eb3f5cc7 5952 "04gqfdlx9fxf97qf0l28x4aaqvl10n6v58qiz5fiaw05sbj1pf1i"))))
89f40c5e
RW		 5953 (properties `((upstream-name . "GOstats")))
5954 (build-system r-build-system)
5955 (propagated-inputs
5956 `(("r-annotate" ,r-annotate)
5957 ("r-annotationdbi" ,r-annotationdbi)
5958 ("r-annotationforge" ,r-annotationforge)
5959 ("r-biobase" ,r-biobase)
5960 ("r-category" ,r-category)
5961 ("r-go-db" ,r-go-db)
5962 ("r-graph" ,r-graph)
eb3f5cc7 5963 ("r-rgraphviz" ,r-rgraphviz)
89f40c5e 5964 ("r-rbgl" ,r-rbgl)))
5713bbf1 5965 (home-page "https://bioconductor.org/packages/GOstats")
89f40c5e
RW		 5966 (synopsis "Tools for manipulating GO and microarrays")
5967 (description
5968 "This package provides a set of tools for interacting with GO and
5969microarray data. A variety of basic manipulation tools for graphs, hypothesis
5970testing and other simple calculations.")
5971 (license license:artistic2.0)))
5972
cb99d457
RW		 5973(define-public r-shortread
5974 (package
5975 (name "r-shortread")
843826e2 5976 (version "1.36.0")
cb99d457
RW		 5977 (source
5978 (origin
5979 (method url-fetch)
5980 (uri (bioconductor-uri "ShortRead" version))
5981 (sha256
5982 (base32
843826e2 5983 "06mknlsmd4hnaxzdjapgvp2kgdnf9w103y500dsac5jgsz4vwzcz"))))
cb99d457
RW		 5984 (properties `((upstream-name . "ShortRead")))
5985 (build-system r-build-system)
5986 (inputs
5987 `(("zlib" ,zlib)))
5988 (propagated-inputs
5989 `(("r-biobase" ,r-biobase)
5990 ("r-biocgenerics" ,r-biocgenerics)
5991 ("r-biocparallel" ,r-biocparallel)
5992 ("r-biostrings" ,r-biostrings)
5993 ("r-genomeinfodb" ,r-genomeinfodb)
5994 ("r-genomicalignments" ,r-genomicalignments)
5995 ("r-genomicranges" ,r-genomicranges)
5996 ("r-hwriter" ,r-hwriter)
5997 ("r-iranges" ,r-iranges)
5998 ("r-lattice" ,r-lattice)
5999 ("r-latticeextra" ,r-latticeextra)
6000 ("r-rsamtools" ,r-rsamtools)
6001 ("r-s4vectors" ,r-s4vectors)
6002 ("r-xvector" ,r-xvector)
6003 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 6004 (home-page "https://bioconductor.org/packages/ShortRead")
cb99d457
RW		 6005 (synopsis "FASTQ input and manipulation tools")
6006 (description
6007 "This package implements sampling, iteration, and input of FASTQ files.
6008It includes functions for filtering and trimming reads, and for generating a
6009quality assessment report. Data are represented as
6010@code{DNAStringSet}-derived objects, and easily manipulated for a diversity of
6011purposes. The package also contains legacy support for early single-end,
6012ungapped alignment formats.")
6013 (license license:artistic2.0)))
6014
7f903d73
RW		 6015(define-public r-systempiper
6016 (package
6017 (name "r-systempiper")
b5375525 6018 (version "1.12.0")
7f903d73
RW		 6019 (source
6020 (origin
6021 (method url-fetch)
6022 (uri (bioconductor-uri "systemPipeR" version))
6023 (sha256
6024 (base32
b5375525 6025 "11mj8pjq5vj25768vmagpzv74fvi3p3kdk5zdlznqyiaggri04cv"))))
7f903d73
RW		 6026 (properties `((upstream-name . "systemPipeR")))
6027 (build-system r-build-system)
6028 (propagated-inputs
6029 `(("r-annotate" ,r-annotate)
6030 ("r-batchjobs" ,r-batchjobs)
6031 ("r-biocgenerics" ,r-biocgenerics)
6032 ("r-biostrings" ,r-biostrings)
6033 ("r-deseq2" ,r-deseq2)
6034 ("r-edger" ,r-edger)
6035 ("r-genomicfeatures" ,r-genomicfeatures)
6036 ("r-genomicranges" ,r-genomicranges)
6037 ("r-ggplot2" ,r-ggplot2)
6038 ("r-go-db" ,r-go-db)
6039 ("r-gostats" ,r-gostats)
6040 ("r-limma" ,r-limma)
6041 ("r-pheatmap" ,r-pheatmap)
6042 ("r-rjson" ,r-rjson)
6043 ("r-rsamtools" ,r-rsamtools)
6044 ("r-shortread" ,r-shortread)
6045 ("r-summarizedexperiment" ,r-summarizedexperiment)
6046 ("r-variantannotation" ,r-variantannotation)))
6047 (home-page "https://github.com/tgirke/systemPipeR")
6048 (synopsis "Next generation sequencing workflow and reporting environment")
6049 (description
6050 "This R package provides tools for building and running automated
6051end-to-end analysis workflows for a wide range of @dfn{next generation
6052sequence} (NGS) applications such as RNA-Seq, ChIP-Seq, VAR-Seq and Ribo-Seq.
6053Important features include a uniform workflow interface across different NGS
6054applications, automated report generation, and support for running both R and
6055command-line software, such as NGS aligners or peak/variant callers, on local
6056computers or compute clusters. Efficient handling of complex sample sets and
6057experimental designs is facilitated by a consistently implemented sample
6058annotation infrastructure.")
6059 (license license:artistic2.0)))
6060
684f29bd
RW		 6061(define-public r-grohmm
6062 (package
6063 (name "r-grohmm")
704fe4d1 6064 (version "1.12.0")
684f29bd
RW		 6065 (source
6066 (origin
6067 (method url-fetch)
6068 (uri (bioconductor-uri "groHMM" version))
6069 (sha256
6070 (base32
704fe4d1 6071 "0cjkj0ypyc4dfi9s8dh88kh6q4xlpnc0wal7njg4b4gqj0l2hva7"))))
684f29bd
RW		 6072 (properties `((upstream-name . "groHMM")))
6073 (build-system r-build-system)
6074 (propagated-inputs
6075 `(("r-genomeinfodb" ,r-genomeinfodb)
6076 ("r-genomicalignments" ,r-genomicalignments)
6077 ("r-genomicranges" ,r-genomicranges)
6078 ("r-iranges" ,r-iranges)
aeb64f3c 6079 ("r-mass" ,r-mass)
684f29bd
RW		 6080 ("r-rtracklayer" ,r-rtracklayer)
6081 ("r-s4vectors" ,r-s4vectors)))
6082 (home-page "https://github.com/Kraus-Lab/groHMM")
6083 (synopsis "GRO-seq analysis pipeline")
6084 (description
6085 "This package provides a pipeline for the analysis of GRO-seq data.")
6086 (license license:gpl3+)))
6087
f3cfe451
RW		 6088(define-public r-txdb-hsapiens-ucsc-hg19-knowngene
6089 (package
6090 (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
6091 (version "3.2.2")
6092 (source (origin
6093 (method url-fetch)
6094 ;; We cannot use bioconductor-uri here because this tarball is
6095 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 6096 (uri (string-append "https://bioconductor.org/packages/"
f3cfe451
RW		 6097 "release/data/annotation/src/contrib"
6098 "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
6099 version ".tar.gz"))
6100 (sha256
6101 (base32
6102 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
6103 (properties
6104 `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
6105 (build-system r-build-system)
6106 ;; As this package provides little more than a very large data file it
6107 ;; doesn't make sense to build substitutes.
6108 (arguments `(#:substitutable? #f))
6109 (propagated-inputs
6110 `(("r-genomicfeatures" ,r-genomicfeatures)))
6111 (home-page
5713bbf1 6112 "https://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
f3cfe451
RW		 6113 (synopsis "Annotation package for human genome in TxDb format")
6114 (description
6115 "This package provides an annotation database of Homo sapiens genome
6116data. It is derived from the UCSC hg19 genome and based on the \"knownGene\"
6117track. The database is exposed as a @code{TxDb} object.")
6118 (license license:artistic2.0)))
6119
325c039c
RJ		 6120(define-public r-sparql
6121 (package
6122 (name "r-sparql")
6123 (version "1.16")
6124 (source (origin
6125 (method url-fetch)
6126 (uri (cran-uri "SPARQL" version))
6127 (sha256
6128 (base32
6129 "0gak1q06yyhdmcxb2n3v0h9gr1vqd0viqji52wpw211qp6r6dcrc"))))
6130 (properties `((upstream-name . "SPARQL")))
6131 (build-system r-build-system)
6132 (propagated-inputs
6133 `(("r-rcurl" ,r-rcurl)
6134 ("r-xml" ,r-xml)))
6135 (home-page "http://cran.r-project.org/web/packages/SPARQL")
6136 (synopsis "SPARQL client for R")
6137 (description "This package provides an interface to use SPARQL to pose
6138SELECT or UPDATE queries to an end-point.")
6139 ;; The only license indication is found in the DESCRIPTION file,
6140 ;; which states GPL-3. So we cannot assume GPLv3+.
6141 (license license:gpl3)))
6142
a2950fa4
BW		 6143(define-public vsearch
6144 (package
6145 (name "vsearch")
7c087150 6146 (version "2.6.0")
a2950fa4
BW		 6147 (source
6148 (origin
6149 (method url-fetch)
6150 (uri (string-append
6151 "https://github.com/torognes/vsearch/archive/v"
6152 version ".tar.gz"))
6153 (file-name (string-append name "-" version ".tar.gz"))
6154 (sha256
6155 (base32
7c087150 6156 "0rplgpvsdkxw0k371ckxrp6i77jn93ckhslaazwbyd85m83nkynr"))
206af46f 6157 (patches (search-patches "vsearch-unbundle-cityhash.patch"))
a2950fa4
BW		 6158 (snippet
6159 '(begin
206af46f
BW		 6160 ;; Remove bundled cityhash sources. The vsearch source is adjusted
6161 ;; for this in the patch.
cf6edaba
BW		 6162 (delete-file "src/city.h")
6163 (delete-file "src/citycrc.h")
6164 (delete-file "src/city.cc")
a2950fa4
BW		 6165 #t))))
6166 (build-system gnu-build-system)
6167 (arguments
6168 `(#:phases
6169 (modify-phases %standard-phases
d10092b8
KK		 6170 (add-after 'unpack 'autogen
6171 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
a2950fa4
BW		 6172 (inputs
6173 `(("zlib" ,zlib)
6174 ("bzip2" ,bzip2)
6175 ("cityhash" ,cityhash)))
6176 (native-inputs
6177 `(("autoconf" ,autoconf)
6178 ("automake" ,automake)))
6179 (synopsis "Sequence search tools for metagenomics")
6180 (description
6181 "VSEARCH supports DNA sequence searching, clustering, chimera detection,
6182dereplication, pairwise alignment, shuffling, subsampling, sorting and
6183masking. The tool takes advantage of parallelism in the form of SIMD
6184vectorization as well as multiple threads to perform accurate alignments at
6185high speed. VSEARCH uses an optimal global aligner (full dynamic programming
6186Needleman-Wunsch).")
6187 (home-page "https://github.com/torognes/vsearch")
6f04e515
BW		 6188 ;; vsearch uses non-portable SSE intrinsics so building fails on other
6189 ;; platforms.
6190 (supported-systems '("x86_64-linux"))
a2950fa4
BW		 6191 ;; Dual licensed; also includes public domain source.
6192 (license (list license:gpl3 license:bsd-2))))
6193
07837874
RW		 6194(define-public pardre
6195 (package
6196 (name "pardre")
7922ab8f
BW		 6197 ;; The source of 1.1.5 changed in place, so we append "-1" to the version.
6198 (version "1.1.5-1")
07837874
RW		 6199 (source
6200 (origin
6201 (method url-fetch)
6202 (uri (string-append "mirror://sourceforge/pardre/ParDRe-rel"
7922ab8f 6203 "1.1.5" ".tar.gz"))
07837874
RW		 6204 (sha256
6205 (base32
7922ab8f 6206 "17j73nc0viq4f6qj50nrndsrif5d6b71q8fl87m54psiv0ilns2b"))))
07837874
RW		 6207 (build-system gnu-build-system)
6208 (arguments
6209 `(#:tests? #f ; no tests included
6210 #:phases
6211 (modify-phases %standard-phases
6212 (delete 'configure)
6213 (replace 'install
6214 (lambda* (#:key outputs #:allow-other-keys)
6215 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
07837874
RW		 6216 (install-file "ParDRe" bin)
6217 #t))))))
6218 (inputs
6219 `(("openmpi" ,openmpi)
6220 ("zlib" ,zlib)))
6221 (synopsis "Parallel tool to remove duplicate DNA reads")
6222 (description
6223 "ParDRe is a parallel tool to remove duplicate genetic sequence reads.
6224Duplicate reads can be seen as identical or nearly identical sequences with
6225some mismatches. This tool lets users avoid the analysis of unnecessary
6226reads, reducing the time of subsequent procedures with the
6227dataset (e.g. assemblies, mappings, etc.). The tool is implemented with MPI
6228in order to exploit the parallel capabilities of multicore clusters. It is
6229faster than multithreaded counterparts (end of 2015) for the same number of
6230cores and, thanks to the message-passing technology, it can be executed on
6231clusters.")
6232 (home-page "https://sourceforge.net/projects/pardre/")
6233 (license license:gpl3+)))
6234
e4a44a6a
BW		 6235(define-public ruby-bio-kseq
6236 (package
6237 (name "ruby-bio-kseq")
6238 (version "0.0.2")
6239 (source
6240 (origin
6241 (method url-fetch)
6242 (uri (rubygems-uri "bio-kseq" version))
6243 (sha256
6244 (base32
6245 "1xyaha46khb5jc6wzkbf7040jagac49jbimn0vcrzid0j8jdikrz"))))
6246 (build-system ruby-build-system)
6247 (arguments
6248 `(#:test-target "spec"))
6249 (native-inputs
6250 `(("bundler" ,bundler)
6251 ("ruby-rspec" ,ruby-rspec)
6252 ("ruby-rake-compiler" ,ruby-rake-compiler)))
6253 (inputs
6254 `(("zlib" ,zlib)))
6255 (synopsis "Ruby bindings for the kseq.h FASTA/Q parser")
6256 (description
6257 "@code{Bio::Kseq} provides ruby bindings to the @code{kseq.h} FASTA and
6258FASTQ parsing code. It provides a fast iterator over sequences and their
6259quality scores.")
6260 (home-page "https://github.com/gusevfe/bio-kseq")
6261 (license license:expat)))
6262
9c38b540
PP		 6263(define-public bio-locus
6264 (package
6265 (name "bio-locus")
6266 (version "0.0.7")
6267 (source
6268 (origin
6269 (method url-fetch)
6270 (uri (rubygems-uri "bio-locus" version))
6271 (sha256
6272 (base32
6273 "02vmrxyimkj9sahsp4zhfhnmbvz6dbbqz1y01vglf8cbwvkajfl0"))))
6274 (build-system ruby-build-system)
6275 (native-inputs
6276 `(("ruby-rspec" ,ruby-rspec)))
6277 (synopsis "Tool for fast querying of genome locations")
6278 (description
6279 "Bio-locus is a tabix-like tool for fast querying of genome
6280locations. Many file formats in bioinformatics contain records that
6281start with a chromosome name and a position for a SNP, or a start-end
6282position for indels. Bio-locus allows users to store this chr+pos or
6283chr+pos+alt information in a database.")
6284 (home-page "https://github.com/pjotrp/bio-locus")
6285 (license license:expat)))
edb15985 6286
b2bddb07
PP		 6287(define-public bio-blastxmlparser
6288 (package
6289 (name "bio-blastxmlparser")
6290 (version "2.0.4")
6291 (source (origin
6292 (method url-fetch)
6293 (uri (rubygems-uri "bio-blastxmlparser" version))
6294 (sha256
6295 (base32
6296 "1wf4qygcmdjgcqm6flmvsagfr1gs9lf63mj32qv3z1f481zc5692"))))
6297 (build-system ruby-build-system)
6298 (propagated-inputs
6299 `(("ruby-bio-logger" ,ruby-bio-logger)
6300 ("ruby-nokogiri" ,ruby-nokogiri)))
6301 (inputs
6302 `(("ruby-rspec" ,ruby-rspec)))
6303 (synopsis "Fast big data BLAST XML parser and library")
6304 (description
6305 "Very fast parallel big-data BLAST XML file parser which can be used as
6306command line utility. Use blastxmlparser to: Parse BLAST XML; filter output;
6307generate FASTA, JSON, YAML, RDF, JSON-LD, HTML, CSV, tabular output etc.")
7bf837fd 6308 (home-page "https://github.com/pjotrp/blastxmlparser")
b2bddb07
PP		 6309 (license license:expat)))
6310
edb15985
PP		 6311(define-public bioruby
6312 (package
6313 (name "bioruby")
dbf9d371 6314 (version "1.5.1")
edb15985
PP		 6315 (source
6316 (origin
6317 (method url-fetch)
6318 (uri (rubygems-uri "bio" version))
6319 (sha256
6320 (base32
dbf9d371 6321 "0hdl0789c9n4mprnx5pgd46bfwl8d000rqpamj5h6kkjgspijv49"))))
edb15985
PP		 6322 (build-system ruby-build-system)
6323 (propagated-inputs
6324 `(("ruby-libxml" ,ruby-libxml)))
6325 (native-inputs
6326 `(("which" ,which))) ; required for test phase
6327 (arguments
6328 `(#:phases
6329 (modify-phases %standard-phases
6330 (add-before 'build 'patch-test-command
6331 (lambda _
6332 (substitute* '("test/functional/bio/test_command.rb")
6333 (("/bin/sh") (which "sh")))
6334 (substitute* '("test/functional/bio/test_command.rb")
6335 (("/bin/ls") (which "ls")))
6336 (substitute* '("test/functional/bio/test_command.rb")
6337 (("which") (which "which")))
6338 (substitute* '("test/functional/bio/test_command.rb",
6339 "test/data/command/echoarg2.sh")
6340 (("/bin/echo") (which "echo")))
6341 #t)))))
6342 (synopsis "Ruby library, shell and utilities for bioinformatics")
6343 (description "BioRuby comes with a comprehensive set of Ruby development
6344tools and libraries for bioinformatics and molecular biology. BioRuby has
6345components for sequence analysis, pathway analysis, protein modelling and
6346phylogenetic analysis; it supports many widely used data formats and provides
6347easy access to databases, external programs and public web services, including
6348BLAST, KEGG, GenBank, MEDLINE and GO.")
6349 (home-page "http://bioruby.org/")
6350 ;; Code is released under Ruby license, except for setup
6351 ;; (LGPLv2.1+) and scripts in samples (which have GPL2 and GPL2+)
6352 (license (list license:ruby license:lgpl2.1+ license:gpl2+ ))))
a5002ae7 6353
9fba89e8
RW		 6354(define-public r-acsnminer
6355 (package
6356 (name "r-acsnminer")
0b54b4c9 6357 (version "0.16.8.25")
9fba89e8
RW		 6358 (source (origin
6359 (method url-fetch)
6360 (uri (cran-uri "ACSNMineR" version))
6361 (sha256
6362 (base32
0b54b4c9 6363 "0gh604s8qall6zfjlwcg2ilxjvz08dplf9k5g47idhv43scm748l"))))
9fba89e8
RW		 6364 (properties `((upstream-name . "ACSNMineR")))
6365 (build-system r-build-system)
6366 (propagated-inputs
6367 `(("r-ggplot2" ,r-ggplot2)
6368 ("r-gridextra" ,r-gridextra)))
6369 (home-page "http://cran.r-project.org/web/packages/ACSNMineR")
6370 (synopsis "Gene enrichment analysis")
6371 (description
6372 "This package provides tools to compute and represent gene set enrichment
6373or depletion from your data based on pre-saved maps from the @dfn{Atlas of
6374Cancer Signalling Networks} (ACSN) or user imported maps. The gene set
6375enrichment can be run with hypergeometric test or Fisher exact test, and can
6376use multiple corrections. Visualization of data can be done either by
6377barplots or heatmaps.")
6378 (license license:gpl2+)))
6379
d29b25c4
RW		 6380(define-public r-biocgenerics
6381 (package
6382 (name "r-biocgenerics")
d61fc4e1 6383 (version "0.24.0")
d29b25c4
RW		 6384 (source (origin
6385 (method url-fetch)
6386 (uri (bioconductor-uri "BiocGenerics" version))
6387 (sha256
6388 (base32
d61fc4e1 6389 "03wxvhxyrhipbgcg83lqlfn7p9gbzzrnl48y0dq7303xgp232zai"))))
d29b25c4 6390 (properties
1d216b6e 6391 `((upstream-name . "BiocGenerics")))
d29b25c4 6392 (build-system r-build-system)
5713bbf1 6393 (home-page "https://bioconductor.org/packages/BiocGenerics")
d29b25c4
RW		 6394 (synopsis "S4 generic functions for Bioconductor")
6395 (description
6396 "This package provides S4 generic functions needed by many Bioconductor
6397packages.")
6398 (license license:artistic2.0)))
6399
eb24341f
RJ		 6400(define-public r-biocinstaller
6401 (package
6402 (name "r-biocinstaller")
32414616 6403 (version "1.28.0")
eb24341f
RJ		 6404 (source (origin
6405 (method url-fetch)
6406 (uri (bioconductor-uri "BiocInstaller" version))
6407 (sha256
6408 (base32
32414616 6409 "19fga27bv6q9v5mpil74y76lahmnwvpg2h33rdx1r79nvljkd19d"))))
eb24341f
RJ		 6410 (properties
6411 `((upstream-name . "BiocInstaller")))
6412 (build-system r-build-system)
5713bbf1 6413 (home-page "https://bioconductor.org/packages/BiocInstaller")
eb24341f
RJ		 6414 (synopsis "Install Bioconductor packages")
6415 (description "This package is used to install and update R packages from
6416Bioconductor, CRAN, and Github.")
6417 (license license:artistic2.0)))
6418
207ce8fb
RJ		 6419(define-public r-biocviews
6420 (package
6421 (name "r-biocviews")
c7782c3f 6422 (version "1.46.0")
207ce8fb
RJ		 6423 (source (origin
6424 (method url-fetch)
6425 (uri (bioconductor-uri "biocViews" version))
6426 (sha256
6427 (base32
c7782c3f 6428 "09zyqj1kqc089lmh9sliy0acanx9zimcasvp71dsrg2bqm08r1md"))))
207ce8fb
RJ		 6429 (properties
6430 `((upstream-name . "biocViews")))
6431 (build-system r-build-system)
6432 (propagated-inputs
6433 `(("r-biobase" ,r-biobase)
6434 ("r-graph" ,r-graph)
6435 ("r-rbgl" ,r-rbgl)
6436 ("r-rcurl" ,r-rcurl)
6437 ("r-xml" ,r-xml)
207ce8fb 6438 ("r-runit" ,r-runit)))
5713bbf1 6439 (home-page "https://bioconductor.org/packages/biocViews")
207ce8fb
RJ		 6440 (synopsis "Bioconductor package categorization helper")
6441 (description "The purpose of biocViews is to create HTML pages that
6442categorize packages in a Bioconductor package repository according to keywords,
6443also known as views, in a controlled vocabulary.")
6444 (license license:artistic2.0)))
6445
2abfc5b8
RJ		 6446(define-public r-bookdown
6447 (package
6448 (name "r-bookdown")
6bd27981 6449 (version "0.5")
2abfc5b8
RJ		 6450 (source (origin
6451 (method url-fetch)
6452 (uri (cran-uri "bookdown" version))
6453 (sha256
6454 (base32
6bd27981 6455 "0zm63kr4f4kja4qpwkzl119zzyciqj7ihajfqgfjpgb4dzaiycxp"))))
2abfc5b8
RJ		 6456 (build-system r-build-system)
6457 (propagated-inputs
6458 `(("r-htmltools" ,r-htmltools)
6459 ("r-knitr" ,r-knitr)
6460 ("r-rmarkdown" ,r-rmarkdown)
6461 ("r-yaml" ,r-yaml)))
6462 (home-page "https://github.com/rstudio/bookdown")
6463 (synopsis "Authoring books and technical documents with R markdown")
6464 (description "This package provides output formats and utilities for
6465authoring books and technical documents with R Markdown.")
6466 (license license:gpl3)))
6467
99df12cd
RJ		 6468(define-public r-biocstyle
6469 (package
6470 (name "r-biocstyle")
2167eafc 6471 (version "2.6.0")
99df12cd
RJ		 6472 (source (origin
6473 (method url-fetch)
6474 (uri (bioconductor-uri "BiocStyle" version))
6475 (sha256
6476 (base32
2167eafc 6477 "05f2j9fx8s5gh4f8qkl6wcz32ghz04wxhqb3xxcn1bj24qd7x1x8"))))
99df12cd
RJ		 6478 (properties
6479 `((upstream-name . "BiocStyle")))
6480 (build-system r-build-system)
3bef24c9
RJ		 6481 (propagated-inputs
6482 `(("r-bookdown" ,r-bookdown)
6483 ("r-knitr" ,r-knitr)
6484 ("r-rmarkdown" ,r-rmarkdown)
6485 ("r-yaml" ,r-yaml)))
5713bbf1 6486 (home-page "https://bioconductor.org/packages/BiocStyle")
99df12cd
RJ		 6487 (synopsis "Bioconductor formatting styles")
6488 (description "This package provides standard formatting styles for
6489Bioconductor PDF and HTML documents. Package vignettes illustrate use and
6490functionality.")
6491 (license license:artistic2.0)))
6492
4644644a
RJ		 6493(define-public r-bioccheck
6494 (package
6495 (name "r-bioccheck")
7373b416 6496 (version "1.14.0")
4644644a
RJ		 6497 (source (origin
6498 (method url-fetch)
6499 (uri (bioconductor-uri "BiocCheck" version))
6500 (sha256
6501 (base32
7373b416 6502 "1nzp8kgw13z9pgf885rplj6k37jcldfhbz0adqclxr2gq0yalmyx"))))
4644644a
RJ		 6503 (properties
6504 `((upstream-name . "BiocCheck")))
6505 (build-system r-build-system)
6506 (arguments
6507 '(#:phases
6508 (modify-phases %standard-phases
6509 ;; This package can be used by calling BiocCheck(<package>) from
6510 ;; within R, or by running R CMD BiocCheck <package>. This phase
6511 ;; makes sure the latter works. For this to work, the BiocCheck
6512 ;; script must be somewhere on the PATH (not the R bin directory).
6513 (add-after 'install 'install-bioccheck-subcommand
6514 (lambda* (#:key outputs #:allow-other-keys)
6515 (let* ((out (assoc-ref outputs "out"))
6516 (dest-dir (string-append out "/bin"))
6517 (script-dir
6518 (string-append out "/site-library/BiocCheck/script/")))
6519 (mkdir-p dest-dir)
6520 (symlink (string-append script-dir "/checkBadDeps.R")
6521 (string-append dest-dir "/checkBadDeps.R"))
6522 (symlink (string-append script-dir "/BiocCheck")
6523 (string-append dest-dir "/BiocCheck")))
6524 #t)))))
4644644a 6525 (propagated-inputs
aeb64f3c
RW		 6526 `(("r-codetools" ,r-codetools)
6527 ("r-graph" ,r-graph)
4644644a
RJ		 6528 ("r-httr" ,r-httr)
6529 ("r-optparse" ,r-optparse)
4644644a 6530 ("r-biocinstaller" ,r-biocinstaller)
7373b416
RW		 6531 ("r-biocviews" ,r-biocviews)
6532 ("r-stringdist" ,r-stringdist)))
5713bbf1 6533 (home-page "https://bioconductor.org/packages/BiocCheck")
4644644a
RJ		 6534 (synopsis "Executes Bioconductor-specific package checks")
6535 (description "This package contains tools to perform additional quality
6536checks on R packages that are to be submitted to the Bioconductor repository.")
6537 (license license:artistic2.0)))
6538
2acaaee5
RJ		 6539(define-public r-getopt
6540 (package
6541 (name "r-getopt")
6542 (version "1.20.0")
6543 (source
6544 (origin
6545 (method url-fetch)
6546 (uri (cran-uri "getopt" version))
6547 (sha256
6548 (base32
6549 "00f57vgnzmg7cz80rjmjz1556xqcmx8nhrlbbhaq4w7gl2ibl87r"))))
6550 (build-system r-build-system)
6551 (home-page "https://github.com/trevorld/getopt")
6552 (synopsis "Command-line option processor for R")
6553 (description
6554 "This package is designed to be used with Rscript to write shebang
6555scripts that accept short and long options. Many users will prefer to
6556use the packages @code{optparse} or @code{argparse} which add extra
6557features like automatically generated help options and usage texts,
6558support for default values, positional argument support, etc.")
6559 (license license:gpl2+)))
6560
c79ad57a
RJ		 6561(define-public r-optparse
6562 (package
6563 (name "r-optparse")
7150f1c3 6564 (version "1.4.4")
c79ad57a
RJ		 6565 (source
6566 (origin
6567 (method url-fetch)
6568 (uri (cran-uri "optparse" version))
6569 (sha256
6570 (base32
7150f1c3 6571 "1ff4wmsszrb3spwfp7ynfs8w11qpy1sdzfxm1wk8dqqvdwris7qb"))))
c79ad57a
RJ		 6572 (build-system r-build-system)
6573 (propagated-inputs
6574 `(("r-getopt" ,r-getopt)))
6575 (home-page
6576 "https://github.com/trevorld/optparse")
6577 (synopsis "Command line option parser")
6578 (description
6579 "This package provides a command line parser inspired by Python's
6580@code{optparse} library to be used with Rscript to write shebang scripts
6581that accept short and long options.")
6582 (license license:gpl2+)))
6583
247d498a
RJ		 6584(define-public r-dnacopy
6585 (package
6586 (name "r-dnacopy")
94fd86e5 6587 (version "1.52.0")
247d498a
RJ		 6588 (source (origin
6589 (method url-fetch)
6590 (uri (bioconductor-uri "DNAcopy" version))
6591 (sha256
6592 (base32
94fd86e5 6593 "127il5rlg1hzjlhwhs64x3nm18p00q1pd9ckb2b9ifl0rax95wai"))))
247d498a
RJ		 6594 (properties
6595 `((upstream-name . "DNAcopy")))
6596 (build-system r-build-system)
6597 (inputs
6598 `(("gfortran" ,gfortran)))
5697fdc3 6599 (home-page "https://bioconductor.org/packages/DNAcopy")
247d498a
RJ		 6600 (synopsis "Implementation of a circular binary segmentation algorithm")
6601 (description "This package implements the circular binary segmentation (CBS)
6602algorithm to segment DNA copy number data and identify genomic regions with
6603abnormal copy number.")
6604 (license license:gpl2+)))
6605
7485129e
RW		 6606(define-public r-s4vectors
6607 (package
6608 (name "r-s4vectors")
41f0f949 6609 (version "0.16.0")
7485129e
RW		 6610 (source (origin
6611 (method url-fetch)
6612 (uri (bioconductor-uri "S4Vectors" version))
6613 (sha256
6614 (base32
41f0f949 6615 "03s8vz33nl6mivjb7dbvj702dkypi340lji1sjban03fyyls0hw0"))))
7485129e 6616 (properties
1d216b6e 6617 `((upstream-name . "S4Vectors")))
7485129e
RW		 6618 (build-system r-build-system)
6619 (propagated-inputs
6620 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6621 (home-page "https://bioconductor.org/packages/S4Vectors")
7485129e
RW		 6622 (synopsis "S4 implementation of vectors and lists")
6623 (description
6624 "The S4Vectors package defines the @code{Vector} and @code{List} virtual
6625classes and a set of generic functions that extend the semantic of ordinary
6626vectors and lists in R. Package developers can easily implement vector-like
6627or list-like objects as concrete subclasses of @code{Vector} or @code{List}.
6628In addition, a few low-level concrete subclasses of general interest (e.g.
6629@code{DataFrame}, @code{Rle}, and @code{Hits}) are implemented in the
6630S4Vectors package itself.")
6631 (license license:artistic2.0)))
6632
274da826
RW		 6633(define-public r-seqinr
6634 (package
6635 (name "r-seqinr")
023aa8ff 6636 (version "3.4-5")
274da826
RW		 6637 (source
6638 (origin
6639 (method url-fetch)
6640 (uri (cran-uri "seqinr" version))
6641 (sha256
6642 (base32
023aa8ff 6643 "17zv0n5cji17izwmwg0jcbxbjl3w5rls91w15svcnlpxjms38ahn"))))
274da826
RW		 6644 (build-system r-build-system)
6645 (propagated-inputs
3b851cd4
RW		 6646 `(("r-ade4" ,r-ade4)
6647 ("r-segmented" ,r-segmented)))
274da826
RW		 6648 (inputs
6649 `(("zlib" ,zlib)))
6650 (home-page "http://seqinr.r-forge.r-project.org/")
6651 (synopsis "Biological sequences retrieval and analysis")
6652 (description
6653 "This package provides tools for exploratory data analysis and data
6654visualization of biological sequence (DNA and protein) data. It also includes
6655utilities for sequence data management under the ACNUC system.")
6656 (license license:gpl2+)))
6657
78addcb0
RW		 6658(define-public r-iranges
6659 (package
6660 (name "r-iranges")
9e482c20 6661 (version "2.12.0")
78addcb0
RW		 6662 (source (origin
6663 (method url-fetch)
6664 (uri (bioconductor-uri "IRanges" version))
6665 (sha256
6666 (base32
9e482c20 6667 "1vqczb9wlxsmpwpqig6j1dmiblcfpq6mgnq8qwzcrvddm4cp47m5"))))
78addcb0 6668 (properties
1d216b6e 6669 `((upstream-name . "IRanges")))
78addcb0
RW		 6670 (build-system r-build-system)
6671 (propagated-inputs
6672 `(("r-biocgenerics" ,r-biocgenerics)
6673 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6674 (home-page "https://bioconductor.org/packages/IRanges")
78addcb0
RW		 6675 (synopsis "Infrastructure for manipulating intervals on sequences")
6676 (description
6677 "This package provides efficient low-level and highly reusable S4 classes
6678for storing ranges of integers, RLE vectors (Run-Length Encoding), and, more
6679generally, data that can be organized sequentially (formally defined as
6680@code{Vector} objects), as well as views on these @code{Vector} objects.
6681Efficient list-like classes are also provided for storing big collections of
6682instances of the basic classes. All classes in the package use consistent
6683naming and share the same rich and consistent \"Vector API\" as much as
6684possible.")
6685 (license license:artistic2.0)))
6686
ffef27f3
RJ		 6687(define-public r-genomeinfodbdata
6688 (package
6689 (name "r-genomeinfodbdata")
6690 (version "0.99.0")
6691 (source (origin
6692 (method url-fetch)
90f83099
EF		 6693 ;; We cannot use bioconductor-uri here because this tarball is
6694 ;; located under "data/annotation/" instead of "bioc/".
6695 (uri (string-append "https://bioconductor.org/packages/release/"
6696 "data/annotation/src/contrib/GenomeInfoDbData_"
6697 version ".tar.gz"))
ffef27f3
RJ		 6698 (sha256
6699 (base32
6700 "120qvhb0pvkzd65lsgja62vyrgc37si6fh68q4cg4w5x9f04jw25"))))
6701 (properties
6702 `((upstream-name . "GenomeInfoDbData")))
6703 (build-system r-build-system)
5713bbf1 6704 (home-page "https://bioconductor.org/packages/GenomeInfoDbData")
ffef27f3
RJ		 6705 (synopsis "Species and taxonomy ID look up tables for GenomeInfoDb")
6706 (description "This package contains data for mapping between NCBI taxonomy
6707ID and species. It is used by functions in the GenomeInfoDb package.")
6708 (license license:artistic2.0)))
6709
bf7764b7
RW		 6710(define-public r-genomeinfodb
6711 (package
6712 (name "r-genomeinfodb")
8aab0235 6713 (version "1.14.0")
bf7764b7
RW		 6714 (source (origin
6715 (method url-fetch)
6716 (uri (bioconductor-uri "GenomeInfoDb" version))
6717 (sha256
6718 (base32
8aab0235 6719 "1jhm0imkac4gvchbjxj408aakk39xdv2fyh818d3lk295bz6bnyp"))))
bf7764b7 6720 (properties
1d216b6e 6721 `((upstream-name . "GenomeInfoDb")))
bf7764b7
RW		 6722 (build-system r-build-system)
6723 (propagated-inputs
6724 `(("r-biocgenerics" ,r-biocgenerics)
38b99ccc 6725 ("r-genomeinfodbdata" ,r-genomeinfodbdata)
bf7764b7 6726 ("r-iranges" ,r-iranges)
4cd07e48 6727 ("r-rcurl" ,r-rcurl)
bf7764b7 6728 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6729 (home-page "https://bioconductor.org/packages/GenomeInfoDb")
bf7764b7
RW		 6730 (synopsis "Utilities for manipulating chromosome identifiers")
6731 (description
6732 "This package contains data and functions that define and allow
6733translation between different chromosome sequence naming conventions (e.g.,
6734\"chr1\" versus \"1\"), including a function that attempts to place sequence
6735names in their natural, rather than lexicographic, order.")
6736 (license license:artistic2.0)))
6737
744004a3
RJ		 6738(define-public r-edger
6739 (package
6740 (name "r-edger")
010ab2ff 6741 (version "3.20.1")
744004a3
RJ		 6742 (source (origin
6743 (method url-fetch)
6744 (uri (bioconductor-uri "edgeR" version))
6745 (sha256
6746 (base32
010ab2ff 6747 "01qnxwr9rmz8r5ga3hvjk632365ga2aygx71mxkk7jiad2pjznsp"))))
744004a3
RJ		 6748 (properties `((upstream-name . "edgeR")))
6749 (build-system r-build-system)
6750 (propagated-inputs
5e48005f 6751 `(("r-limma" ,r-limma)
47055b27 6752 ("r-locfit" ,r-locfit)
010ab2ff 6753 ("r-rcpp" ,r-rcpp)
47055b27 6754 ("r-statmod" ,r-statmod))) ;for estimateDisp
744004a3
RJ		 6755 (home-page "http://bioinf.wehi.edu.au/edgeR")
6756 (synopsis "EdgeR does empirical analysis of digital gene expression data")
6757 (description "This package can do differential expression analysis of
6758RNA-seq expression profiles with biological replication. It implements a range
6759of statistical methodology based on the negative binomial distributions,
6760including empirical Bayes estimation, exact tests, generalized linear models
6761and quasi-likelihood tests. It be applied to differential signal analysis of
6762other types of genomic data that produce counts, including ChIP-seq, SAGE and
6763CAGE.")
6764 (license license:gpl2+)))
6765
b669d9c4
RJ		 6766(define-public r-variantannotation
6767 (package
6768 (name "r-variantannotation")
7f1bac35 6769 (version "1.24.0")
b669d9c4
RJ		 6770 (source (origin
6771 (method url-fetch)
6772 (uri (bioconductor-uri "VariantAnnotation" version))
6773 (sha256
6774 (base32
7f1bac35 6775 "1lllp2vgyfbrar1yg28ji7am470hfzrzxm1bgdk68xpnrwcgcl25"))))
b669d9c4
RJ		 6776 (properties
6777 `((upstream-name . "VariantAnnotation")))
6778 (inputs
6779 `(("zlib" ,zlib)))
6780 (propagated-inputs
6781 `(("r-annotationdbi" ,r-annotationdbi)
37d96f1d 6782 ("r-biobase" ,r-biobase)
b669d9c4 6783 ("r-biocgenerics" ,r-biocgenerics)
37d96f1d 6784 ("r-biostrings" ,r-biostrings)
b669d9c4
RJ		 6785 ("r-bsgenome" ,r-bsgenome)
6786 ("r-dbi" ,r-dbi)
6787 ("r-genomeinfodb" ,r-genomeinfodb)
6788 ("r-genomicfeatures" ,r-genomicfeatures)
6789 ("r-genomicranges" ,r-genomicranges)
37d96f1d 6790 ("r-iranges" ,r-iranges)
b669d9c4
RJ		 6791 ("r-summarizedexperiment" ,r-summarizedexperiment)
6792 ("r-rsamtools" ,r-rsamtools)
37d96f1d
RW		 6793 ("r-rtracklayer" ,r-rtracklayer)
6794 ("r-s4vectors" ,r-s4vectors)
6795 ("r-xvector" ,r-xvector)
b669d9c4
RJ		 6796 ("r-zlibbioc" ,r-zlibbioc)))
6797 (build-system r-build-system)
6798 (home-page "https://bioconductor.org/packages/VariantAnnotation")
6799 (synopsis "Package for annotation of genetic variants")
6800 (description "This R package can annotate variants, compute amino acid
6801coding changes and predict coding outcomes.")
6802 (license license:artistic2.0)))
6803
7d4224d7
RJ		 6804(define-public r-limma
6805 (package
6806 (name "r-limma")
f642c0e4 6807 (version "3.34.0")
7d4224d7
RJ		 6808 (source (origin
6809 (method url-fetch)
6810 (uri (bioconductor-uri "limma" version))
6811 (sha256
6812 (base32
f642c0e4 6813 "0a15gsaky0hfrkx8wrrmp0labzxpq6m2hrd33zl206wyas8bqzcs"))))
7d4224d7
RJ		 6814 (build-system r-build-system)
6815 (home-page "http://bioinf.wehi.edu.au/limma")
6816 (synopsis "Package for linear models for microarray and RNA-seq data")
6817 (description "This package can be used for the analysis of gene expression
6818studies, especially the use of linear models for analysing designed experiments
6819and the assessment of differential expression. The analysis methods apply to
6820different technologies, including microarrays, RNA-seq, and quantitative PCR.")
6821 (license license:gpl2+)))
6822
0e7d5560
RW		 6823(define-public r-xvector
6824 (package
6825 (name "r-xvector")
2b6ae8bf 6826 (version "0.18.0")
0e7d5560
RW		 6827 (source (origin
6828 (method url-fetch)
6829 (uri (bioconductor-uri "XVector" version))
6830 (sha256
6831 (base32
2b6ae8bf 6832 "1i4i3kdxr78lr1kcxq657p11ybi7kq10c8kyaqyh6gfc8i9rhvmk"))))
0e7d5560 6833 (properties
1d216b6e 6834 `((upstream-name . "XVector")))
0e7d5560
RW		 6835 (build-system r-build-system)
6836 (arguments
6837 `(#:phases
6838 (modify-phases %standard-phases
6839 (add-after 'unpack 'use-system-zlib
6840 (lambda _
6841 (substitute* "DESCRIPTION"
6842 (("zlibbioc, ") ""))
6843 (substitute* "NAMESPACE"
6844 (("import\\(zlibbioc\\)") ""))
6845 #t)))))
6846 (inputs
6847 `(("zlib" ,zlib)))
6848 (propagated-inputs
6849 `(("r-biocgenerics" ,r-biocgenerics)
6850 ("r-iranges" ,r-iranges)
6851 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6852 (home-page "https://bioconductor.org/packages/XVector")
0e7d5560
RW		 6853 (synopsis "Representation and manpulation of external sequences")
6854 (description
6855 "This package provides memory efficient S4 classes for storing sequences
6856\"externally\" (behind an R external pointer, or on disk).")
6857 (license license:artistic2.0)))
6858
e2cd1d0f
RW		 6859(define-public r-genomicranges
6860 (package
6861 (name "r-genomicranges")
aa5c1f61 6862 (version "1.30.0")
e2cd1d0f
RW		 6863 (source (origin
6864 (method url-fetch)
6865 (uri (bioconductor-uri "GenomicRanges" version))
6866 (sha256
6867 (base32
aa5c1f61 6868 "10ra2sjn17h6gilm9iz0cygp9ijpgbirljlc4drwrnivnw9cmi2a"))))
e2cd1d0f 6869 (properties
1d216b6e 6870 `((upstream-name . "GenomicRanges")))
e2cd1d0f
RW		 6871 (build-system r-build-system)
6872 (propagated-inputs
6873 `(("r-biocgenerics" ,r-biocgenerics)
6874 ("r-genomeinfodb" ,r-genomeinfodb)
92a740af
RW		 6875 ("r-iranges" ,r-iranges)
6876 ("r-s4vectors" ,r-s4vectors)
e2cd1d0f 6877 ("r-xvector" ,r-xvector)))
5713bbf1 6878 (home-page "https://bioconductor.org/packages/GenomicRanges")
e2cd1d0f
RW		 6879 (synopsis "Representation and manipulation of genomic intervals")
6880 (description
6881 "This package provides tools to efficiently represent and manipulate
6882genomic annotations and alignments is playing a central role when it comes to
6883analyzing high-throughput sequencing data (a.k.a. NGS data). The
6884GenomicRanges package defines general purpose containers for storing and
6885manipulating genomic intervals and variables defined along a genome.")
6886 (license license:artistic2.0)))
6887
555e3399
RW		 6888(define-public r-biobase
6889 (package
6890 (name "r-biobase")
ca521236 6891 (version "2.38.0")
555e3399
RW		 6892 (source (origin
6893 (method url-fetch)
6894 (uri (bioconductor-uri "Biobase" version))
6895 (sha256
6896 (base32
ca521236 6897 "1cgm1ja1kp56zdlzyy9ggbkfn8r2vbsd4hncmz8g4hjd47fg18kg"))))
555e3399
RW		 6898 (properties
6899 `((upstream-name . "Biobase")))
6900 (build-system r-build-system)
6901 (propagated-inputs
6902 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6903 (home-page "https://bioconductor.org/packages/Biobase")
555e3399
RW		 6904 (synopsis "Base functions for Bioconductor")
6905 (description
6906 "This package provides functions that are needed by many other packages
6907on Bioconductor or which replace R functions.")
6908 (license license:artistic2.0)))
6909
8b7bce74
RW		 6910(define-public r-annotationdbi
6911 (package
6912 (name "r-annotationdbi")
0f8d98f2 6913 (version "1.40.0")
8b7bce74
RW		 6914 (source (origin
6915 (method url-fetch)
6916 (uri (bioconductor-uri "AnnotationDbi" version))
6917 (sha256
6918 (base32
0f8d98f2 6919 "1dh4qs1a757n640gs34lf6z2glc96nan86x0sqaw5csadl2rhnlc"))))
8b7bce74
RW		 6920 (properties
6921 `((upstream-name . "AnnotationDbi")))
6922 (build-system r-build-system)
6923 (propagated-inputs
6924 `(("r-biobase" ,r-biobase)
6925 ("r-biocgenerics" ,r-biocgenerics)
6926 ("r-dbi" ,r-dbi)
6927 ("r-iranges" ,r-iranges)
6928 ("r-rsqlite" ,r-rsqlite)
6929 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6930 (home-page "https://bioconductor.org/packages/AnnotationDbi")
8b7bce74
RW		 6931 (synopsis "Annotation database interface")
6932 (description
6933 "This package provides user interface and database connection code for
6934annotation data packages using SQLite data storage.")
6935 (license license:artistic2.0)))
6936
c465fa72
RW		 6937(define-public r-biomart
6938 (package
6939 (name "r-biomart")
42e11d33 6940 (version "2.34.0")
c465fa72
RW		 6941 (source (origin
6942 (method url-fetch)
6943 (uri (bioconductor-uri "biomaRt" version))
6944 (sha256
6945 (base32
42e11d33 6946 "1dn3ysf0vb3mmg2b3380g0j1ajf88x4rh7fddfp990h2xlnsy2cx"))))
c465fa72
RW		 6947 (properties
6948 `((upstream-name . "biomaRt")))
6949 (build-system r-build-system)
6950 (propagated-inputs
6951 `(("r-annotationdbi" ,r-annotationdbi)
42e11d33 6952 ("r-progress" ,r-progress)
c465fa72 6953 ("r-rcurl" ,r-rcurl)
42e11d33 6954 ("r-stringr" ,r-stringr)
c465fa72 6955 ("r-xml" ,r-xml)))
5713bbf1 6956 (home-page "https://bioconductor.org/packages/biomaRt")
c465fa72
RW		 6957 (synopsis "Interface to BioMart databases")
6958 (description
6959 "biomaRt provides an interface to a growing collection of databases
6960implementing the @url{BioMart software suite, http://www.biomart.org}. The
6961package enables retrieval of large amounts of data in a uniform way without
6962the need to know the underlying database schemas or write complex SQL queries.
6963Examples of BioMart databases are Ensembl, COSMIC, Uniprot, HGNC, Gramene,
6964Wormbase and dbSNP mapped to Ensembl. These major databases give biomaRt
6965users direct access to a diverse set of data and enable a wide range of
6966powerful online queries from gene annotation to database mining.")
6967 (license license:artistic2.0)))
6968
e91d362e
RW		 6969(define-public r-biocparallel
6970 (package
6971 (name "r-biocparallel")
a044c7f4 6972 (version "1.12.0")
e91d362e
RW		 6973 (source (origin
6974 (method url-fetch)
6975 (uri (bioconductor-uri "BiocParallel" version))
6976 (sha256
6977 (base32
a044c7f4 6978 "13ng3n2wsgl3fh0v6jnz3vg51k5c1sh44pqdvblcrcd1qyjmmqhd"))))
e91d362e
RW		 6979 (properties
6980 `((upstream-name . "BiocParallel")))
6981 (build-system r-build-system)
6982 (propagated-inputs
6983 `(("r-futile-logger" ,r-futile-logger)
a044c7f4
RW		 6984 ("r-snow" ,r-snow)
6985 ("r-bh" ,r-bh)))
5713bbf1 6986 (home-page "https://bioconductor.org/packages/BiocParallel")
e91d362e
RW		 6987 (synopsis "Bioconductor facilities for parallel evaluation")
6988 (description
6989 "This package provides modified versions and novel implementation of
6990functions for parallel evaluation, tailored to use with Bioconductor
6991objects.")
6992 (license (list license:gpl2+ license:gpl3+))))
6993
bf159353
RW		 6994(define-public r-biostrings
6995 (package
6996 (name "r-biostrings")
b719435e 6997 (version "2.46.0")
bf159353
RW		 6998 (source (origin
6999 (method url-fetch)
7000 (uri (bioconductor-uri "Biostrings" version))
7001 (sha256
7002 (base32
b719435e 7003 "0vg50qdlxqcm2d6axjnzg8wh8pr4c5gz03l8bdl0llmwzp0zclzk"))))
bf159353
RW		 7004 (properties
7005 `((upstream-name . "Biostrings")))
7006 (build-system r-build-system)
7007 (propagated-inputs
7008 `(("r-biocgenerics" ,r-biocgenerics)
7009 ("r-iranges" ,r-iranges)
7010 ("r-s4vectors" ,r-s4vectors)
7011 ("r-xvector" ,r-xvector)))
5713bbf1 7012 (home-page "https://bioconductor.org/packages/Biostrings")
bf159353
RW		 7013 (synopsis "String objects and algorithms for biological sequences")
7014 (description
7015 "This package provides memory efficient string containers, string
7016matching algorithms, and other utilities, for fast manipulation of large
7017biological sequences or sets of sequences.")
7018 (license license:artistic2.0)))
7019
f8d74f70
RW		 7020(define-public r-rsamtools
7021 (package
7022 (name "r-rsamtools")
f8068419 7023 (version "1.30.0")
f8d74f70
RW		 7024 (source (origin
7025 (method url-fetch)
7026 (uri (bioconductor-uri "Rsamtools" version))
7027 (sha256
7028 (base32
f8068419 7029 "0pjny5fjvbnfdyhl3bwxin678sha2drvs00sivxh3l772cn6yams"))))
f8d74f70
RW		 7030 (properties
7031 `((upstream-name . "Rsamtools")))
7032 (build-system r-build-system)
7033 (arguments
7034 `(#:phases
7035 (modify-phases %standard-phases
7036 (add-after 'unpack 'use-system-zlib
7037 (lambda _
7038 (substitute* "DESCRIPTION"
7039 (("zlibbioc, ") ""))
7040 (substitute* "NAMESPACE"
7041 (("import\\(zlibbioc\\)") ""))
7042 #t)))))
7043 (inputs
7044 `(("zlib" ,zlib)))
7045 (propagated-inputs
7046 `(("r-biocgenerics" ,r-biocgenerics)
7047 ("r-biocparallel" ,r-biocparallel)
7048 ("r-biostrings" ,r-biostrings)
7049 ("r-bitops" ,r-bitops)
7050 ("r-genomeinfodb" ,r-genomeinfodb)
7051 ("r-genomicranges" ,r-genomicranges)
7052 ("r-iranges" ,r-iranges)
7053 ("r-s4vectors" ,r-s4vectors)
7054 ("r-xvector" ,r-xvector)))
5713bbf1 7055 (home-page "https://bioconductor.org/packages/release/bioc/html/Rsamtools.html")
f8d74f70
RW		 7056 (synopsis "Interface to samtools, bcftools, and tabix")
7057 (description
7058 "This package provides an interface to the 'samtools', 'bcftools', and
7059'tabix' utilities for manipulating SAM (Sequence Alignment / Map), FASTA,
7060binary variant call (BCF) and compressed indexed tab-delimited (tabix)
7061files.")
7062 (license license:expat)))
7063
71e34e6b
RJ		 7064(define-public r-delayedarray
7065 (package
7066 (name "r-delayedarray")
cb0a9a74 7067 (version "0.4.1")
71e34e6b
RJ		 7068 (source (origin
7069 (method url-fetch)
7070 (uri (bioconductor-uri "DelayedArray" version))
7071 (sha256
7072 (base32
cb0a9a74 7073 "0s7h2giyvz04cg6248kbbzpwhxdrpnsvl2s8k5c8ricisd9aaz4b"))))
71e34e6b
RJ		 7074 (properties
7075 `((upstream-name . "DelayedArray")))
7076 (build-system r-build-system)
7077 (propagated-inputs
7078 `(("r-biocgenerics" ,r-biocgenerics)
7079 ("r-s4vectors" ,r-s4vectors)
7080 ("r-iranges" ,r-iranges)
7081 ("r-matrixstats" ,r-matrixstats)))
5713bbf1 7082 (home-page "https://bioconductor.org/packages/DelayedArray")
71e34e6b
RJ		 7083 (synopsis "Delayed operations on array-like objects")
7084 (description
7085 "Wrapping an array-like object (typically an on-disk object) in a
7086@code{DelayedArray} object allows one to perform common array operations on it
7087without loading the object in memory. In order to reduce memory usage and
7088optimize performance, operations on the object are either delayed or executed
7089using a block processing mechanism. Note that this also works on in-memory
7090array-like objects like @code{DataFrame} objects (typically with Rle columns),
7091@code{Matrix} objects, and ordinary arrays and data frames.")
7092 (license license:artistic2.0)))
7093
6e76dda2
RW		 7094(define-public r-summarizedexperiment
7095 (package
7096 (name "r-summarizedexperiment")
6fcbed73 7097 (version "1.8.0")
6e76dda2
RW		 7098 (source (origin
7099 (method url-fetch)
7100 (uri (bioconductor-uri "SummarizedExperiment" version))
7101 (sha256
7102 (base32
6fcbed73 7103 "1011r8l0k8420j31bmh4xdcp6ka5bzf4bqhip84v5b6alpkcbvmf"))))
6e76dda2
RW		 7104 (properties
7105 `((upstream-name . "SummarizedExperiment")))
7106 (build-system r-build-system)
7107 (propagated-inputs
7108 `(("r-biobase" ,r-biobase)
7109 ("r-biocgenerics" ,r-biocgenerics)
d006ee31 7110 ("r-delayedarray" ,r-delayedarray)
6e76dda2
RW		 7111 ("r-genomeinfodb" ,r-genomeinfodb)
7112 ("r-genomicranges" ,r-genomicranges)
7113 ("r-iranges" ,r-iranges)
aeb64f3c 7114 ("r-matrix" ,r-matrix)
6e76dda2 7115 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 7116 (home-page "https://bioconductor.org/packages/SummarizedExperiment")
6e76dda2
RW		 7117 (synopsis "Container for representing genomic ranges by sample")
7118 (description
7119 "The SummarizedExperiment container contains one or more assays, each
7120represented by a matrix-like object of numeric or other mode. The rows
7121typically represent genomic ranges of interest and the columns represent
7122samples.")
7123 (license license:artistic2.0)))
7124
d8a828af
RW		 7125(define-public r-genomicalignments
7126 (package
7127 (name "r-genomicalignments")
96fb4651 7128 (version "1.14.0")
d8a828af
RW		 7129 (source (origin
7130 (method url-fetch)
7131 (uri (bioconductor-uri "GenomicAlignments" version))
7132 (sha256
7133 (base32
96fb4651 7134 "0sw30lj11wv7ifzypqm04lcah987crqwvj48wz3flaw3biw41zfi"))))
d8a828af
RW		 7135 (properties
7136 `((upstream-name . "GenomicAlignments")))
7137 (build-system r-build-system)
7138 (propagated-inputs
7139 `(("r-biocgenerics" ,r-biocgenerics)
7140 ("r-biocparallel" ,r-biocparallel)
7141 ("r-biostrings" ,r-biostrings)
7142 ("r-genomeinfodb" ,r-genomeinfodb)
7143 ("r-genomicranges" ,r-genomicranges)
7144 ("r-iranges" ,r-iranges)
7145 ("r-rsamtools" ,r-rsamtools)
7146 ("r-s4vectors" ,r-s4vectors)
7147 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 7148 (home-page "https://bioconductor.org/packages/GenomicAlignments")
d8a828af
RW		 7149 (synopsis "Representation and manipulation of short genomic alignments")
7150 (description
7151 "This package provides efficient containers for storing and manipulating
7152short genomic alignments (typically obtained by aligning short reads to a
7153reference genome). This includes read counting, computing the coverage,
7154junction detection, and working with the nucleotide content of the
7155alignments.")
7156 (license license:artistic2.0)))
7157
317755ff
RW		 7158(define-public r-rtracklayer
7159 (package
7160 (name "r-rtracklayer")
4dd469ef 7161 (version "1.38.0")
317755ff
RW		 7162 (source (origin
7163 (method url-fetch)
7164 (uri (bioconductor-uri "rtracklayer" version))
7165 (sha256
7166 (base32
4dd469ef 7167 "12al1ygzy9p4myxa1fd817m28x2fj6f863znk9bw3hp7knbi98dh"))))
317755ff
RW		 7168 (build-system r-build-system)
7169 (arguments
7170 `(#:phases
7171 (modify-phases %standard-phases
7172 (add-after 'unpack 'use-system-zlib
7173 (lambda _
7174 (substitute* "DESCRIPTION"
4dd469ef 7175 ((" zlibbioc,") ""))
317755ff
RW		 7176 (substitute* "NAMESPACE"
7177 (("import\\(zlibbioc\\)") ""))
7178 #t)))))
7179 (inputs
7180 `(("zlib" ,zlib)))
7181 (propagated-inputs
7182 `(("r-biocgenerics" ,r-biocgenerics)
7183 ("r-biostrings" ,r-biostrings)
7184 ("r-genomeinfodb" ,r-genomeinfodb)
7185 ("r-genomicalignments" ,r-genomicalignments)
7186 ("r-genomicranges" ,r-genomicranges)
7187 ("r-iranges" ,r-iranges)
7188 ("r-rcurl" ,r-rcurl)
7189 ("r-rsamtools" ,r-rsamtools)
7190 ("r-s4vectors" ,r-s4vectors)
7191 ("r-xml" ,r-xml)
7192 ("r-xvector" ,r-xvector)))
5713bbf1 7193 (home-page "https://bioconductor.org/packages/rtracklayer")
317755ff
RW		 7194 (synopsis "R interface to genome browsers and their annotation tracks")
7195 (description
7196 "rtracklayer is an extensible framework for interacting with multiple
7197genome browsers (currently UCSC built-in) and manipulating annotation tracks
7198in various formats (currently GFF, BED, bedGraph, BED15, WIG, BigWig and 2bit
7199built-in). The user may export/import tracks to/from the supported browsers,
7200as well as query and modify the browser state, such as the current viewport.")
7201 (license license:artistic2.0)))
7202
2fd7c049
RW		 7203(define-public r-genomicfeatures
7204 (package
7205 (name "r-genomicfeatures")
158b6743 7206 (version "1.30.0")
2fd7c049
RW		 7207 (source (origin
7208 (method url-fetch)
7209 (uri (bioconductor-uri "GenomicFeatures" version))
7210 (sha256
7211 (base32
158b6743 7212 "1khjvq1ffhqavkwf8n7bilknci60lxbg52icrcf2vnb9k8rlpghs"))))
2fd7c049
RW		 7213 (properties
7214 `((upstream-name . "GenomicFeatures")))
7215 (build-system r-build-system)
7216 (propagated-inputs
7217 `(("r-annotationdbi" ,r-annotationdbi)
7218 ("r-biobase" ,r-biobase)
7219 ("r-biocgenerics" ,r-biocgenerics)
7220 ("r-biomart" ,r-biomart)
7221 ("r-biostrings" ,r-biostrings)
7222 ("r-dbi" ,r-dbi)
7223 ("r-genomeinfodb" ,r-genomeinfodb)
7224 ("r-genomicranges" ,r-genomicranges)
7225 ("r-iranges" ,r-iranges)
7226 ("r-rcurl" ,r-rcurl)
7227 ("r-rsqlite" ,r-rsqlite)
158b6743 7228 ("r-rmysql" ,r-rmysql)
2fd7c049
RW		 7229 ("r-rtracklayer" ,r-rtracklayer)
7230 ("r-s4vectors" ,r-s4vectors)
7231 ("r-xvector" ,r-xvector)))
5713bbf1 7232 (home-page "https://bioconductor.org/packages/GenomicFeatures")
2fd7c049
RW		 7233 (synopsis "Tools for working with transcript centric annotations")
7234 (description
7235 "This package provides a set of tools and methods for making and
7236manipulating transcript centric annotations. With these tools the user can
7237easily download the genomic locations of the transcripts, exons and cds of a
7238given organism, from either the UCSC Genome Browser or a BioMart
7239database (more sources will be supported in the future). This information is
7240then stored in a local database that keeps track of the relationship between
7241transcripts, exons, cds and genes. Flexible methods are provided for
7242extracting the desired features in a convenient format.")
7243 (license license:artistic2.0)))
7244
fb25d880
RW		 7245(define-public r-go-db
7246 (package
7247 (name "r-go-db")
d1b1587c 7248 (version "3.4.0")
fb25d880
RW		 7249 (source (origin
7250 (method url-fetch)
5713bbf1 7251 (uri (string-append "https://www.bioconductor.org/packages/"
f82c8c3c
PP		 7252 "release/data/annotation/src/contrib/GO.db_"
7253 version ".tar.gz"))
fb25d880
RW		 7254 (sha256
7255 (base32
d1b1587c 7256 "02cj8kqi5w39jwcs8gp1dgj08sah262ppxnkz4h3qd0w191y8yyl"))))
fb25d880
RW		 7257 (properties
7258 `((upstream-name . "GO.db")))
7259 (build-system r-build-system)
3141b83d
RW		 7260 (propagated-inputs
7261 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7262 (home-page "https://bioconductor.org/packages/GO.db")
fb25d880
RW		 7263 (synopsis "Annotation maps describing the entire Gene Ontology")
7264 (description
7265 "The purpose of this GO.db annotation package is to provide detailed
7266information about the latest version of the Gene Ontologies.")
7267 (license license:artistic2.0)))
7268
d1dbde6a
RW		 7269(define-public r-graph
7270 (package
7271 (name "r-graph")
aeb73879 7272 (version "1.56.0")
d1dbde6a
RW		 7273 (source (origin
7274 (method url-fetch)
7275 (uri (bioconductor-uri "graph" version))
7276 (sha256
7277 (base32
aeb73879 7278 "15aajjp8h2z14p80c8hyd4rrmr9vqsm7bvwb989jxjl4k6g52an1"))))
d1dbde6a
RW		 7279 (build-system r-build-system)
7280 (propagated-inputs
7281 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 7282 (home-page "https://bioconductor.org/packages/graph")
d1dbde6a
RW		 7283 (synopsis "Handle graph data structures in R")
7284 (description
7285 "This package implements some simple graph handling capabilities for R.")
7286 (license license:artistic2.0)))
7287
d547ce5e
RW		 7288(define-public r-topgo
7289 (package
7290 (name "r-topgo")
d13a3fea 7291 (version "2.30.0")
d547ce5e
RW		 7292 (source (origin
7293 (method url-fetch)
7294 (uri (bioconductor-uri "topGO" version))
7295 (sha256
7296 (base32
d13a3fea 7297 "1hqffz5qp7glxdvjp37005g8qk5nam3f9wpf6d1wjnzpar04f3dz"))))
d547ce5e
RW		 7298 (properties
7299 `((upstream-name . "topGO")))
7300 (build-system r-build-system)
7301 (propagated-inputs
7302 `(("r-annotationdbi" ,r-annotationdbi)
30ec4de7 7303 ("r-dbi" ,r-dbi)
d547ce5e
RW		 7304 ("r-biobase" ,r-biobase)
7305 ("r-biocgenerics" ,r-biocgenerics)
7306 ("r-go-db" ,r-go-db)
6d415db2 7307 ("r-graph" ,r-graph)
aeb64f3c
RW		 7308 ("r-lattice" ,r-lattice)
7309 ("r-matrixstats" ,r-matrixstats)
d547ce5e 7310 ("r-sparsem" ,r-sparsem)))
5713bbf1 7311 (home-page "https://bioconductor.org/packages/topGO")
d547ce5e
RW		 7312 (synopsis "Enrichment analysis for gene ontology")
7313 (description
7314 "The topGO package provides tools for testing @dfn{gene ontology} (GO)
7315terms while accounting for the topology of the GO graph. Different test
7316statistics and different methods for eliminating local similarities and
7317dependencies between GO terms can be implemented and applied.")
7318 ;; Any version of the LGPL applies.
7319 (license license:lgpl2.1+)))
7320
c63cef66
RW		 7321(define-public r-bsgenome
7322 (package
7323 (name "r-bsgenome")
e67850b4 7324 (version "1.46.0")
c63cef66
RW		 7325 (source (origin
7326 (method url-fetch)
7327 (uri (bioconductor-uri "BSgenome" version))
7328 (sha256
7329 (base32
e67850b4 7330 "1jbzq7lm2iajajn2bifxnkss0k9fdvgqr30mral17cbhp5f6w4lq"))))
c63cef66
RW		 7331 (properties
7332 `((upstream-name . "BSgenome")))
7333 (build-system r-build-system)
7334 (propagated-inputs
7335 `(("r-biocgenerics" ,r-biocgenerics)
7336 ("r-biostrings" ,r-biostrings)
7337 ("r-genomeinfodb" ,r-genomeinfodb)
7338 ("r-genomicranges" ,r-genomicranges)
7339 ("r-iranges" ,r-iranges)
7340 ("r-rsamtools" ,r-rsamtools)
7341 ("r-rtracklayer" ,r-rtracklayer)
7342 ("r-s4vectors" ,r-s4vectors)
7343 ("r-xvector" ,r-xvector)))
5713bbf1 7344 (home-page "https://bioconductor.org/packages/BSgenome")
c63cef66
RW		 7345 (synopsis "Infrastructure for Biostrings-based genome data packages")
7346 (description
7347 "This package provides infrastructure shared by all Biostrings-based
7348genome data packages and support for efficient SNP representation.")
7349 (license license:artistic2.0)))
7350
aa3eeeb5
RJ		 7351(define-public r-bsgenome-hsapiens-1000genomes-hs37d5
7352 (package
7353 (name "r-bsgenome-hsapiens-1000genomes-hs37d5")
7354 (version "0.99.1")
7355 (source (origin
7356 (method url-fetch)
7357 ;; We cannot use bioconductor-uri here because this tarball is
7358 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7359 (uri (string-append "https://www.bioconductor.org/packages/"
aa3eeeb5
RJ		 7360 "release/data/annotation/src/contrib/"
7361 "BSgenome.Hsapiens.1000genomes.hs37d5_"
7362 version ".tar.gz"))
7363 (sha256
7364 (base32
7365 "1cg0g5fqmsvwyw2p9hp2yy4ilk21jkbbrnpgqvb5c36ihjwvc7sr"))))
7366 (properties
7367 `((upstream-name . "BSgenome.Hsapiens.1000genomes.hs37d5")))
7368 (build-system r-build-system)
7369 ;; As this package provides little more than a very large data file it
7370 ;; doesn't make sense to build substitutes.
7371 (arguments `(#:substitutable? #f))
7372 (propagated-inputs
7373 `(("r-bsgenome" ,r-bsgenome)))
7374 (home-page
5713bbf1 7375 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.1000genomes.hs37d5/")
aa3eeeb5
RJ		 7376 (synopsis "Full genome sequences for Homo sapiens")
7377 (description
7378 "This package provides full genome sequences for Homo sapiens from
73791000genomes phase2 reference genome sequence (hs37d5), based on NCBI GRCh37.")
7380 (license license:artistic2.0)))
7381
c43a011d
RW		 7382(define-public r-impute
7383 (package
7384 (name "r-impute")
e6ce4bf0 7385 (version "1.52.0")
c43a011d
RW		 7386 (source (origin
7387 (method url-fetch)
7388 (uri (bioconductor-uri "impute" version))
7389 (sha256
7390 (base32
e6ce4bf0 7391 "0b8r4swvyx3cjcc2ky8yn0ncpzlbi1pgfsn3wpbjmhh7sqrffm2n"))))
c43a011d
RW		 7392 (inputs
7393 `(("gfortran" ,gfortran)))
7394 (build-system r-build-system)
5713bbf1 7395 (home-page "https://bioconductor.org/packages/impute")
c43a011d
RW		 7396 (synopsis "Imputation for microarray data")
7397 (description
7398 "This package provides a function to impute missing gene expression
7399microarray data, using nearest neighbor averaging.")
7400 (license license:gpl2+)))
7401
03ea5a35
RW		 7402(define-public r-seqpattern
7403 (package
7404 (name "r-seqpattern")
0f948b11 7405 (version "1.10.0")
03ea5a35
RW		 7406 (source (origin
7407 (method url-fetch)
7408 (uri (bioconductor-uri "seqPattern" version))
7409 (sha256
7410 (base32
0f948b11 7411 "1kcm5w83q7w0v0vs7nyp4gq5z86c6n6pqy9zmyyhxcrns7f597pm"))))
03ea5a35
RW		 7412 (properties
7413 `((upstream-name . "seqPattern")))
7414 (build-system r-build-system)
7415 (propagated-inputs
7416 `(("r-biostrings" ,r-biostrings)
7417 ("r-genomicranges" ,r-genomicranges)
7418 ("r-iranges" ,r-iranges)
e92dd6f5 7419 ("r-kernsmooth" ,r-kernsmooth)
03ea5a35 7420 ("r-plotrix" ,r-plotrix)))
5713bbf1 7421 (home-page "https://bioconductor.org/packages/seqPattern")
03ea5a35
RW		 7422 (synopsis "Visualising oligonucleotide patterns and motif occurrences")
7423 (description
7424 "This package provides tools to visualize oligonucleotide patterns and
7425sequence motif occurrences across a large set of sequences centred at a common
7426reference point and sorted by a user defined feature.")
7427 (license license:gpl3+)))
7428
cb933df6
RW		 7429(define-public r-genomation
7430 (package
7431 (name "r-genomation")
e127c918 7432 (version "1.10.0")
cb933df6
RW		 7433 (source (origin
7434 (method url-fetch)
7435 (uri (bioconductor-uri "genomation" version))
7436 (sha256
7437 (base32
e127c918 7438 "1ddd8c9w1f1i1ga9rpbwiic8rsaws1chdxx4j38bpyaiy4zhz1ca"))))
cb933df6
RW		 7439 (build-system r-build-system)
7440 (propagated-inputs
7441 `(("r-biostrings" ,r-biostrings)
7442 ("r-bsgenome" ,r-bsgenome)
7443 ("r-data-table" ,r-data-table)
7444 ("r-genomeinfodb" ,r-genomeinfodb)
7445 ("r-genomicalignments" ,r-genomicalignments)
7446 ("r-genomicranges" ,r-genomicranges)
7447 ("r-ggplot2" ,r-ggplot2)
7448 ("r-gridbase" ,r-gridbase)
7449 ("r-impute" ,r-impute)
7450 ("r-iranges" ,r-iranges)
7451 ("r-matrixstats" ,r-matrixstats)
7452 ("r-plotrix" ,r-plotrix)
7453 ("r-plyr" ,r-plyr)
51c3c490 7454 ("r-rcpp" ,r-rcpp)
cb933df6
RW		 7455 ("r-readr" ,r-readr)
7456 ("r-reshape2" ,r-reshape2)
7457 ("r-rsamtools" ,r-rsamtools)
7458 ("r-rtracklayer" ,r-rtracklayer)
51c3c490
RW		 7459 ("r-runit" ,r-runit)
7460 ("r-s4vectors" ,r-s4vectors)
cb933df6
RW		 7461 ("r-seqpattern" ,r-seqpattern)))
7462 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7463 (synopsis "Summary, annotation and visualization of genomic data")
7464 (description
7465 "This package provides a package for summary and annotation of genomic
7466intervals. Users can visualize and quantify genomic intervals over
7467pre-defined functional regions, such as promoters, exons, introns, etc. The
7468genomic intervals represent regions with a defined chromosome position, which
7469may be associated with a score, such as aligned reads from HT-seq experiments,
7470TF binding sites, methylation scores, etc. The package can use any tabular
7471genomic feature data as long as it has minimal information on the locations of
7472genomic intervals. In addition, it can use BAM or BigWig files as input.")
7473 (license license:artistic2.0)))
7474
64efa307
RW		 7475(define-public r-genomationdata
7476 (package
7477 (name "r-genomationdata")
59198f8f 7478 (version "1.6.0")
64efa307
RW		 7479 (source (origin
7480 (method url-fetch)
7481 ;; We cannot use bioconductor-uri here because this tarball is
7482 ;; located under "data/annotation/" instead of "bioc/".
7483 (uri (string-append "https://bioconductor.org/packages/"
7484 "release/data/experiment/src/contrib/"
7485 "genomationData_" version ".tar.gz"))
7486 (sha256
7487 (base32
59198f8f 7488 "16dqwb7wx1igx77zdbcskx5m1hs4g4gp2hl56zzm70hcagnlkz8y"))))
64efa307
RW		 7489 (build-system r-build-system)
7490 ;; As this package provides little more than large data files, it doesn't
7491 ;; make sense to build substitutes.
7492 (arguments `(#:substitutable? #f))
7493 (native-inputs
7494 `(("r-knitr" ,r-knitr)))
7495 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7496 (synopsis "Experimental data for use with the genomation package")
7497 (description
7498 "This package contains experimental genetic data for use with the
7499genomation package. Included are Chip Seq, Methylation and Cage data,
7500downloaded from Encode.")
7501 (license license:gpl3+)))
7502
486da491
RW		 7503(define-public r-org-hs-eg-db
7504 (package
7505 (name "r-org-hs-eg-db")
83f9a6fb 7506 (version "3.4.0")
486da491
RW		 7507 (source (origin
7508 (method url-fetch)
7509 ;; We cannot use bioconductor-uri here because this tarball is
7510 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7511 (uri (string-append "https://www.bioconductor.org/packages/"
486da491
RW		 7512 "release/data/annotation/src/contrib/"
7513 "org.Hs.eg.db_" version ".tar.gz"))
7514 (sha256
7515 (base32
83f9a6fb 7516 "19mg64pw8zcvb9yxzzyf7caz1kvdrkfsj1hd84bzq7crrh8kc4y6"))))
486da491
RW		 7517 (properties
7518 `((upstream-name . "org.Hs.eg.db")))
7519 (build-system r-build-system)
7520 (propagated-inputs
7521 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7522 (home-page "https://www.bioconductor.org/packages/org.Hs.eg.db/")
486da491
RW		 7523 (synopsis "Genome wide annotation for Human")
7524 (description
7525 "This package provides mappings from Entrez gene identifiers to various
7526annotations for the human genome.")
7527 (license license:artistic2.0)))
7528
fefedf98
RW		 7529(define-public r-org-ce-eg-db
7530 (package
7531 (name "r-org-ce-eg-db")
f04a3eff 7532 (version "3.4.0")
fefedf98
RW		 7533 (source (origin
7534 (method url-fetch)
7535 ;; We cannot use bioconductor-uri here because this tarball is
7536 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7537 (uri (string-append "https://www.bioconductor.org/packages/"
fefedf98
RW		 7538 "release/data/annotation/src/contrib/"
7539 "org.Ce.eg.db_" version ".tar.gz"))
7540 (sha256
7541 (base32
f04a3eff 7542 "12llfzrrc09kj2wzbisdspv38qzkzgpsbn8kv7qkwg746k3pq436"))))
fefedf98
RW		 7543 (properties
7544 `((upstream-name . "org.Ce.eg.db")))
7545 (build-system r-build-system)
7546 (propagated-inputs
7547 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7548 (home-page "https://www.bioconductor.org/packages/org.Ce.eg.db/")
fefedf98
RW		 7549 (synopsis "Genome wide annotation for Worm")
7550 (description
7551 "This package provides mappings from Entrez gene identifiers to various
7552annotations for the genome of the model worm Caenorhabditis elegans.")
7553 (license license:artistic2.0)))
7554
16c53a1e
RW		 7555(define-public r-org-dm-eg-db
7556 (package
7557 (name "r-org-dm-eg-db")
0cfac6f2 7558 (version "3.4.0")
16c53a1e
RW		 7559 (source (origin
7560 (method url-fetch)
7561 ;; We cannot use bioconductor-uri here because this tarball is
7562 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7563 (uri (string-append "https://www.bioconductor.org/packages/"
16c53a1e
RW		 7564 "release/data/annotation/src/contrib/"
7565 "org.Dm.eg.db_" version ".tar.gz"))
7566 (sha256
7567 (base32
0cfac6f2 7568 "1vzbphbrh1cf7xi5cksia9xy9a9l42js2z2qsajvjxvddiphrb7j"))))
16c53a1e
RW		 7569 (properties
7570 `((upstream-name . "org.Dm.eg.db")))
7571 (build-system r-build-system)
7572 (propagated-inputs
7573 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7574 (home-page "https://www.bioconductor.org/packages/org.Dm.eg.db/")
16c53a1e
RW		 7575 (synopsis "Genome wide annotation for Fly")
7576 (description
7577 "This package provides mappings from Entrez gene identifiers to various
7578annotations for the genome of the model fruit fly Drosophila melanogaster.")
7579 (license license:artistic2.0)))
7580
e761beb9
RW		 7581(define-public r-org-mm-eg-db
7582 (package
7583 (name "r-org-mm-eg-db")
7c45670b 7584 (version "3.4.0")
e761beb9
RW		 7585 (source (origin
7586 (method url-fetch)
7587 ;; We cannot use bioconductor-uri here because this tarball is
7588 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7589 (uri (string-append "https://www.bioconductor.org/packages/"
e761beb9
RW		 7590 "release/data/annotation/src/contrib/"
7591 "org.Mm.eg.db_" version ".tar.gz"))
7592 (sha256
7593 (base32
7c45670b 7594 "1lykjqjaf01fmgg3cvfcvwd5xjq6zc5vbxnm5r4l32fzvl89q50c"))))
e761beb9
RW		 7595 (properties
7596 `((upstream-name . "org.Mm.eg.db")))
7597 (build-system r-build-system)
7598 (propagated-inputs
7599 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7600 (home-page "https://www.bioconductor.org/packages/org.Mm.eg.db/")
e761beb9
RW		 7601 (synopsis "Genome wide annotation for Mouse")
7602 (description
7603 "This package provides mappings from Entrez gene identifiers to various
7604annotations for the genome of the model mouse Mus musculus.")
7605 (license license:artistic2.0)))
7606
936e7d67
RW		 7607(define-public r-seqlogo
7608 (package
7609 (name "r-seqlogo")
61770089 7610 (version "1.44.0")
936e7d67
RW		 7611 (source
7612 (origin
7613 (method url-fetch)
7614 (uri (bioconductor-uri "seqLogo" version))
7615 (sha256
7616 (base32
61770089 7617 "1ql4q4vx0j61a893dqc3c8zxmgs8sqhy3j1qhyfdvbd01vw9w1kq"))))
936e7d67
RW		 7618 (properties `((upstream-name . "seqLogo")))
7619 (build-system r-build-system)
5713bbf1 7620 (home-page "https://bioconductor.org/packages/seqLogo")
936e7d67
RW		 7621 (synopsis "Sequence logos for DNA sequence alignments")
7622 (description
7623 "seqLogo takes the position weight matrix of a DNA sequence motif and
7624plots the corresponding sequence logo as introduced by Schneider and
7625Stephens (1990).")
7626 (license license:lgpl2.0+)))
7627
c90a4baf
RW		 7628(define-public r-bsgenome-hsapiens-ucsc-hg19
7629 (package
7630 (name "r-bsgenome-hsapiens-ucsc-hg19")
7631 (version "1.4.0")
7632 (source (origin
7633 (method url-fetch)
7634 ;; We cannot use bioconductor-uri here because this tarball is
7635 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7636 (uri (string-append "https://www.bioconductor.org/packages/"
c90a4baf
RW		 7637 "release/data/annotation/src/contrib/"
7638 "BSgenome.Hsapiens.UCSC.hg19_"
7639 version ".tar.gz"))
7640 (sha256
7641 (base32
9d217d27 7642 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
c90a4baf
RW		 7643 (properties
7644 `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
7645 (build-system r-build-system)
7646 ;; As this package provides little more than a very large data file it
7647 ;; doesn't make sense to build substitutes.
7648 (arguments `(#:substitutable? #f))
7649 (propagated-inputs
7650 `(("r-bsgenome" ,r-bsgenome)))
7651 (home-page
5713bbf1 7652 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
c90a4baf
RW		 7653 (synopsis "Full genome sequences for Homo sapiens")
7654 (description
7655 "This package provides full genome sequences for Homo sapiens as provided
7656by UCSC (hg19, February 2009) and stored in Biostrings objects.")
7657 (license license:artistic2.0)))
7658
a3e90287
RW		 7659(define-public r-bsgenome-mmusculus-ucsc-mm9
7660 (package
7661 (name "r-bsgenome-mmusculus-ucsc-mm9")
7662 (version "1.4.0")
7663 (source (origin
7664 (method url-fetch)
7665 ;; We cannot use bioconductor-uri here because this tarball is
7666 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7667 (uri (string-append "https://www.bioconductor.org/packages/"
a3e90287
RW		 7668 "release/data/annotation/src/contrib/"
7669 "BSgenome.Mmusculus.UCSC.mm9_"
7670 version ".tar.gz"))
7671 (sha256
7672 (base32
7673 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
7674 (properties
7675 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
7676 (build-system r-build-system)
7677 ;; As this package provides little more than a very large data file it
7678 ;; doesn't make sense to build substitutes.
7679 (arguments `(#:substitutable? #f))
7680 (propagated-inputs
7681 `(("r-bsgenome" ,r-bsgenome)))
7682 (home-page
5713bbf1 7683 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
a3e90287
RW		 7684 (synopsis "Full genome sequences for Mouse")
7685 (description
7686 "This package provides full genome sequences for Mus musculus (Mouse) as
7687provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
7688 (license license:artistic2.0)))
7689
4714d521
RW		 7690(define-public r-bsgenome-mmusculus-ucsc-mm10
7691 (package
7692 (name "r-bsgenome-mmusculus-ucsc-mm10")
7693 (version "1.4.0")
7694 (source (origin
7695 (method url-fetch)
7696 ;; We cannot use bioconductor-uri here because this tarball is
7697 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7698 (uri (string-append "https://www.bioconductor.org/packages/"
4714d521
RW		 7699 "release/data/annotation/src/contrib/"
7700 "BSgenome.Mmusculus.UCSC.mm10_"
7701 version ".tar.gz"))
7702 (sha256
7703 (base32
7704 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
7705 (properties
7706 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
7707 (build-system r-build-system)
7708 ;; As this package provides little more than a very large data file it
7709 ;; doesn't make sense to build substitutes.
7710 (arguments `(#:substitutable? #f))
7711 (propagated-inputs
7712 `(("r-bsgenome" ,r-bsgenome)))
7713 (home-page
5713bbf1 7714 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
4714d521
RW		 7715 (synopsis "Full genome sequences for Mouse")
7716 (description
7717 "This package provides full genome sequences for Mus
7718musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
7719in Biostrings objects.")
7720 (license license:artistic2.0)))
7721
c5173d74
RJ		 7722(define-public r-txdb-mmusculus-ucsc-mm10-knowngene
7723 (package
7724 (name "r-txdb-mmusculus-ucsc-mm10-knowngene")
7725 (version "3.4.0")
7726 (source (origin
7727 (method url-fetch)
7728 ;; We cannot use bioconductor-uri here because this tarball is
7729 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7730 (uri (string-append "https://www.bioconductor.org/packages/"
c5173d74
RJ		 7731 "release/data/annotation/src/contrib/"
7732 "TxDb.Mmusculus.UCSC.mm10.knownGene_"
7733 version ".tar.gz"))
7734 (sha256
7735 (base32
7736 "08gava9wsvpcqz51k2sni3pj03n5155v32d9riqbf305nbirqbkb"))))
7737 (properties
7738 `((upstream-name . "TxDb.Mmusculus.UCSC.mm10.knownGene")))
7739 (build-system r-build-system)
7740 ;; As this package provides little more than a very large data file it
7741 ;; doesn't make sense to build substitutes.
7742 (arguments `(#:substitutable? #f))
7743 (propagated-inputs
7744 `(("r-bsgenome" ,r-bsgenome)
7745 ("r-genomicfeatures" ,r-genomicfeatures)
7746 ("r-annotationdbi" ,r-annotationdbi)))
7747 (home-page
5713bbf1 7748 "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm10.knownGene/")
c5173d74
RJ		 7749 (synopsis "Annotation package for TxDb knownGene object(s) for Mouse")
7750 (description
7751 "This package loads a TxDb object, which is an R interface to
7752prefabricated databases contained in this package. This package provides
7753the TxDb object of Mouse data as provided by UCSC (mm10, December 2011)
7754based on the knownGene track.")
7755 (license license:artistic2.0)))
7756
943bd627
RW		 7757(define-public r-bsgenome-celegans-ucsc-ce6
7758 (package
7759 (name "r-bsgenome-celegans-ucsc-ce6")
7760 (version "1.4.0")
7761 (source (origin
7762 (method url-fetch)
7763 ;; We cannot use bioconductor-uri here because this tarball is
7764 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7765 (uri (string-append "https://www.bioconductor.org/packages/"
943bd627
RW		 7766 "release/data/annotation/src/contrib/"
7767 "BSgenome.Celegans.UCSC.ce6_"
7768 version ".tar.gz"))
7769 (sha256
7770 (base32
7771 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
7772 (properties
7773 `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
7774 (build-system r-build-system)
7775 ;; As this package provides little more than a very large data file it
7776 ;; doesn't make sense to build substitutes.
7777 (arguments `(#:substitutable? #f))
7778 (propagated-inputs
7779 `(("r-bsgenome" ,r-bsgenome)))
7780 (home-page
5713bbf1 7781 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
943bd627
RW		 7782 (synopsis "Full genome sequences for Worm")
7783 (description
7784 "This package provides full genome sequences for Caenorhabditis
7785elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
7786objects.")
7787 (license license:artistic2.0)))
7788
fc47c7d6
RW		 7789(define-public r-bsgenome-celegans-ucsc-ce10
7790 (package
7791 (name "r-bsgenome-celegans-ucsc-ce10")
7792 (version "1.4.0")
7793 (source (origin
7794 (method url-fetch)
7795 ;; We cannot use bioconductor-uri here because this tarball is
7796 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7797 (uri (string-append "https://www.bioconductor.org/packages/"
fc47c7d6
RW		 7798 "release/data/annotation/src/contrib/"
7799 "BSgenome.Celegans.UCSC.ce10_"
7800 version ".tar.gz"))
7801 (sha256
7802 (base32
7803 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
7804 (properties
7805 `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
7806 (build-system r-build-system)
7807 ;; As this package provides little more than a very large data file it
7808 ;; doesn't make sense to build substitutes.
7809 (arguments `(#:substitutable? #f))
7810 (propagated-inputs
7811 `(("r-bsgenome" ,r-bsgenome)))
7812 (home-page
5713bbf1 7813 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
fc47c7d6
RW		 7814 (synopsis "Full genome sequences for Worm")
7815 (description
7816 "This package provides full genome sequences for Caenorhabditis
7817elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
7818objects.")
7819 (license license:artistic2.0)))
7820
6dc60998
RW		 7821(define-public r-bsgenome-dmelanogaster-ucsc-dm3
7822 (package
7823 (name "r-bsgenome-dmelanogaster-ucsc-dm3")
7824 (version "1.4.0")
7825 (source (origin
7826 (method url-fetch)
7827 ;; We cannot use bioconductor-uri here because this tarball is
7828 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7829 (uri (string-append "https://www.bioconductor.org/packages/"
6dc60998
RW		 7830 "release/data/annotation/src/contrib/"
7831 "BSgenome.Dmelanogaster.UCSC.dm3_"
7832 version ".tar.gz"))
7833 (sha256
7834 (base32
7835 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
7836 (properties
7837 `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
7838 (build-system r-build-system)
7839 ;; As this package provides little more than a very large data file it
7840 ;; doesn't make sense to build substitutes.
7841 (arguments `(#:substitutable? #f))
7842 (propagated-inputs
7843 `(("r-bsgenome" ,r-bsgenome)))
7844 (home-page
5713bbf1 7845 "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
6dc60998
RW		 7846 (synopsis "Full genome sequences for Fly")
7847 (description
7848 "This package provides full genome sequences for Drosophila
7849melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
7850Biostrings objects.")
7851 (license license:artistic2.0)))
7852
ae2462f7
RW		 7853(define-public r-motifrg
7854 (package
7855 (name "r-motifrg")
ce59d625 7856 (version "1.22.0")
ae2462f7
RW		 7857 (source
7858 (origin
7859 (method url-fetch)
7860 (uri (bioconductor-uri "motifRG" version))
7861 (sha256
7862 (base32
ce59d625 7863 "193zl2rlzwxv9p9q5i7rilj3w05ndqfyp9bdpvagp5s5cin4hf44"))))
ae2462f7
RW		 7864 (properties `((upstream-name . "motifRG")))
7865 (build-system r-build-system)
7866 (propagated-inputs
7867 `(("r-biostrings" ,r-biostrings)
7868 ("r-bsgenome" ,r-bsgenome)
007424b5 7869 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
ae2462f7
RW		 7870 ("r-iranges" ,r-iranges)
7871 ("r-seqlogo" ,r-seqlogo)
7872 ("r-xvector" ,r-xvector)))
5713bbf1 7873 (home-page "https://bioconductor.org/packages/motifRG")
ae2462f7
RW		 7874 (synopsis "Discover motifs in high throughput sequencing data")
7875 (description
7876 "This package provides tools for discriminative motif discovery in high
7877throughput genetic sequencing data sets using regression methods.")
7878 (license license:artistic2.0)))
7879
a5002ae7
AE		 7880(define-public r-qtl
7881 (package
7882 (name "r-qtl")
c8a9b2bc 7883 (version "1.41-6")
a5002ae7
AE		 7884 (source
7885 (origin
7886 (method url-fetch)
7887 (uri (string-append "mirror://cran/src/contrib/qtl_"
7888 version ".tar.gz"))
7889 (sha256
7890 (base32
c8a9b2bc 7891 "067az4v432zxp6lxck8d7vlh9w4r13r0mvw5zsglyaqwsh3d9sad"))))
a5002ae7
AE		 7892 (build-system r-build-system)
7893 (home-page "http://rqtl.org/")
7894 (synopsis "R package for analyzing QTL experiments in genetics")
7895 (description "R/qtl is an extension library for the R statistics
7896system. It is used to analyze experimental crosses for identifying
7897genes contributing to variation in quantitative traits (so-called
7898quantitative trait loci, QTLs).
7899
7900Using a hidden Markov model, R/qtl allows to estimate genetic maps, to
7901identify genotyping errors, and to perform single-QTL and two-QTL,
7902two-dimensional genome scans.")
7903 (license license:gpl3)))
d1e32822 7904
9e3ba31c
RJ		 7905(define-public r-zlibbioc
7906 (package
7907 (name "r-zlibbioc")
5c184700 7908 (version "1.24.0")
9e3ba31c
RJ		 7909 (source (origin
7910 (method url-fetch)
7911 (uri (bioconductor-uri "zlibbioc" version))
7912 (sha256
7913 (base32
5c184700 7914 "1zr9hbh55hglfpy15cpxwmddxblhyb0an15953l3rbhmlh2vpy92"))))
9e3ba31c
RJ		 7915 (properties
7916 `((upstream-name . "zlibbioc")))
7917 (build-system r-build-system)
7918 (home-page "https://bioconductor.org/packages/zlibbioc")
7919 (synopsis "Provider for zlib-1.2.5 to R packages")
7920 (description "This package uses the source code of zlib-1.2.5 to create
7921libraries for systems that do not have these available via other means.")
7922 (license license:artistic2.0)))
7923
e619a5c2
RW		 7924(define-public r-r4rna
7925 (package
7926 (name "r-r4rna")
7927 (version "0.1.4")
7928 (source
7929 (origin
7930 (method url-fetch)
7931 (uri (string-append "http://www.e-rna.org/r-chie/files/R4RNA_"
7932 version ".tar.gz"))
7933 (sha256
7934 (base32
7935 "1p0i78wh76jfgmn9jphbwwaz6yy6pipzfg08xs54cxavxg2j81p5"))))
7936 (build-system r-build-system)
7937 (propagated-inputs
7938 `(("r-optparse" ,r-optparse)
7939 ("r-rcolorbrewer" ,r-rcolorbrewer)))
7940 (home-page "http://www.e-rna.org/r-chie/index.cgi")
7941 (synopsis "Analysis framework for RNA secondary structure")
7942 (description
7943 "The R4RNA package aims to be a general framework for the analysis of RNA
7944secondary structure and comparative analysis in R.")
7945 (license license:gpl3+)))
7946
52765a63
RW		 7947(define-public r-rhtslib
7948 (package
7949 (name "r-rhtslib")
1d0263b4 7950 (version "1.10.0")
52765a63
RW		 7951 (source
7952 (origin
7953 (method url-fetch)
7954 (uri (bioconductor-uri "Rhtslib" version))
7955 (sha256
7956 (base32
1d0263b4 7957 "1dw3p44bfr0m7w39ckc2k37sjcp1zz0b9g12mr8am15jaj6v0q2j"))))
52765a63
RW		 7958 (properties `((upstream-name . "Rhtslib")))
7959 (build-system r-build-system)
7960 (propagated-inputs
7961 `(("r-zlibbioc" ,r-zlibbioc)))
7962 (inputs
7963 `(("zlib" ,zlib)))
53ca52f0
RW		 7964 (native-inputs
7965 `(("autoconf" ,autoconf)))
52765a63
RW		 7966 (home-page "https://github.com/nhayden/Rhtslib")
7967 (synopsis "High-throughput sequencing library as an R package")
7968 (description
7969 "This package provides the HTSlib C library for high-throughput
7970nucleotide sequence analysis. The package is primarily useful to developers
7971of other R packages who wish to make use of HTSlib.")
7972 (license license:lgpl2.0+)))
7973
fe02c4c9
RW		 7974(define-public r-bamsignals
7975 (package
7976 (name "r-bamsignals")
da153b7f 7977 (version "1.10.0")
fe02c4c9
RW		 7978 (source
7979 (origin
7980 (method url-fetch)
7981 (uri (bioconductor-uri "bamsignals" version))
7982 (sha256
7983 (base32
da153b7f 7984 "15id6mkj95skb4kfafvfs2j7ylydal60c3pspcl7llhwpq6vcqvl"))))
fe02c4c9
RW		 7985 (build-system r-build-system)
7986 (propagated-inputs
7987 `(("r-biocgenerics" ,r-biocgenerics)
7988 ("r-genomicranges" ,r-genomicranges)
7989 ("r-iranges" ,r-iranges)
7990 ("r-rcpp" ,r-rcpp)
7991 ("r-rhtslib" ,r-rhtslib)
7992 ("r-zlibbioc" ,r-zlibbioc)))
7993 (inputs
7994 `(("zlib" ,zlib)))
5713bbf1 7995 (home-page "https://bioconductor.org/packages/bamsignals")
fe02c4c9
RW		 7996 (synopsis "Extract read count signals from bam files")
7997 (description
7998 "This package allows to efficiently obtain count vectors from indexed bam
7999files. It counts the number of nucleotide sequence reads in given genomic
8000ranges and it computes reads profiles and coverage profiles. It also handles
8001paired-end data.")
8002 (license license:gpl2+)))
8003
89984be4
RW		 8004(define-public r-rcas
8005 (package
8006 (name "r-rcas")
d82937fb 8007 (version "1.3.4")
89984be4
RW		 8008 (source (origin
8009 (method url-fetch)
8010 (uri (string-append "https://github.com/BIMSBbioinfo/RCAS/archive/v"
8011 version ".tar.gz"))
8012 (file-name (string-append name "-" version ".tar.gz"))
8013 (sha256
8014 (base32
d82937fb 8015 "1qgc7vi6fpzl440yg7jhiycg5q336kd4pxqzx10yx2zcq3bq3msg"))))
89984be4
RW		 8016 (build-system r-build-system)
8017 (native-inputs
8018 `(("r-knitr" ,r-knitr)
8019 ("r-testthat" ,r-testthat)
8020 ;; During vignette building knitr checks that "pandoc-citeproc"
8021 ;; is in the PATH.
8022 ("ghc-pandoc-citeproc" ,ghc-pandoc-citeproc)))
8023 (propagated-inputs
8024 `(("r-data-table" ,r-data-table)
8025 ("r-biomart" ,r-biomart)
8026 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
8027 ("r-org-ce-eg-db" ,r-org-ce-eg-db)
8028 ("r-org-dm-eg-db" ,r-org-dm-eg-db)
8029 ("r-org-mm-eg-db" ,r-org-mm-eg-db)
8030 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8031 ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
8032 ("r-bsgenome-celegans-ucsc-ce10" ,r-bsgenome-celegans-ucsc-ce10)
8033 ("r-bsgenome-dmelanogaster-ucsc-dm3" ,r-bsgenome-dmelanogaster-ucsc-dm3)
8034 ("r-topgo" ,r-topgo)
8035 ("r-dt" ,r-dt)
ebfd6a71 8036 ("r-pbapply" ,r-pbapply)
89984be4 8037 ("r-plotly" ,r-plotly)
2c8d6c0b 8038 ("r-plotrix" ,r-plotrix)
89984be4
RW		 8039 ("r-motifrg" ,r-motifrg)
8040 ("r-genomation" ,r-genomation)
8041 ("r-genomicfeatures" ,r-genomicfeatures)
8042 ("r-rtracklayer" ,r-rtracklayer)
8043 ("r-rmarkdown" ,r-rmarkdown)))
8044 (synopsis "RNA-centric annotation system")
8045 (description
8046 "RCAS aims to be a standalone RNA-centric annotation system that provides
8047intuitive reports and publication-ready graphics. This package provides the R
8048library implementing most of the pipeline's features.")
8049 (home-page "https://github.com/BIMSBbioinfo/RCAS")
75690c9f 8050 (license license:artistic2.0)))
89984be4 8051
50937297
RW		 8052(define-public rcas-web
8053 (package
8054 (name "rcas-web")
01d87d3c 8055 (version "0.0.4")
50937297
RW		 8056 (source
8057 (origin
8058 (method url-fetch)
8059 (uri (string-append "https://github.com/BIMSBbioinfo/rcas-web/"
8060 "releases/download/v" version
8061 "/rcas-web-" version ".tar.gz"))
8062 (sha256
8063 (base32
01d87d3c 8064 "1p16frfys41a8yaa4gkm457nzkqhqs2pc3lkac0ds457w9w5j1gm"))))
50937297
RW		 8065 (build-system gnu-build-system)
8066 (arguments
8067 `(#:phases
8068 (modify-phases %standard-phases
8069 (add-after 'install 'wrap-executable
8070 (lambda* (#:key inputs outputs #:allow-other-keys)
8071 (let* ((out (assoc-ref outputs "out"))
8072 (json (assoc-ref inputs "guile-json"))
8073 (redis (assoc-ref inputs "guile-redis"))
8074 (path (string-append
8075 json "/share/guile/site/2.2:"
8076 redis "/share/guile/site/2.2")))
8077 (wrap-program (string-append out "/bin/rcas-web")
8078 `("GUILE_LOAD_PATH" ":" = (,path))
8079 `("GUILE_LOAD_COMPILED_PATH" ":" = (,path))
8080 `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
8081 #t)))))
8082 (inputs
2d7c4ae3 8083 `(("r-minimal" ,r-minimal)
50937297 8084 ("r-rcas" ,r-rcas)
f6396d86 8085 ("guile-next" ,guile-2.2)
2252f087 8086 ("guile-json" ,guile-json)
50937297
RW		 8087 ("guile-redis" ,guile2.2-redis)))
8088 (native-inputs
8089 `(("pkg-config" ,pkg-config)))
8090 (home-page "https://github.com/BIMSBbioinfo/rcas-web")
8091 (synopsis "Web interface for RNA-centric annotation system (RCAS)")
8092 (description "This package provides a simple web interface for the
8093@dfn{RNA-centric annotation system} (RCAS).")
8094 (license license:agpl3+)))
8095
7500e42b
RJ		 8096(define-public r-mutationalpatterns
8097 (package
8098 (name "r-mutationalpatterns")
39d9098d 8099 (version "1.4.0")
7500e42b
RJ		 8100 (source
8101 (origin
8102 (method url-fetch)
8103 (uri (bioconductor-uri "MutationalPatterns" version))
8104 (sha256
8105 (base32
39d9098d 8106 "0sqbrswg8ylkjb9q3vqcb5ggwixynwj6hyv2n4sk7snyk61z3fq9"))))
7500e42b
RJ		 8107 (build-system r-build-system)
8108 (propagated-inputs
8109 `(("r-biocgenerics" ,r-biocgenerics)
8110 ("r-biostrings" ,r-biostrings)
cf4ac4e4
RJ		 8111 ("r-bsgenome-hsapiens-1000g" ,r-bsgenome-hsapiens-1000genomes-hs37d5)
8112 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7500e42b
RJ		 8113 ("r-genomicranges" ,r-genomicranges)
8114 ("r-genomeinfodb" ,r-genomeinfodb)
8115 ("r-ggplot2" ,r-ggplot2)
8116 ("r-gridextra" ,r-gridextra)
8117 ("r-iranges" ,r-iranges)
8118 ("r-nmf" ,r-nmf)
8119 ("r-plyr" ,r-plyr)
8120 ("r-pracma" ,r-pracma)
8121 ("r-reshape2" ,r-reshape2)
39d9098d
RW		 8122 ("r-cowplot" ,r-cowplot)
8123 ("r-ggdendro" ,r-ggdendro)
8124 ("r-s4vectors" ,r-s4vectors)
7500e42b
RJ		 8125 ("r-summarizedexperiment" ,r-summarizedexperiment)
8126 ("r-variantannotation" ,r-variantannotation)))
5713bbf1 8127 (home-page "https://bioconductor.org/packages/MutationalPatterns/")
7500e42b
RJ		 8128 (synopsis "Extract and visualize mutational patterns in genomic data")
8129 (description "This package provides an extensive toolset for the
8130characterization and visualization of a wide range of mutational patterns
8131in SNV base substitution data.")
8132 (license license:expat)))
8133
d7160529
RW		 8134(define-public r-wgcna
8135 (package
8136 (name "r-wgcna")
1b22ecda 8137 (version "1.61")
d7160529
RW		 8138 (source
8139 (origin
8140 (method url-fetch)
8141 (uri (cran-uri "WGCNA" version))
8142 (sha256
8143 (base32
1b22ecda 8144 "1vrc2k33a196hrrl7k0z534fp96vv0shmigcr65ny1q0v6lq0h6i"))))
d7160529
RW		 8145 (properties `((upstream-name . "WGCNA")))
8146 (build-system r-build-system)
8147 (propagated-inputs
8148 `(("r-annotationdbi" ,r-annotationdbi)
8149 ("r-doparallel" ,r-doparallel)
8150 ("r-dynamictreecut" ,r-dynamictreecut)
8151 ("r-fastcluster" ,r-fastcluster)
8152 ("r-foreach" ,r-foreach)
8153 ("r-go-db" ,r-go-db)
8154 ("r-hmisc" ,r-hmisc)
8155 ("r-impute" ,r-impute)
1b22ecda
RW		 8156 ("r-rcpp" ,r-rcpp)
8157 ("r-robust" ,r-robust)
8158 ("r-survival" ,r-survival)
d7160529
RW		 8159 ("r-matrixstats" ,r-matrixstats)
8160 ("r-preprocesscore" ,r-preprocesscore)))
8161 (home-page
8162 "http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/Rpackages/WGCNA/")
8163 (synopsis "Weighted correlation network analysis")
8164 (description
8165 "This package provides functions necessary to perform Weighted
8166Correlation Network Analysis on high-dimensional data. It includes functions
8167for rudimentary data cleaning, construction and summarization of correlation
8168networks, module identification and functions for relating both variables and
8169modules to sample traits. It also includes a number of utility functions for
8170data manipulation and visualization.")
8171 (license license:gpl2+)))
8172
c827f202
RW		 8173(define-public r-chipkernels
8174 (let ((commit "c9cfcacb626b1221094fb3490ea7bac0fd625372")
8175 (revision "1"))
8176 (package
8177 (name "r-chipkernels")
8178 (version (string-append "1.1-" revision "." (string-take commit 9)))
8179 (source
8180 (origin
8181 (method git-fetch)
8182 (uri (git-reference
8183 (url "https://github.com/ManuSetty/ChIPKernels.git")
8184 (commit commit)))
8185 (file-name (string-append name "-" version))
8186 (sha256
8187 (base32
8188 "14bj5qhjm1hsm9ay561nfbqi9wxsa7y487df2idsaaf6z10nw4v0"))))
8189 (build-system r-build-system)
8190 (propagated-inputs
8191 `(("r-iranges" ,r-iranges)
8192 ("r-xvector" ,r-xvector)
8193 ("r-biostrings" ,r-biostrings)
8194 ("r-bsgenome" ,r-bsgenome)
8195 ("r-gtools" ,r-gtools)
8196 ("r-genomicranges" ,r-genomicranges)
8197 ("r-sfsmisc" ,r-sfsmisc)
8198 ("r-kernlab" ,r-kernlab)
8199 ("r-s4vectors" ,r-s4vectors)
8200 ("r-biocgenerics" ,r-biocgenerics)))
8201 (home-page "https://github.com/ManuSetty/ChIPKernels")
8202 (synopsis "Build string kernels for DNA Sequence analysis")
8203 (description "ChIPKernels is an R package for building different string
8204kernels used for DNA Sequence analysis. A dictionary of the desired kernel
8205must be built and this dictionary can be used for determining kernels for DNA
8206Sequences.")
8207 (license license:gpl2+))))
8208
2d9fb170
RW		 8209(define-public r-seqgl
8210 (package
8211 (name "r-seqgl")
8212 (version "1.1.4")
8213 (source
8214 (origin
8215 (method url-fetch)
8216 (uri (string-append "https://github.com/ManuSetty/SeqGL/"
8217 "archive/" version ".tar.gz"))
8218 (file-name (string-append name "-" version ".tar.gz"))
8219 (sha256
8220 (base32
8221 "0pnk1p3sci5yipyc8xnb6jbmydpl80fld927xgnbcv104hy8h8yh"))))
8222 (build-system r-build-system)
8223 (propagated-inputs
8224 `(("r-biostrings" ,r-biostrings)
8225 ("r-chipkernels" ,r-chipkernels)
8226 ("r-genomicranges" ,r-genomicranges)
8227 ("r-spams" ,r-spams)
8228 ("r-wgcna" ,r-wgcna)
8229 ("r-fastcluster" ,r-fastcluster)))
8230 (home-page "https://github.com/ManuSetty/SeqGL")
8231 (synopsis "Group lasso for Dnase/ChIP-seq data")
8232 (description "SeqGL is a group lasso based algorithm to extract
8233transcription factor sequence signals from ChIP, DNase and ATAC-seq profiles.
8234This package presents a method which uses group lasso to discriminate between
8235bound and non bound genomic regions to accurately identify transcription
8236factors bound at the specific regions.")
8237 (license license:gpl2+)))
8238
bd3be46e
RW		 8239(define-public r-gkmsvm
8240 (package
8241 (name "r-gkmsvm")
8242 (version "0.71.0")
8243 (source
8244 (origin
8245 (method url-fetch)
8246 (uri (cran-uri "gkmSVM" version))
8247 (sha256
8248 (base32
8249 "1zpxgxmf2nd5j5wn00ps6kfxr8wxh7d1swr1rr4spq7sj5z5z0k0"))))
8250 (properties `((upstream-name . "gkmSVM")))
8251 (build-system r-build-system)
8252 (propagated-inputs
8253 `(("r-biocgenerics" ,r-biocgenerics)
8254 ("r-biostrings" ,r-biostrings)
8255 ("r-genomeinfodb" ,r-genomeinfodb)
8256 ("r-genomicranges" ,r-genomicranges)
8257 ("r-iranges" ,r-iranges)
8258 ("r-kernlab" ,r-kernlab)
8259 ("r-rcpp" ,r-rcpp)
8260 ("r-rocr" ,r-rocr)
8261 ("r-rtracklayer" ,r-rtracklayer)
8262 ("r-s4vectors" ,r-s4vectors)
8263 ("r-seqinr" ,r-seqinr)))
8264 (home-page "http://cran.r-project.org/web/packages/gkmSVM")
8265 (synopsis "Gapped-kmer support vector machine")
8266 (description
8267 "This R package provides tools for training gapped-kmer SVM classifiers
8268for DNA and protein sequences. This package supports several sequence
8269kernels, including: gkmSVM, kmer-SVM, mismatch kernel and wildcard kernel.")
8270 (license license:gpl2+)))
8271
d4af25b5
RJPB		 8272(define-public r-tximport
8273 (package
8274 (name "r-tximport")
19e8929c 8275 (version "1.6.0")
d4af25b5
RJPB		 8276 (source (origin
8277 (method url-fetch)
8278 (uri (bioconductor-uri "tximport" version))
8279 (sha256
8280 (base32
19e8929c 8281 "1gyqcm91hxg1kgjqcz2qw1n56yp9pymjzs50rwcpb2893dr8sp2h"))))
d4af25b5 8282 (build-system r-build-system)
5713bbf1 8283 (home-page "https://bioconductor.org/packages/tximport")
d4af25b5
RJPB		 8284 (synopsis "Import and summarize transcript-level estimates for gene-level analysis")
8285 (description
8286 "This package provides tools to import transcript-level abundance,
8287estimated counts and transcript lengths, and to summarize them into matrices
8288for use with downstream gene-level analysis packages. Average transcript
8289length, weighted by sample-specific transcript abundance estimates, is
8290provided as a matrix which can be used as an offset for different expression
8291of gene-level counts.")
8292 (license license:gpl2+)))
8293
69f2b3bd
RJPB		 8294(define-public r-rhdf5
8295 (package
8296 (name "r-rhdf5")
e6b332d4 8297 (version "2.22.0")
69f2b3bd
RJPB		 8298 (source (origin
8299 (method url-fetch)
8300 (uri (bioconductor-uri "rhdf5" version))
8301 (sha256
8302 (base32
e6b332d4 8303 "145858qg1xan6imxcbprzq3yn3mdf532aahdr6cibvdjg47hs4c1"))))
69f2b3bd
RJPB		 8304 (build-system r-build-system)
8305 (arguments
8306 `(#:phases
8307 (modify-phases %standard-phases
8308 (add-after 'unpack 'unpack-smallhdf5
8309 (lambda* (#:key outputs #:allow-other-keys)
8310 (system* "tar" "-xzvf"
8311 "src/hdf5source/hdf5small.tgz" "-C" "src/" )
8312 (substitute* "src/Makevars"
8313 (("^.*cd hdf5source &&.*$") "")
8314 (("^.*gunzip -dc hdf5small.tgz.*$") "")
8315 (("^.*rm -rf hdf5.*$") "")
8316 (("^.*mv hdf5source/hdf5 ..*$") ""))
8317 (substitute* "src/hdf5/configure"
8318 (("/bin/mv") "mv"))
8319 #t)))))
8320 (propagated-inputs
8321 `(("r-zlibbioc" ,r-zlibbioc)))
8322 (inputs
8323 `(("perl" ,perl)
8324 ("zlib" ,zlib)))
5713bbf1 8325 (home-page "https://bioconductor.org/packages/rhdf5")
69f2b3bd
RJPB		 8326 (synopsis "HDF5 interface to R")
8327 (description
8328 "This R/Bioconductor package provides an interface between HDF5 and R.
8329HDF5's main features are the ability to store and access very large and/or
8330complex datasets and a wide variety of metadata on mass storage (disk) through
8331a completely portable file format. The rhdf5 package is thus suited for the
8332exchange of large and/or complex datasets between R and other software
8333package, and for letting R applications work on datasets that are larger than
8334the available RAM.")
8335 (license license:artistic2.0)))
8336
17cddc17
RW		 8337(define-public r-annotationfilter
8338 (package
8339 (name "r-annotationfilter")
598dacf6 8340 (version "1.2.0")
17cddc17
RW		 8341 (source (origin
8342 (method url-fetch)
8343 (uri (bioconductor-uri "AnnotationFilter" version))
8344 (sha256
8345 (base32
598dacf6 8346 "04zf864c1fvdlaay2r5cn30fc1n5i3czh31fs62qlrvs61wjiscs"))))
17cddc17
RW		 8347 (properties
8348 `((upstream-name . "AnnotationFilter")))
8349 (build-system r-build-system)
8350 (propagated-inputs
8351 `(("r-genomicranges" ,r-genomicranges)
8352 ("r-lazyeval" ,r-lazyeval)))
8353 (home-page "https://github.com/Bioconductor/AnnotationFilter")
8354 (synopsis "Facilities for filtering Bioconductor annotation resources")
8355 (description
8356 "This package provides classes and other infrastructure to implement
8357filters for manipulating Bioconductor annotation resources. The filters are
8358used by @code{ensembldb}, @code{Organism.dplyr}, and other packages.")
8359 (license license:artistic2.0)))
8360
66e40e00
RW		 8361(define-public emboss
8362 (package
8363 (name "emboss")
8364 (version "6.5.7")
8365 (source (origin
8366 (method url-fetch)
8367 (uri (string-append "ftp://emboss.open-bio.org/pub/EMBOSS/old/"
8368 (version-major+minor version) ".0/"
8369 "EMBOSS-" version ".tar.gz"))
8370 (sha256
8371 (base32
8372 "0vsmz96gc411yj2iyzdrsmg4l2n1nhgmp7vrgzlxx3xixv9xbf0q"))))
8373 (build-system gnu-build-system)
8374 (arguments
8375 `(#:configure-flags
8376 (list (string-append "--with-hpdf="
8377 (assoc-ref %build-inputs "libharu")))
8378 #:phases
8379 (modify-phases %standard-phases
8380 (add-after 'unpack 'fix-checks
8381 (lambda _
8382 ;; The PNGDRIVER tests check for the presence of libgd, libpng
8383 ;; and zlib, but assume that they are all found at the same
8384 ;; prefix.
8385 (substitute* "configure.in"
8386 (("CHECK_PNGDRIVER")
8387 "LIBS=\"$LIBS -lgd -lpng -lz -lm\"
8388AC_DEFINE([PLD_png], [1], [Define to 1 if PNG support is available])
8389AM_CONDITIONAL(AMPNG, true)"))
8390 #t))
d10092b8 8391 (add-after 'fix-checks 'disable-update-check
66e40e00
RW		 8392 (lambda _
8393 ;; At build time there is no connection to the Internet, so
8394 ;; looking for updates will not work.
8395 (substitute* "Makefile.am"
8396 (("\\$\\(bindir\\)/embossupdate") ""))
8397 #t))
d10092b8 8398 (add-after 'disable-update-check 'autogen
66e40e00
RW		 8399 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
8400 (inputs
8401 `(("perl" ,perl)
8402 ("libpng" ,libpng)
8403 ("gd" ,gd)
8404 ("libx11" ,libx11)
8405 ("libharu" ,libharu)
8406 ("zlib" ,zlib)))
8407 (native-inputs
8408 `(("autoconf" ,autoconf)
8409 ("automake" ,automake)
8410 ("libtool" ,libtool)
8411 ("pkg-config" ,pkg-config)))
8412 (home-page "http://emboss.sourceforge.net")
8413 (synopsis "Molecular biology analysis suite")
8414 (description "EMBOSS is the \"European Molecular Biology Open Software
8415Suite\". EMBOSS is an analysis package specially developed for the needs of
8416the molecular biology (e.g. EMBnet) user community. The software
8417automatically copes with data in a variety of formats and even allows
8418transparent retrieval of sequence data from the web. It also provides a
8419number of libraries for the development of software in the field of molecular
8420biology. EMBOSS also integrates a range of currently available packages and
8421tools for sequence analysis into a seamless whole.")
8422 (license license:gpl2+)))
8423
1f1b20b8
RW		 8424(define-public bits
8425 (let ((revision "1")
8426 (commit "3cc4567896d9d6442923da944beb704750a08d2d"))
8427 (package
8428 (name "bits")
8429 ;; The version is 2.13.0 even though no release archives have been
8430 ;; published as yet.
8431 (version (string-append "2.13.0-" revision "." (string-take commit 9)))
8432 (source (origin
8433 (method git-fetch)
8434 (uri (git-reference
8435 (url "https://github.com/arq5x/bits.git")
8436 (commit commit)))
8437 (file-name (string-append name "-" version "-checkout"))
8438 (sha256
8439 (base32
8440 "17n2kffk4kmhivd8c98g2vr6y1s23vbg4sxlxs689wni66797hbs"))))
8441 (build-system gnu-build-system)
8442 (arguments
8443 `(#:tests? #f ;no tests included
8444 #:phases
8445 (modify-phases %standard-phases
8446 (delete 'configure)
8447 (add-after 'unpack 'remove-cuda
8448 (lambda _
8449 (substitute* "Makefile"
8450 ((".*_cuda") "")
8451 (("(bits_test_intersections) \\\\" _ match) match))
8452 #t))
8453 (replace 'install
8454 (lambda* (#:key outputs #:allow-other-keys)
8455 (copy-recursively
8456 "bin" (string-append (assoc-ref outputs "out") "/bin"))
8457 #t)))))
8458 (inputs
8459 `(("gsl" ,gsl)
8460 ("zlib" ,zlib)))
8461 (home-page "https://github.com/arq5x/bits")
8462 (synopsis "Implementation of binary interval search algorithm")
8463 (description "This package provides an implementation of the
8464BITS (Binary Interval Search) algorithm, an approach to interval set
8465intersection. It is especially suited for the comparison of diverse genomic
8466datasets and the exploration of large datasets of genome
8467intervals (e.g. genes, sequence alignments).")
8468 (license license:gpl2))))
8469
e62ffce5 8470(define-public piranha
883302da
RW		 8471 ;; There is no release tarball for the latest version. The latest commit is
8472 ;; older than one year at the time of this writing.
8473 (let ((revision "1")
8474 (commit "0466d364b71117d01e4471b74c514436cc281233"))
8475 (package
8476 (name "piranha")
8477 (version (string-append "1.2.1-" revision "." (string-take commit 9)))
8478 (source (origin
8479 (method git-fetch)
8480 (uri (git-reference
8481 (url "https://github.com/smithlabcode/piranha.git")
8482 (commit commit)))
8483 (sha256
8484 (base32
8485 "117dc0zf20c61jam69sk4abl57ah6yi6i7qra7d7y5zrbgk12q5n"))))
8486 (build-system gnu-build-system)
8487 (arguments
8488 `(#:test-target "test"
8489 #:phases
8490 (modify-phases %standard-phases
8491 (add-after 'unpack 'copy-smithlab-cpp
8492 (lambda* (#:key inputs #:allow-other-keys)
e62ffce5 8493 (for-each (lambda (file)
883302da
RW		 8494 (install-file file "./src/smithlab_cpp/"))
8495 (find-files (assoc-ref inputs "smithlab-cpp")))
8496 #t))
8497 (add-after 'install 'install-to-store
8498 (lambda* (#:key outputs #:allow-other-keys)
8499 (let* ((out (assoc-ref outputs "out"))
8500 (bin (string-append out "/bin")))
883302da
RW		 8501 (for-each (lambda (file)
8502 (install-file file bin))
8503 (find-files "bin" ".*")))
8504 #t)))
8505 #:configure-flags
8506 (list (string-append "--with-bam_tools_headers="
8507 (assoc-ref %build-inputs "bamtools") "/include/bamtools")
8508 (string-append "--with-bam_tools_library="
8509 (assoc-ref %build-inputs "bamtools") "/lib/bamtools"))))
8510 (inputs
8511 `(("bamtools" ,bamtools)
8512 ("samtools" ,samtools-0.1)
8513 ("gsl" ,gsl)
8514 ("smithlab-cpp"
8515 ,(let ((commit "3723e2db438c51501d0423429ff396c3035ba46a"))
8516 (origin
8517 (method git-fetch)
8518 (uri (git-reference
8519 (url "https://github.com/smithlabcode/smithlab_cpp.git")
8520 (commit commit)))
8521 (file-name (string-append "smithlab_cpp-" commit "-checkout"))
8522 (sha256
8523 (base32
8524 "0l4gvbwslw5ngziskja41c00x1r06l3yidv7y0xw9djibhykzy0g")))))))
8525 (native-inputs
8526 `(("python" ,python-2)))
8527 (home-page "https://github.com/smithlabcode/piranha")
8528 (synopsis "Peak-caller for CLIP-seq and RIP-seq data")
8529 (description
8530 "Piranha is a peak-caller for genomic data produced by CLIP-seq and
e62ffce5
RW		 8531RIP-seq experiments. It takes input in BED or BAM format and identifies
8532regions of statistically significant read enrichment. Additional covariates
8533may optionally be provided to further inform the peak-calling process.")
883302da 8534 (license license:gpl3+))))
e62ffce5 8535
d1e32822
RW		 8536(define-public pepr
8537 (package
8538 (name "pepr")
8539 (version "1.0.9")
8540 (source (origin
8541 (method url-fetch)
8542 (uri (string-append "https://pypi.python.org/packages/source/P"
8543 "/PePr/PePr-" version ".tar.gz"))
8544 (sha256
8545 (base32
8546 "0qxjfdpl1b1y53nccws2d85f6k74zwmx8y8sd9rszcqhfayx6gdx"))))
8547 (build-system python-build-system)
8548 (arguments
8549 `(#:python ,python-2 ; python2 only
b41a05ce 8550 #:tests? #f)) ; no tests included
d1e32822
RW		 8551 (propagated-inputs
8552 `(("python2-numpy" ,python2-numpy)
8553 ("python2-scipy" ,python2-scipy)
8554 ("python2-pysam" ,python2-pysam)))
0c6c9c00 8555 (home-page "https://github.com/shawnzhangyx/PePr")
d1e32822
RW		 8556 (synopsis "Peak-calling and prioritization pipeline for ChIP-Seq data")
8557 (description
8558 "PePr is a ChIP-Seq peak calling or differential binding analysis tool
8559that is primarily designed for data with biological replicates. It uses a
8560negative binomial distribution to model the read counts among the samples in
8561the same group, and look for consistent differences between ChIP and control
8562group or two ChIP groups run under different conditions.")
8563 (license license:gpl3+)))
6b49a37e
RJ		 8564
8565(define-public filevercmp
8566 (let ((commit "1a9b779b93d0b244040274794d402106907b71b7"))
8567 (package
8568 (name "filevercmp")
8569 (version (string-append "0-1." (string-take commit 7)))
8570 (source (origin
8571 (method url-fetch)
8572 (uri (string-append "https://github.com/ekg/filevercmp/archive/"
8573 commit ".tar.gz"))
8574 (file-name (string-append name "-" version ".tar.gz"))
8575 (sha256
8576 (base32 "0yp5jswf5j2pqc6517x277s4s6h1ss99v57kxw9gy0jkfl3yh450"))))
8577 (build-system gnu-build-system)
8578 (arguments
8579 `(#:tests? #f ; There are no tests to run.
8580 #:phases
8581 (modify-phases %standard-phases
8582 (delete 'configure) ; There is no configure phase.
8583 (replace 'install
8584 (lambda* (#:key outputs #:allow-other-keys)
8585 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
8586 (install-file "filevercmp" bin)))))))
8587 (home-page "https://github.com/ekg/filevercmp")
8588 (synopsis "This program compares version strings")
8589 (description "This program compares version strings. It intends to be a
8590replacement for strverscmp.")
8591 (license license:gpl3+))))
5fb5dffb
RW		 8592
8593(define-public multiqc
8594 (package
8595 (name "multiqc")
66865ab8 8596 (version "1.3")
5fb5dffb
RW		 8597 (source
8598 (origin
8599 (method url-fetch)
8600 (uri (pypi-uri "multiqc" version))
8601 (sha256
8602 (base32
66865ab8 8603 "0fx1sx53znbgzfhbbiyd8j6cg5llpcsl5q5c45jy2c81d12piqfd"))))
5fb5dffb
RW		 8604 (build-system python-build-system)
8605 (propagated-inputs
66865ab8
RW		 8606 `(("python-enum34" ,python-enum34)
8607 ("python-jinja2" ,python-jinja2)
5fb5dffb
RW		 8608 ("python-simplejson" ,python-simplejson)
8609 ("python-pyyaml" ,python-pyyaml)
8610 ("python-click" ,python-click)
799247d2
RW		 8611 ("python-spectra" ,python-spectra)
8612 ("python-requests" ,python-requests)
8613 ("python-markdown" ,python-markdown)
8614 ("python-lzstring" ,python-lzstring)
5fb5dffb 8615 ("python-matplotlib" ,python-matplotlib)
1778ea74
RW		 8616 ("python-numpy" ,python-numpy)
8617 ;; MultQC checks for the presence of nose at runtime.
8618 ("python-nose" ,python-nose)))
5fb5dffb
RW		 8619 (home-page "http://multiqc.info")
8620 (synopsis "Aggregate bioinformatics analysis reports")
8621 (description
8622 "MultiQC is a tool to aggregate bioinformatics results across many
8623samples into a single report. It contains modules for a large number of
8624common bioinformatics tools.")
66865ab8 8625 (license license:gpl3+)))
6193563a
RW		 8626
8627(define-public r-chipseq
8628 (package
8629 (name "r-chipseq")
eeaa6967 8630 (version "1.28.0")
6193563a
RW		 8631 (source
8632 (origin
8633 (method url-fetch)
8634 (uri (bioconductor-uri "chipseq" version))
8635 (sha256
8636 (base32
eeaa6967 8637 "1ymcq77krwjzrkzzcw7i9909cmkqa7c0675z9wzvrrk81hgdssfq"))))
6193563a
RW		 8638 (build-system r-build-system)
8639 (propagated-inputs
8640 `(("r-biocgenerics" ,r-biocgenerics)
8641 ("r-genomicranges" ,r-genomicranges)
8642 ("r-iranges" ,r-iranges)
eeaa6967 8643 ("r-lattice" ,r-lattice)
6193563a
RW		 8644 ("r-s4vectors" ,r-s4vectors)
8645 ("r-shortread" ,r-shortread)))
5713bbf1 8646 (home-page "https://bioconductor.org/packages/chipseq")
6193563a
RW		 8647 (synopsis "Package for analyzing ChIPseq data")
8648 (description
8649 "This package provides tools for processing short read data from ChIPseq
8650experiments.")
8651 (license license:artistic2.0)))
d407bdb9
RW		 8652
8653(define-public r-copyhelper
8654 (package
8655 (name "r-copyhelper")
8656 (version "1.6.0")
8657 (source
8658 (origin
8659 (method url-fetch)
5713bbf1 8660 (uri (string-append "https://bioconductor.org/packages/release/"
d407bdb9
RW		 8661 "data/experiment/src/contrib/CopyhelpeR_"
8662 version ".tar.gz"))
8663 (sha256
8664 (base32
8665 "0x7cyynjmxls9as2gg0iyp9x5fpalxmdjq914ss7i84i9zyk5bhq"))))
8666 (properties `((upstream-name . "CopyhelpeR")))
8667 (build-system r-build-system)
5713bbf1 8668 (home-page "https://bioconductor.org/packages/CopyhelpeR/")
d407bdb9
RW		 8669 (synopsis "Helper files for CopywriteR")
8670 (description
8671 "This package contains the helper files that are required to run the
8672Bioconductor package CopywriteR. It contains pre-assembled 1kb bin GC-content
8673and mappability files for the reference genomes hg18, hg19, hg38, mm9 and
8674mm10. In addition, it contains a blacklist filter to remove regions that
8675display copy number variation. Files are stored as GRanges objects from the
8676GenomicRanges Bioconductor package.")
8677 (license license:gpl2)))
3a3bf2f8
RW		 8678
8679(define-public r-copywriter
8680 (package
8681 (name "r-copywriter")
c126ba69 8682 (version "2.10.0")
3a3bf2f8
RW		 8683 (source
8684 (origin
8685 (method url-fetch)
8686 (uri (bioconductor-uri "CopywriteR" version))
8687 (sha256
8688 (base32
c126ba69 8689 "17fy2lc5yf3nh6v077kv87h53n263hqz2540lzrl0vjiqrl2plca"))))
3a3bf2f8
RW		 8690 (properties `((upstream-name . "CopywriteR")))
8691 (build-system r-build-system)
8692 (propagated-inputs
8693 `(("r-biocparallel" ,r-biocparallel)
8694 ("r-chipseq" ,r-chipseq)
8695 ("r-copyhelper" ,r-copyhelper)
8696 ("r-data-table" ,r-data-table)
8697 ("r-dnacopy" ,r-dnacopy)
8698 ("r-futile-logger" ,r-futile-logger)
8699 ("r-genomeinfodb" ,r-genomeinfodb)
8700 ("r-genomicalignments" ,r-genomicalignments)
8701 ("r-genomicranges" ,r-genomicranges)
8702 ("r-gtools" ,r-gtools)
8703 ("r-iranges" ,r-iranges)
8704 ("r-matrixstats" ,r-matrixstats)
8705 ("r-rsamtools" ,r-rsamtools)
8706 ("r-s4vectors" ,r-s4vectors)))
8707 (home-page "https://github.com/PeeperLab/CopywriteR")
8708 (synopsis "Copy number information from targeted sequencing")
8709 (description
8710 "CopywriteR extracts DNA copy number information from targeted sequencing
8711by utilizing off-target reads. It allows for extracting uniformly distributed
8712copy number information, can be used without reference, and can be applied to
8713sequencing data obtained from various techniques including chromatin
8714immunoprecipitation and target enrichment on small gene panels. Thereby,
8715CopywriteR constitutes a widely applicable alternative to available copy
8716number detection tools.")
8717 (license license:gpl2)))
dd42a330 8718
bf3fa996
RW		 8719(define-public r-methylkit
8720 (package
8721 (name "r-methylkit")
91e18679 8722 (version "1.4.0")
bf3fa996
RW		 8723 (source (origin
8724 (method url-fetch)
8725 (uri (bioconductor-uri "methylKit" version))
8726 (sha256
8727 (base32
91e18679 8728 "0h53w2mrjrg2n0ndi12k9j6cwclgwcgpy25nz7nyj971aisw02xn"))))
bf3fa996
RW		 8729 (properties `((upstream-name . "methylKit")))
8730 (build-system r-build-system)
8731 (propagated-inputs
8732 `(("r-data-table" ,r-data-table)
8733 ("r-emdbook" ,r-emdbook)
8734 ("r-fastseg" ,r-fastseg)
8735 ("r-genomeinfodb" ,r-genomeinfodb)
8736 ("r-genomicranges" ,r-genomicranges)
8737 ("r-gtools" ,r-gtools)
8738 ("r-iranges" ,r-iranges)
8739 ("r-kernsmooth" ,r-kernsmooth)
8740 ("r-limma" ,r-limma)
8741 ("r-mclust" ,r-mclust)
8742 ("r-qvalue" ,r-qvalue)
8743 ("r-r-utils" ,r-r-utils)
8744 ("r-rcpp" ,r-rcpp)
8745 ("r-rhtslib" ,r-rhtslib)
8746 ("r-rsamtools" ,r-rsamtools)
8747 ("r-rtracklayer" ,r-rtracklayer)
8748 ("r-s4vectors" ,r-s4vectors)
8749 ("r-zlibbioc" ,r-zlibbioc)))
8750 (inputs
8751 `(("zlib" ,zlib)))
11ed4108 8752 (home-page "https://github.com/al2na/methylKit")
bf3fa996
RW		 8753 (synopsis
8754 "DNA methylation analysis from high-throughput bisulfite sequencing results")
8755 (description
8756 "MethylKit is an R package for DNA methylation analysis and annotation
8757from high-throughput bisulfite sequencing. The package is designed to deal
8758with sequencing data from @dfn{Reduced representation bisulfite
8759sequencing} (RRBS) and its variants, but also target-capture methods and whole
8760genome bisulfite sequencing. It also has functions to analyze base-pair
8761resolution 5hmC data from experimental protocols such as oxBS-Seq and
8762TAB-Seq.")
8763 (license license:artistic2.0)))
8764
dd42a330
RJPB		 8765(define-public r-sva
8766 (package
8767 (name "r-sva")
fad7c0c7 8768 (version "3.26.0")
dd42a330
RJPB		 8769 (source
8770 (origin
8771 (method url-fetch)
8772 (uri (bioconductor-uri "sva" version))
8773 (sha256
8774 (base32
fad7c0c7 8775 "0q5xb68wfcnchy8rkv5ma67pmz1i91lsnvmwmj8f1c3w4xan3pgw"))))
dd42a330
RJPB		 8776 (build-system r-build-system)
8777 (propagated-inputs
aeb64f3c 8778 `(("r-genefilter" ,r-genefilter)
bfa0d1e7
RJ		 8779 ("r-mgcv" ,r-mgcv)
8780 ("r-biocparallel" ,r-biocparallel)
8781 ("r-matrixstats" ,r-matrixstats)
8782 ("r-limma" ,r-limma)))
5713bbf1 8783 (home-page "https://bioconductor.org/packages/sva")
dd42a330
RJPB		 8784 (synopsis "Surrogate variable analysis")
8785 (description
8786 "This package contains functions for removing batch effects and other
8787unwanted variation in high-throughput experiment. It also contains functions
8788for identifying and building surrogate variables for high-dimensional data
8789sets. Surrogate variables are covariates constructed directly from
8790high-dimensional data like gene expression/RNA sequencing/methylation/brain
8791imaging data that can be used in subsequent analyses to adjust for unknown,
8792unmodeled, or latent sources of noise.")
8793 (license license:artistic2.0)))
fb1e528e
RW		 8794
8795(define-public r-seqminer
8796 (package
8797 (name "r-seqminer")
5e0de298 8798 (version "6.0")
fb1e528e
RW		 8799 (source
8800 (origin
8801 (method url-fetch)
8802 (uri (cran-uri "seqminer" version))
8803 (sha256
8804 (base32
5e0de298 8805 "057j1l6dip35l1aivilapl2zv9db677b3di2pb3sfgq2sxg0ps3l"))))
fb1e528e
RW		 8806 (build-system r-build-system)
8807 (inputs
8808 `(("zlib" ,zlib)))
8809 (home-page "http://seqminer.genomic.codes")
8810 (synopsis "Read nucleotide sequence data (VCF, BCF, and METAL formats)")
8811 (description
8812 "This package provides tools to integrate nucleotide sequencing
8813data (variant call format, e.g. VCF or BCF) or meta-analysis results in R.")
8814 ;; Any version of the GPL is acceptable
8815 (license (list license:gpl2+ license:gpl3+))))
1f40e73c
RW		 8816
8817(define-public r-raremetals2
8818 (package
8819 (name "r-raremetals2")
8820 (version "0.1")
8821 (source
8822 (origin
8823 (method url-fetch)
8824 (uri (string-append "http://genome.sph.umich.edu/w/images/"
8825 "b/b7/RareMETALS2_" version ".tar.gz"))
8826 (sha256
8827 (base32
8828 "0z5ljcgvnm06ja9lm85a3cniq7slxcy37aqqkxrdidr79an5fs4s"))))
8829 (properties `((upstream-name . "RareMETALS2")))
8830 (build-system r-build-system)
8831 (propagated-inputs
8832 `(("r-seqminer" ,r-seqminer)
8833 ("r-mvtnorm" ,r-mvtnorm)
aeb64f3c 8834 ("r-mass" ,r-mass)
1f40e73c
RW		 8835 ("r-compquadform" ,r-compquadform)
8836 ("r-getopt" ,r-getopt)))
8837 (home-page "http://genome.sph.umich.edu/wiki/RareMETALS2")
8838 (synopsis "Analyze gene-level association tests for binary trait")
8839 (description
8840 "The R package rareMETALS2 is an extension of the R package rareMETALS.
8841It was designed to meta-analyze gene-level association tests for binary trait.
8842While rareMETALS offers a near-complete solution for meta-analysis of
8843gene-level tests for quantitative trait, it does not offer the optimal
8844solution for binary trait. The package rareMETALS2 offers improved features
8845for analyzing gene-level association tests in meta-analyses for binary
8846trait.")
8847 (license license:gpl3)))
0e7d058e
RW		 8848
8849(define-public r-maldiquant
8850 (package
8851 (name "r-maldiquant")
7b92acdf 8852 (version "1.16.4")
0e7d058e
RW		 8853 (source
8854 (origin
8855 (method url-fetch)
8856 (uri (cran-uri "MALDIquant" version))
8857 (sha256
8858 (base32
7b92acdf 8859 "1pmhsfvd45a44xdiml4zx3zd5fhygqyziqvygahkk9yibnyhv4cv"))))
0e7d058e
RW		 8860 (properties `((upstream-name . "MALDIquant")))
8861 (build-system r-build-system)
8862 (home-page "http://cran.r-project.org/web/packages/MALDIquant")
8863 (synopsis "Quantitative analysis of mass spectrometry data")
8864 (description
8865 "This package provides a complete analysis pipeline for matrix-assisted
8866laser desorption/ionization-time-of-flight (MALDI-TOF) and other
8867two-dimensional mass spectrometry data. In addition to commonly used plotting
8868and processing methods it includes distinctive features, namely baseline
8869subtraction methods such as morphological filters (TopHat) or the
8870statistics-sensitive non-linear iterative peak-clipping algorithm (SNIP), peak
8871alignment using warping functions, handling of replicated measurements as well
8872as allowing spectra with different resolutions.")
8873 (license license:gpl3+)))
ae262174
RW		 8874
8875(define-public r-protgenerics
8876 (package
8877 (name "r-protgenerics")
9a665dfb 8878 (version "1.10.0")
ae262174
RW		 8879 (source
8880 (origin
8881 (method url-fetch)
8882 (uri (bioconductor-uri "ProtGenerics" version))
8883 (sha256
8884 (base32
9a665dfb 8885 "16ijp50448wnabp43klx943rhdvh7x45hvy7cnpq1s4dckxhhyni"))))
ae262174
RW		 8886 (properties `((upstream-name . "ProtGenerics")))
8887 (build-system r-build-system)
8888 (home-page "https://github.com/lgatto/ProtGenerics")
8889 (synopsis "S4 generic functions for proteomics infrastructure")
8890 (description
8891 "This package provides S4 generic functions needed by Bioconductor
8892proteomics packages.")
8893 (license license:artistic2.0)))
71676a1f
RW		 8894
8895(define-public r-mzr
8896 (package
8897 (name "r-mzr")
34b96754 8898 (version "2.12.0")
71676a1f
RW		 8899 (source
8900 (origin
8901 (method url-fetch)
8902 (uri (bioconductor-uri "mzR" version))
8903 (sha256
8904 (base32
15ca1959
RW		 8905 "1x3gp30sfxz2v3k3swih9kff9b2rvk7hzhnlkp6ywlnn2wgb0q8c"))
8906 (modules '((guix build utils)))
8907 (snippet
8908 '(begin
8909 (delete-file-recursively "src/boost")
8910 #t))))
71676a1f
RW		 8911 (properties `((upstream-name . "mzR")))
8912 (build-system r-build-system)
15ca1959
RW		 8913 (arguments
8914 `(#:phases
8915 (modify-phases %standard-phases
8916 (add-after 'unpack 'use-system-boost
8917 (lambda _
8918 (substitute* "src/Makevars"
8919 (("\\./boost/libs.*") "")
8920 (("ARCH_OBJS=" line)
8921 (string-append line
8922 "\nARCH_LIBS=-lboost_system -lboost_regex \
8923-lboost_iostreams -lboost_thread -lboost_filesystem -lboost_chrono\n")))
8924 #t)))))
71676a1f 8925 (inputs
34b96754 8926 `(("boost" ,boost) ; use this instead of the bundled boost sources
d534f3a4 8927 ("netcdf" ,netcdf)))
71676a1f
RW		 8928 (propagated-inputs
8929 `(("r-biobase" ,r-biobase)
8930 ("r-biocgenerics" ,r-biocgenerics)
8931 ("r-protgenerics" ,r-protgenerics)
8932 ("r-rcpp" ,r-rcpp)
8933 ("r-zlibbioc" ,r-zlibbioc)))
8934 (home-page "https://github.com/sneumann/mzR/")
8935 (synopsis "Parser for mass spectrometry data files")
8936 (description
8937 "The mzR package provides a unified API to the common file formats and
8938parsers available for mass spectrometry data. It comes with a wrapper for the
8939ISB random access parser for mass spectrometry mzXML, mzData and mzML files.
8940The package contains the original code written by the ISB, and a subset of the
8941proteowizard library for mzML and mzIdentML. The netCDF reading code has
8942previously been used in XCMS.")
8943 (license license:artistic2.0)))
5c9d1505
RW		 8944
8945(define-public r-affyio
8946 (package
8947 (name "r-affyio")
8a8138d2 8948 (version "1.48.0")
5c9d1505
RW		 8949 (source
8950 (origin
8951 (method url-fetch)
8952 (uri (bioconductor-uri "affyio" version))
8953 (sha256
8954 (base32
8a8138d2 8955 "1pzzp3d3dbmyf34gvivfiprkpscn36rgvhrq853a1d3avcwr5ak9"))))
5c9d1505
RW		 8956 (build-system r-build-system)
8957 (propagated-inputs
8958 `(("r-zlibbioc" ,r-zlibbioc)))
8959 (inputs
8960 `(("zlib" ,zlib)))
8961 (home-page "https://github.com/bmbolstad/affyio")
8962 (synopsis "Tools for parsing Affymetrix data files")
8963 (description
8964 "This package provides routines for parsing Affymetrix data files based
8965upon file format information. The primary focus is on accessing the CEL and
8966CDF file formats.")
8967 (license license:lgpl2.0+)))
05c7e5fb
RW		 8968
8969(define-public r-affy
8970 (package
8971 (name "r-affy")
dad5f838 8972 (version "1.56.0")
05c7e5fb
RW		 8973 (source
8974 (origin
8975 (method url-fetch)
8976 (uri (bioconductor-uri "affy" version))
8977 (sha256
8978 (base32
dad5f838 8979 "0jmbkimma5ffsdkk3xp03g4lpz84gd95nkqakif2nqq6wmx0syrj"))))
05c7e5fb
RW		 8980 (build-system r-build-system)
8981 (propagated-inputs
8982 `(("r-affyio" ,r-affyio)
8983 ("r-biobase" ,r-biobase)
8984 ("r-biocgenerics" ,r-biocgenerics)
8985 ("r-biocinstaller" ,r-biocinstaller)
8986 ("r-preprocesscore" ,r-preprocesscore)
8987 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 8988 (home-page "https://bioconductor.org/packages/affy")
05c7e5fb
RW		 8989 (synopsis "Methods for affymetrix oligonucleotide arrays")
8990 (description
8991 "This package contains functions for exploratory oligonucleotide array
8992analysis.")
8993 (license license:lgpl2.0+)))
7c08afaf
RW		 8994
8995(define-public r-vsn
8996 (package
8997 (name "r-vsn")
1be0f536 8998 (version "3.46.0")
7c08afaf
RW		 8999 (source
9000 (origin
9001 (method url-fetch)
9002 (uri (bioconductor-uri "vsn" version))
9003 (sha256
9004 (base32
1be0f536 9005 "18y62phzirj75gg6v5l41jwybmk23ia6w7qhch0kxc4bl2rysw6j"))))
7c08afaf
RW		 9006 (build-system r-build-system)
9007 (propagated-inputs
9008 `(("r-affy" ,r-affy)
9009 ("r-biobase" ,r-biobase)
9010 ("r-ggplot2" ,r-ggplot2)
aeb64f3c 9011 ("r-lattice" ,r-lattice)
7c08afaf 9012 ("r-limma" ,r-limma)))
5713bbf1 9013 (home-page "https://bioconductor.org/packages/release/bioc/html/vsn.html")
7c08afaf
RW		 9014 (synopsis "Variance stabilization and calibration for microarray data")
9015 (description
9016 "The package implements a method for normalising microarray intensities,
9017and works for single- and multiple-color arrays. It can also be used for data
9018from other technologies, as long as they have similar format. The method uses
9019a robust variant of the maximum-likelihood estimator for an
9020additive-multiplicative error model and affine calibration. The model
9021incorporates data calibration step (a.k.a. normalization), a model for the
9022dependence of the variance on the mean intensity and a variance stabilizing
9023data transformation. Differences between transformed intensities are
9024analogous to \"normalized log-ratios\". However, in contrast to the latter,
9025their variance is independent of the mean, and they are usually more sensitive
9026and specific in detecting differential transcription.")
9027 (license license:artistic2.0)))
4aa7d592
RW		 9028
9029(define-public r-mzid
9030 (package
9031 (name "r-mzid")
b500dbcd 9032 (version "1.16.0")
4aa7d592
RW		 9033 (source
9034 (origin
9035 (method url-fetch)
9036 (uri (bioconductor-uri "mzID" version))
9037 (sha256
9038 (base32
b500dbcd 9039 "0yk70dka56zd8w62f03ggx3mandj91gfa767h9ajj0sd3mjmfqb9"))))
4aa7d592
RW		 9040 (properties `((upstream-name . "mzID")))
9041 (build-system r-build-system)
9042 (propagated-inputs
9043 `(("r-doparallel" ,r-doparallel)
9044 ("r-foreach" ,r-foreach)
9045 ("r-iterators" ,r-iterators)
9046 ("r-plyr" ,r-plyr)
9047 ("r-protgenerics" ,r-protgenerics)
9048 ("r-rcpp" ,r-rcpp)
9049 ("r-xml" ,r-xml)))
5713bbf1 9050 (home-page "https://bioconductor.org/packages/mzID")
4aa7d592
RW		 9051 (synopsis "Parser for mzIdentML files")
9052 (description
9053 "This package provides a parser for mzIdentML files implemented using the
9054XML package. The parser tries to be general and able to handle all types of
9055mzIdentML files with the drawback of having less pretty output than a vendor
9056specific parser.")
9057 (license license:gpl2+)))
6a67e181
RW		 9058
9059(define-public r-pcamethods
9060 (package
9061 (name "r-pcamethods")
a998c9a4 9062 (version "1.70.0")
6a67e181
RW		 9063 (source
9064 (origin
9065 (method url-fetch)
9066 (uri (bioconductor-uri "pcaMethods" version))
9067 (sha256
9068 (base32
a998c9a4 9069 "0ii235g0x0492kh8cfrf28ni0b6vd6fh7kizkqmczzqggd6b1bk8"))))
6a67e181
RW		 9070 (properties `((upstream-name . "pcaMethods")))
9071 (build-system r-build-system)
9072 (propagated-inputs
9073 `(("r-biobase" ,r-biobase)
9074 ("r-biocgenerics" ,r-biocgenerics)
aeb64f3c 9075 ("r-mass" ,r-mass)
6a67e181
RW		 9076 ("r-rcpp" ,r-rcpp)))
9077 (home-page "https://github.com/hredestig/pcamethods")
9078 (synopsis "Collection of PCA methods")
9079 (description
9080 "This package provides Bayesian PCA, Probabilistic PCA, Nipals PCA,
9081Inverse Non-Linear PCA and the conventional SVD PCA. A cluster based method
9082for missing value estimation is included for comparison. BPCA, PPCA and
9083NipalsPCA may be used to perform PCA on incomplete data as well as for
9084accurate missing value estimation. A set of methods for printing and plotting
9085the results is also provided. All PCA methods make use of the same data
9086structure (pcaRes) to provide a common interface to the PCA results.")
9087 (license license:gpl3+)))
11879284
RW		 9088
9089(define-public r-msnbase
9090 (package
9091 (name "r-msnbase")
baa00ee9 9092 (version "2.4.0")
11879284
RW		 9093 (source
9094 (origin
9095 (method url-fetch)
9096 (uri (bioconductor-uri "MSnbase" version))
9097 (sha256
9098 (base32
baa00ee9 9099 "0dqfimljhrx3gac8d1k72gppx27lz8yckyb12v4604nbviw7xd3r"))))
11879284
RW		 9100 (properties `((upstream-name . "MSnbase")))
9101 (build-system r-build-system)
9102 (propagated-inputs
9103 `(("r-affy" ,r-affy)
9104 ("r-biobase" ,r-biobase)
9105 ("r-biocgenerics" ,r-biocgenerics)
9106 ("r-biocparallel" ,r-biocparallel)
9107 ("r-digest" ,r-digest)
9108 ("r-ggplot2" ,r-ggplot2)
9109 ("r-impute" ,r-impute)
9110 ("r-iranges" ,r-iranges)
baa00ee9 9111 ("r-lattice" ,r-lattice)
11879284
RW		 9112 ("r-maldiquant" ,r-maldiquant)
9113 ("r-mzid" ,r-mzid)
9114 ("r-mzr" ,r-mzr)
9115 ("r-pcamethods" ,r-pcamethods)
9116 ("r-plyr" ,r-plyr)
9117 ("r-preprocesscore" ,r-preprocesscore)
9118 ("r-protgenerics" ,r-protgenerics)
9119 ("r-rcpp" ,r-rcpp)
11879284
RW		 9120 ("r-s4vectors" ,r-s4vectors)
9121 ("r-vsn" ,r-vsn)
9122 ("r-xml" ,r-xml)))
9123 (home-page "https://github.com/lgatto/MSnbase")
9124 (synopsis "Base functions and classes for MS-based proteomics")
9125 (description
9126 "This package provides basic plotting, data manipulation and processing
e614d639 9127of mass spectrometry based proteomics data.")
11879284 9128 (license license:artistic2.0)))
66da3a9f
RW		 9129
9130(define-public r-msnid
9131 (package
9132 (name "r-msnid")
22752bef 9133 (version "1.11.0")
66da3a9f
RW		 9134 (source
9135 (origin
9136 (method url-fetch)
9137 (uri (bioconductor-uri "MSnID" version))
9138 (sha256
9139 (base32
22752bef 9140 "1vi4ngwbayrv2jkfb4pbmdp37xn04y07rh1jcklqfh0fcrm1jdig"))))
66da3a9f
RW		 9141 (properties `((upstream-name . "MSnID")))
9142 (build-system r-build-system)
9143 (propagated-inputs
9144 `(("r-biobase" ,r-biobase)
9145 ("r-data-table" ,r-data-table)
9146 ("r-doparallel" ,r-doparallel)
9147 ("r-dplyr" ,r-dplyr)
9148 ("r-foreach" ,r-foreach)
9149 ("r-iterators" ,r-iterators)
9150 ("r-msnbase" ,r-msnbase)
9151 ("r-mzid" ,r-mzid)
9152 ("r-mzr" ,r-mzr)
9153 ("r-protgenerics" ,r-protgenerics)
9154 ("r-r-cache" ,r-r-cache)
9155 ("r-rcpp" ,r-rcpp)
9156 ("r-reshape2" ,r-reshape2)))
5713bbf1 9157 (home-page "https://bioconductor.org/packages/MSnID")
66da3a9f
RW		 9158 (synopsis "Utilities for LC-MSn proteomics identifications")
9159 (description
2923f3e5
RW		 9160 "This package extracts @dfn{tandem mass spectrometry} (MS/MS) ID data
9161from mzIdentML (leveraging the mzID package) or text files. After collating
9162the search results from multiple datasets it assesses their identification
9163quality and optimize filtering criteria to achieve the maximum number of
9164identifications while not exceeding a specified false discovery rate. It also
9165contains a number of utilities to explore the MS/MS results and assess missed
9166and irregular enzymatic cleavages, mass measurement accuracy, etc.")
66da3a9f 9167 (license license:artistic2.0)))
5ded35d8 9168
2441c284
RJPB		 9169(define-public r-seurat
9170 ;; Source releases are only made for new x.0 versions. All newer versions
9171 ;; are only released as pre-built binaries. At the time of this writing the
9172 ;; latest binary release is 1.4.0.12, which is equivalent to this commit.
9173 (let ((commit "fccb77d1452c35ee47e47ebf8e87bddb59f3b08d")
9174 (revision "1"))
9175 (package
9176 (name "r-seurat")
9177 (version (string-append "1.4.0.12-" revision "." (string-take commit 7)))
9178 (source (origin
9179 (method git-fetch)
9180 (uri (git-reference
9181 (url "https://github.com/satijalab/seurat")
9182 (commit commit)))
9183 (file-name (string-append name "-" version "-checkout"))
9184 (sha256
9185 (base32
9186 "101wq3aqrdmbfi3lqmq4iivk9iwbf10d4z216ss25hf7n9091cyl"))
9187 ;; Delete pre-built jar.
9188 (snippet
9189 '(begin (delete-file "inst/java/ModularityOptimizer.jar")
9190 #t))))
9191 (build-system r-build-system)
9192 (arguments
9193 `(#:phases
9194 (modify-phases %standard-phases
9195 (add-after 'unpack 'build-jar
9196 (lambda* (#:key inputs #:allow-other-keys)
9197 (let ((classesdir "tmp-classes"))
9198 (setenv "JAVA_HOME" (assoc-ref inputs "jdk"))
9199 (mkdir classesdir)
9200 (and (zero? (apply system* `("javac" "-d" ,classesdir
9201 ,@(find-files "java" "\\.java$"))))
9202 (zero? (system* "jar"
9203 "-cf" "inst/java/ModularityOptimizer.jar"
9204 "-C" classesdir ".")))))))))
9205 (native-inputs
9206 `(("jdk" ,icedtea "jdk")))
9207 (propagated-inputs
9208 `(("r-ape" ,r-ape)
9209 ("r-caret" ,r-caret)
9210 ("r-cowplot" ,r-cowplot)
9211 ("r-dplyr" ,r-dplyr)
9212 ("r-fastica" ,r-fastica)
9213 ("r-fnn" ,r-fnn)
9214 ("r-fpc" ,r-fpc)
9215 ("r-gdata" ,r-gdata)
9216 ("r-ggplot2" ,r-ggplot2)
9217 ("r-gplots" ,r-gplots)
9218 ("r-gridextra" ,r-gridextra)
9219 ("r-igraph" ,r-igraph)
9220 ("r-irlba" ,r-irlba)
9221 ("r-lars" ,r-lars)
9222 ("r-mixtools" ,r-mixtools)
9223 ("r-pbapply" ,r-pbapply)
9224 ("r-plyr" ,r-plyr)
9225 ("r-ranger" ,r-ranger)
9226 ("r-rcolorbrewer" ,r-rcolorbrewer)
9227 ("r-rcpp" ,r-rcpp)
9228 ("r-rcppeigen" ,r-rcppeigen)
9229 ("r-rcppprogress" ,r-rcppprogress)
9230 ("r-reshape2" ,r-reshape2)
9231 ("r-rocr" ,r-rocr)
9232 ("r-rtsne" ,r-rtsne)
9233 ("r-stringr" ,r-stringr)
9234 ("r-tclust" ,r-tclust)
9235 ("r-tsne" ,r-tsne)
9236 ("r-vgam" ,r-vgam)))
9237 (home-page "http://www.satijalab.org/seurat")
9238 (synopsis "Seurat is an R toolkit for single cell genomics")
9239 (description
9240 "This package is an R package designed for QC, analysis, and
9241exploration of single cell RNA-seq data. It easily enables widely-used
9242analytical techniques, including the identification of highly variable genes,
9243dimensionality reduction; PCA, ICA, t-SNE, standard unsupervised clustering
9244algorithms; density clustering, hierarchical clustering, k-means, and the
9245discovery of differentially expressed genes and markers.")
9246 (license license:gpl3))))
9247
d4b83651
RW		 9248(define-public r-aroma-light
9249 (package
9250 (name "r-aroma-light")
531e8090 9251 (version "3.8.0")
d4b83651
RW		 9252 (source
9253 (origin
9254 (method url-fetch)
9255 (uri (bioconductor-uri "aroma.light" version))
9256 (sha256
9257 (base32
531e8090 9258 "0crnk6851jwypqr5l5jcbbay0vi5vvdjyisaf6z2d69c39wmr6sc"))))
d4b83651
RW		 9259 (properties `((upstream-name . "aroma.light")))
9260 (build-system r-build-system)
9261 (propagated-inputs
9262 `(("r-matrixstats" ,r-matrixstats)
9263 ("r-r-methodss3" ,r-r-methodss3)
9264 ("r-r-oo" ,r-r-oo)
9265 ("r-r-utils" ,r-r-utils)))
9266 (home-page "https://github.com/HenrikBengtsson/aroma.light")
9267 (synopsis "Methods for normalization and visualization of microarray data")
9268 (description
9269 "This package provides methods for microarray analysis that take basic
9270data types such as matrices and lists of vectors. These methods can be used
9271standalone, be utilized in other packages, or be wrapped up in higher-level
9272classes.")
9273 (license license:gpl2+)))
9274
7b465327
RW		 9275(define-public r-deseq
9276 (package
9277 (name "r-deseq")
8140f680 9278 (version "1.30.0")
7b465327
RW		 9279 (source
9280 (origin
9281 (method url-fetch)
9282 (uri (bioconductor-uri "DESeq" version))
9283 (sha256
9284 (base32
8140f680 9285 "0mn5w3cy16iwwk8zxs7za6aa6cnrca75z0g45zd5zh1py5d7nfv9"))))
7b465327
RW		 9286 (properties `((upstream-name . "DESeq")))
9287 (build-system r-build-system)
9288 (propagated-inputs
9289 `(("r-biobase" ,r-biobase)
9290 ("r-biocgenerics" ,r-biocgenerics)
9291 ("r-genefilter" ,r-genefilter)
9292 ("r-geneplotter" ,r-geneplotter)
9293 ("r-lattice" ,r-lattice)
9294 ("r-locfit" ,r-locfit)
9295 ("r-mass" ,r-mass)
9296 ("r-rcolorbrewer" ,r-rcolorbrewer)))
9297 (home-page "http://www-huber.embl.de/users/anders/DESeq")
9298 (synopsis "Differential gene expression analysis")
9299 (description
9300 "This package provides tools for estimating variance-mean dependence in
9301count data from high-throughput genetic sequencing assays and for testing for
9302differential expression based on a model using the negative binomial
9303distribution.")
9304 (license license:gpl3+)))
9305
296b3c63
RW		 9306(define-public r-edaseq
9307 (package
9308 (name "r-edaseq")
00961279 9309 (version "2.12.0")
296b3c63
RW		 9310 (source
9311 (origin
9312 (method url-fetch)
9313 (uri (bioconductor-uri "EDASeq" version))
9314 (sha256
9315 (base32
00961279 9316 "07zm89zcivyn2261aq9grqmly8ji482kr9h9dyfknfdfrpv7jpwv"))))
296b3c63
RW		 9317 (properties `((upstream-name . "EDASeq")))
9318 (build-system r-build-system)
9319 (propagated-inputs
9320 `(("r-annotationdbi" ,r-annotationdbi)
9321 ("r-aroma-light" ,r-aroma-light)
9322 ("r-biobase" ,r-biobase)
9323 ("r-biocgenerics" ,r-biocgenerics)
9324 ("r-biomart" ,r-biomart)
9325 ("r-biostrings" ,r-biostrings)
9326 ("r-deseq" ,r-deseq)
9327 ("r-genomicfeatures" ,r-genomicfeatures)
9328 ("r-genomicranges" ,r-genomicranges)
9329 ("r-iranges" ,r-iranges)
9330 ("r-rsamtools" ,r-rsamtools)
9331 ("r-shortread" ,r-shortread)))
9332 (home-page "https://github.com/drisso/EDASeq")
9333 (synopsis "Exploratory data analysis and normalization for RNA-Seq")
9334 (description
9335 "This package provides support for numerical and graphical summaries of
9336RNA-Seq genomic read data. Provided within-lane normalization procedures to
9337adjust for GC-content effect (or other gene-level effects) on read counts:
9338loess robust local regression, global-scaling, and full-quantile
9339normalization. Between-lane normalization procedures to adjust for
9340distributional differences between lanes (e.g., sequencing depth):
9341global-scaling and full-quantile normalization.")
9342 (license license:artistic2.0)))
9343
5f96f30f
RW		 9344(define-public r-interactivedisplaybase
9345 (package
9346 (name "r-interactivedisplaybase")
1567c8d9 9347 (version "1.16.0")
5f96f30f
RW		 9348 (source
9349 (origin
9350 (method url-fetch)
9351 (uri (bioconductor-uri "interactiveDisplayBase" version))
9352 (sha256
9353 (base32
1567c8d9 9354 "01yb945jqqimwjgriza6yy4dnp303cdirxrhl4hjyprfdlmnz5p5"))))
5f96f30f
RW		 9355 (properties
9356 `((upstream-name . "interactiveDisplayBase")))
9357 (build-system r-build-system)
9358 (propagated-inputs
9359 `(("r-biocgenerics" ,r-biocgenerics)
9360 ("r-shiny" ,r-shiny)))
5713bbf1 9361 (home-page "https://bioconductor.org/packages/interactiveDisplayBase")
5f96f30f
RW		 9362 (synopsis "Base package for web displays of Bioconductor objects")
9363 (description
9364 "This package contains the basic methods needed to generate interactive
9365Shiny-based display methods for Bioconductor objects.")
9366 (license license:artistic2.0)))
9367
4a007ffa
RW		 9368(define-public r-annotationhub
9369 (package
9370 (name "r-annotationhub")
14b3c719 9371 (version "2.10.0")
4a007ffa
RW		 9372 (source
9373 (origin
9374 (method url-fetch)
9375 (uri (bioconductor-uri "AnnotationHub" version))
9376 (sha256
9377 (base32
14b3c719 9378 "1arfka3czw8hkv6n2d85bgibq81s2rgkwhmpaxzhy6nw39vv7y8b"))))
4a007ffa
RW		 9379 (properties `((upstream-name . "AnnotationHub")))
9380 (build-system r-build-system)
9381 (propagated-inputs
9382 `(("r-annotationdbi" ,r-annotationdbi)
9383 ("r-biocgenerics" ,r-biocgenerics)
9384 ("r-biocinstaller" ,r-biocinstaller)
14b3c719 9385 ("r-curl" ,r-curl)
4a007ffa
RW		 9386 ("r-httr" ,r-httr)
9387 ("r-interactivedisplaybase" ,r-interactivedisplaybase)
9388 ("r-rsqlite" ,r-rsqlite)
9389 ("r-s4vectors" ,r-s4vectors)
9390 ("r-yaml" ,r-yaml)))
5713bbf1 9391 (home-page "https://bioconductor.org/packages/AnnotationHub")
4a007ffa
RW		 9392 (synopsis "Client to access AnnotationHub resources")
9393 (description
9394 "This package provides a client for the Bioconductor AnnotationHub web
9395resource. The AnnotationHub web resource provides a central location where
9396genomic files (e.g. VCF, bed, wig) and other resources from standard
9397locations (e.g. UCSC, Ensembl) can be discovered. The resource includes
9398metadata about each resource, e.g., a textual description, tags, and date of
9399modification. The client creates and manages a local cache of files retrieved
9400by the user, helping with quick and reproducible access.")
9401 (license license:artistic2.0)))
9402
d4a1e918
RW		 9403(define-public r-fastseg
9404 (package
9405 (name "r-fastseg")
87fa5afa 9406 (version "1.24.0")
d4a1e918
RW		 9407 (source
9408 (origin
9409 (method url-fetch)
9410 (uri (bioconductor-uri "fastseg" version))
9411 (sha256
9412 (base32
87fa5afa 9413 "0dd7nr3klwz9ailwshnbynhd62lwb8zbbpj6jf3igpb94yi6x2jp"))))
d4a1e918
RW		 9414 (build-system r-build-system)
9415 (propagated-inputs
9416 `(("r-biobase" ,r-biobase)
9417 ("r-biocgenerics" ,r-biocgenerics)
9418 ("r-genomicranges" ,r-genomicranges)
9419 ("r-iranges" ,r-iranges)
9420 ("r-s4vectors" ,r-s4vectors)))
9421 (home-page "http://www.bioinf.jku.at/software/fastseg/index.html")
9422 (synopsis "Fast segmentation algorithm for genetic sequencing data")
9423 (description
9424 "Fastseg implements a very fast and efficient segmentation algorithm.
9425It can segment data from DNA microarrays and data from next generation
9426sequencing for example to detect copy number segments. Further it can segment
9427data from RNA microarrays like tiling arrays to identify transcripts. Most
9428generally, it can segment data given as a matrix or as a vector. Various data
9429formats can be used as input to fastseg like expression set objects for
9430microarrays or GRanges for sequencing data.")
9431 (license license:lgpl2.0+)))
9432
3121814e
RW		 9433(define-public r-keggrest
9434 (package
9435 (name "r-keggrest")
9436 (version "1.18.0")
9437 (source
9438 (origin
9439 (method url-fetch)
9440 (uri (bioconductor-uri "KEGGREST" version))
9441 (sha256
9442 (base32
9443 "1i3i88lj57wvpgjf75a23msgfsjv8pr2b4j1faga276p4fsblkhj"))))
9444 (properties `((upstream-name . "KEGGREST")))
9445 (build-system r-build-system)
9446 (propagated-inputs
9447 `(("r-biostrings" ,r-biostrings)
9448 ("r-httr" ,r-httr)
9449 ("r-png" ,r-png)))
9450 (home-page "https://bioconductor.org/packages/KEGGREST")
9451 (synopsis "Client-side REST access to KEGG")
9452 (description
9453 "This package provides a package that provides a client interface to the
9454@dfn{Kyoto Encyclopedia of Genes and Genomes} (KEGG) REST server.")
9455 (license license:artistic2.0)))
9456
e160aa1f
RW		 9457(define-public r-gage
9458 (package
9459 (name "r-gage")
9460 (version "2.28.0")
9461 (source
9462 (origin
9463 (method url-fetch)
9464 (uri (bioconductor-uri "gage" version))
9465 (sha256
9466 (base32
9467 "1r14p88q3y736pkqm4pdimf1izy1xy3xgivmj3cr4dv65kjny1zk"))))
9468 (build-system r-build-system)
9469 (propagated-inputs
9470 `(("r-annotationdbi" ,r-annotationdbi)
9471 ("r-graph" ,r-graph)
9472 ("r-keggrest" ,r-keggrest)))
9473 (home-page "http://www.biomedcentral.com/1471-2105/10/161")
9474 (synopsis "Generally applicable gene-set enrichment for pathway analysis")
9475 (description
9476 "GAGE is a published method for gene set (enrichment or GSEA) or pathway
9477analysis. GAGE is generally applicable independent of microarray or RNA-Seq
9478data attributes including sample sizes, experimental designs, assay platforms,
9479and other types of heterogeneity. The gage package provides functions for
9480basic GAGE analysis, result processing and presentation. In addition, it
9481provides demo microarray data and commonly used gene set data based on KEGG
9482pathways and GO terms. These funtions and data are also useful for gene set
9483analysis using other methods.")
9484 (license license:gpl2+)))
9485
902fb15d
RW		 9486(define-public r-genomicfiles
9487 (package
9488 (name "r-genomicfiles")
9489 (version "1.14.0")
9490 (source
9491 (origin
9492 (method url-fetch)
9493 (uri (bioconductor-uri "GenomicFiles" version))
9494 (sha256
9495 (base32
9496 "0r0wmrs5jycf1kckhnc2sgjmp336srlcjdkpbb1ymm7kazdd0s9n"))))
9497 (properties `((upstream-name . "GenomicFiles")))
9498 (build-system r-build-system)
9499 (propagated-inputs
9500 `(("r-biocgenerics" ,r-biocgenerics)
9501 ("r-biocparallel" ,r-biocparallel)
9502 ("r-genomeinfodb" ,r-genomeinfodb)
9503 ("r-genomicalignments" ,r-genomicalignments)
9504 ("r-genomicranges" ,r-genomicranges)
9505 ("r-iranges" ,r-iranges)
9506 ("r-rsamtools" ,r-rsamtools)
9507 ("r-rtracklayer" ,r-rtracklayer)
9508 ("r-s4vectors" ,r-s4vectors)
9509 ("r-summarizedexperiment" ,r-summarizedexperiment)
9510 ("r-variantannotation" ,r-variantannotation)))
9511 (home-page "https://bioconductor.org/packages/GenomicFiles")
9512 (synopsis "Distributed computing by file or by range")
9513 (description
9514 "This package provides infrastructure for parallel computations
9515distributed by file or by range. User defined mapper and reducer functions
9516provide added flexibility for data combination and manipulation.")
9517 (license license:artistic2.0)))
9518
3af149f5
RW		 9519(define-public r-complexheatmap
9520 (package
9521 (name "r-complexheatmap")
9522 (version "1.17.1")
9523 (source
9524 (origin
9525 (method url-fetch)
9526 (uri (bioconductor-uri "ComplexHeatmap" version))
9527 (sha256
9528 (base32
9529 "1x6kp55iqqsd8bhdl3qch95nfiy2y46ldbbsx1sj1v8f0b0ywwcy"))))
9530 (properties
9531 `((upstream-name . "ComplexHeatmap")))
9532 (build-system r-build-system)
9533 (propagated-inputs
9534 `(("r-circlize" ,r-circlize)
9535 ("r-colorspace" ,r-colorspace)
9536 ("r-getoptlong" ,r-getoptlong)
9537 ("r-globaloptions" ,r-globaloptions)
9538 ("r-rcolorbrewer" ,r-rcolorbrewer)))
9539 (home-page
9540 "https://github.com/jokergoo/ComplexHeatmap")
9541 (synopsis "Making Complex Heatmaps")
9542 (description
9543 "Complex heatmaps are efficient to visualize associations between
9544different sources of data sets and reveal potential structures. This package
9545provides a highly flexible way to arrange multiple heatmaps and supports
9546self-defined annotation graphics.")
9547 (license license:gpl2+)))
9548
04e2a2e9
RW		 9549(define-public r-dirichletmultinomial
9550 (package
9551 (name "r-dirichletmultinomial")
9552 (version "1.20.0")
9553 (source
9554 (origin
9555 (method url-fetch)
9556 (uri (bioconductor-uri "DirichletMultinomial" version))
9557 (sha256
9558 (base32
9559 "1c4s6x0qm20556grcd1xys9kkpnlzpasaai474malwcg6qvgi4x1"))))
9560 (properties
9561 `((upstream-name . "DirichletMultinomial")))
9562 (build-system r-build-system)
9563 (inputs
9564 `(("gsl" ,gsl)))
9565 (propagated-inputs
9566 `(("r-biocgenerics" ,r-biocgenerics)
9567 ("r-iranges" ,r-iranges)
9568 ("r-s4vectors" ,r-s4vectors)))
9569 (home-page "https://bioconductor.org/packages/DirichletMultinomial")
9570 (synopsis "Dirichlet-Multinomial mixture models for microbiome data")
9571 (description
9572 "Dirichlet-multinomial mixture models can be used to describe variability
9573in microbial metagenomic data. This package is an interface to code
9574originally made available by Holmes, Harris, and Quince, 2012, PLoS ONE 7(2):
95751-15.")
9576 (license license:lgpl3)))
9577
763e9810
RW		 9578(define-public r-annotationfilter
9579 (package
9580 (name "r-annotationfilter")
9581 (version "1.2.0")
9582 (source (origin
9583 (method url-fetch)
9584 (uri (bioconductor-uri "AnnotationFilter" version))
9585 (sha256
9586 (base32
9587 "04zf864c1fvdlaay2r5cn30fc1n5i3czh31fs62qlrvs61wjiscs"))))
9588 (properties
9589 `((upstream-name . "AnnotationFilter")))
9590 (build-system r-build-system)
9591 (propagated-inputs
9592 `(("r-genomicranges" ,r-genomicranges)
9593 ("r-lazyeval" ,r-lazyeval)))
9594 (home-page "https://github.com/Bioconductor/AnnotationFilter")
9595 (synopsis "Facilities for filtering Bioconductor annotation resources")
9596 (description
9597 "This package provides classes and other infrastructure to implement
9598filters for manipulating Bioconductor annotation resources. The filters are
9599used by @code{ensembldb}, @code{Organism.dplyr}, and other packages.")
9600 (license license:artistic2.0)))
9601
aa30ab5a
RW		 9602(define-public r-ensembldb
9603 (package
9604 (name "r-ensembldb")
9605 (version "2.2.0")
9606 (source
9607 (origin
9608 (method url-fetch)
9609 (uri (bioconductor-uri "ensembldb" version))
9610 (sha256
9611 (base32
9612 "1w0lca3ws5j770bmls91cn93lznvv2pc8s42nybdzz3vdxjvb4m1"))))
9613 (build-system r-build-system)
9614 (propagated-inputs
9615 `(("r-annotationdbi" ,r-annotationdbi)
9616 ("r-annotationfilter" ,r-annotationfilter)
9617 ("r-annotationhub" ,r-annotationhub)
9618 ("r-biobase" ,r-biobase)
9619 ("r-biocgenerics" ,r-biocgenerics)
9620 ("r-biostrings" ,r-biostrings)
9621 ("r-curl" ,r-curl)
9622 ("r-dbi" ,r-dbi)
9623 ("r-genomeinfodb" ,r-genomeinfodb)
9624 ("r-genomicfeatures" ,r-genomicfeatures)
9625 ("r-genomicranges" ,r-genomicranges)
9626 ("r-iranges" ,r-iranges)
9627 ("r-protgenerics" ,r-protgenerics)
9628 ("r-rsamtools" ,r-rsamtools)
9629 ("r-rsqlite" ,r-rsqlite)
9630 ("r-rtracklayer" ,r-rtracklayer)
9631 ("r-s4vectors" ,r-s4vectors)))
9632 (home-page "https://github.com/jotsetung/ensembldb")
9633 (synopsis "Utilities to create and use Ensembl-based annotation databases")
9634 (description
9635 "The package provides functions to create and use transcript-centric
9636annotation databases/packages. The annotation for the databases are directly
9637fetched from Ensembl using their Perl API. The functionality and data is
9638similar to that of the TxDb packages from the @code{GenomicFeatures} package,
9639but, in addition to retrieve all gene/transcript models and annotations from
9640the database, the @code{ensembldb} package also provides a filter framework
9641allowing to retrieve annotations for specific entries like genes encoded on a
9642chromosome region or transcript models of lincRNA genes.")
9643 ;; No version specified
9644 (license license:lgpl3+)))
9645
2f6f61fd
RW		 9646(define-public r-organismdbi
9647 (package
9648 (name "r-organismdbi")
9649 (version "1.20.0")
9650 (source
9651 (origin
9652 (method url-fetch)
9653 (uri (bioconductor-uri "OrganismDbi" version))
9654 (sha256
9655 (base32
9656 "0yxvhwn0m53wfwp0zi81x96argdf7cf1lpymc2as51apvfcnjdl8"))))
9657 (properties `((upstream-name . "OrganismDbi")))
9658 (build-system r-build-system)
9659 (propagated-inputs
9660 `(("r-annotationdbi" ,r-annotationdbi)
9661 ("r-biobase" ,r-biobase)
9662 ("r-biocgenerics" ,r-biocgenerics)
9663 ("r-biocinstaller" ,r-biocinstaller)
9664 ("r-dbi" ,r-dbi)
9665 ("r-genomicfeatures" ,r-genomicfeatures)
9666 ("r-genomicranges" ,r-genomicranges)
9667 ("r-graph" ,r-graph)
9668 ("r-iranges" ,r-iranges)
9669 ("r-rbgl" ,r-rbgl)
9670 ("r-s4vectors" ,r-s4vectors)))
9671 (home-page "https://bioconductor.org/packages/OrganismDbi")
9672 (synopsis "Software to enable the smooth interfacing of database packages")
9673 (description "The package enables a simple unified interface to several
9674annotation packages each of which has its own schema by taking advantage of
9675the fact that each of these packages implements a select methods.")
9676 (license license:artistic2.0)))
9677
7af1f403
RW		 9678(define-public r-biovizbase
9679 (package
9680 (name "r-biovizbase")
9681 (version "1.26.0")
9682 (source
9683 (origin
9684 (method url-fetch)
9685 (uri (bioconductor-uri "biovizBase" version))
9686 (sha256
9687 (base32
9688 "14l4vhj0a4ssr9m9zdzz3qpd4qw1mhgq5bmxq7jhrq3j9kmd6i2f"))))
9689 (properties `((upstream-name . "biovizBase")))
9690 (build-system r-build-system)
9691 (propagated-inputs
9692 `(("r-annotationdbi" ,r-annotationdbi)
9693 ("r-annotationfilter" ,r-annotationfilter)
9694 ("r-biocgenerics" ,r-biocgenerics)
9695 ("r-biostrings" ,r-biostrings)
9696 ("r-dichromat" ,r-dichromat)
9697 ("r-ensembldb" ,r-ensembldb)
9698 ("r-genomeinfodb" ,r-genomeinfodb)
9699 ("r-genomicalignments" ,r-genomicalignments)
9700 ("r-genomicfeatures" ,r-genomicfeatures)
9701 ("r-genomicranges" ,r-genomicranges)
9702 ("r-hmisc" ,r-hmisc)
9703 ("r-iranges" ,r-iranges)
9704 ("r-rcolorbrewer" ,r-rcolorbrewer)
9705 ("r-rsamtools" ,r-rsamtools)
9706 ("r-s4vectors" ,r-s4vectors)
9707 ("r-scales" ,r-scales)
9708 ("r-summarizedexperiment" ,r-summarizedexperiment)
9709 ("r-variantannotation" ,r-variantannotation)))
9710 (home-page "https://bioconductor.org/packages/biovizBase")
9711 (synopsis "Basic graphic utilities for visualization of genomic data")
9712 (description
9713 "The biovizBase package is designed to provide a set of utilities, color
9714schemes and conventions for genomic data. It serves as the base for various
9715high-level packages for biological data visualization. This saves development
9716effort and encourages consistency.")
9717 (license license:artistic2.0)))
9718
60faf945
RW		 9719(define-public r-ggbio
9720 (package
9721 (name "r-ggbio")
9722 (version "1.26.0")
9723 (source
9724 (origin
9725 (method url-fetch)
9726 (uri (bioconductor-uri "ggbio" version))
9727 (sha256
9728 (base32
9729 "1bqxfqy0hff87ax92z4lfbjz01ndrz7x8pzm6dlkdmi52p30krm9"))))
9730 (build-system r-build-system)
9731 (propagated-inputs
9732 `(("r-annotationdbi" ,r-annotationdbi)
9733 ("r-annotationfilter" ,r-annotationfilter)
9734 ("r-biobase" ,r-biobase)
9735 ("r-biocgenerics" ,r-biocgenerics)
9736 ("r-biostrings" ,r-biostrings)
9737 ("r-biovizbase" ,r-biovizbase)
9738 ("r-bsgenome" ,r-bsgenome)
9739 ("r-ensembldb" ,r-ensembldb)
9740 ("r-genomeinfodb" ,r-genomeinfodb)
9741 ("r-genomicalignments" ,r-genomicalignments)
9742 ("r-genomicfeatures" ,r-genomicfeatures)
9743 ("r-genomicranges" ,r-genomicranges)
9744 ("r-ggally" ,r-ggally)
9745 ("r-ggplot2" ,r-ggplot2)
9746 ("r-gridextra" ,r-gridextra)
9747 ("r-gtable" ,r-gtable)
9748 ("r-hmisc" ,r-hmisc)
9749 ("r-iranges" ,r-iranges)
9750 ("r-organismdbi" ,r-organismdbi)
9751 ("r-reshape2" ,r-reshape2)
9752 ("r-rsamtools" ,r-rsamtools)
9753 ("r-rtracklayer" ,r-rtracklayer)
9754 ("r-s4vectors" ,r-s4vectors)
9755 ("r-scales" ,r-scales)
9756 ("r-summarizedexperiment" ,r-summarizedexperiment)
9757 ("r-variantannotation" ,r-variantannotation)))
9758 (home-page "http://www.tengfei.name/ggbio/")
9759 (synopsis "Visualization tools for genomic data")
9760 (description
9761 "The ggbio package extends and specializes the grammar of graphics for
9762biological data. The graphics are designed to answer common scientific
9763questions, in particular those often asked of high throughput genomics data.
9764All core Bioconductor data structures are supported, where appropriate. The
9765package supports detailed views of particular genomic regions, as well as
9766genome-wide overviews. Supported overviews include ideograms and grand linear
9767views. High-level plots include sequence fragment length, edge-linked
9768interval to data view, mismatch pileup, and several splicing summaries.")
9769 (license license:artistic2.0)))
9770
96a392a0
RW		 9771(define-public r-gprofiler
9772 (package
9773 (name "r-gprofiler")
9774 (version "0.6.1")
9775 (source
9776 (origin
9777 (method url-fetch)
9778 (uri (cran-uri "gProfileR" version))
9779 (sha256
9780 (base32
9781 "1qix15d0wa9nspdclcawml94mng4qmr2jciv7d24py315wfsvv8p"))))
9782 (properties `((upstream-name . "gProfileR")))
9783 (build-system r-build-system)
9784 (propagated-inputs
9785 `(("r-plyr" ,r-plyr)
9786 ("r-rcurl" ,r-rcurl)))
9787 (home-page "http://cran.r-project.org/web/packages/gProfileR/")
9788 (synopsis "Interface to the g:Profiler toolkit")
9789 (description
9790 "This package provides tools for functional enrichment analysis,
9791gene identifier conversion and mapping homologous genes across related
9792organisms via the @code{g:Profiler} toolkit.")
9793 (license license:gpl2+)))
9794
e2b92799
RW		 9795(define-public r-gqtlbase
9796 (package
9797 (name "r-gqtlbase")
9798 (version "1.10.0")
9799 (source
9800 (origin
9801 (method url-fetch)
9802 (uri (bioconductor-uri "gQTLBase" version))
9803 (sha256
9804 (base32
9805 "1756vfcj2dkkgcmfkkg7qdaig36dv9gfvpypn9rbrky56wm1p035"))))
9806 (properties `((upstream-name . "gQTLBase")))
9807 (build-system r-build-system)
9808 (propagated-inputs
9809 `(("r-batchjobs" ,r-batchjobs)
9810 ("r-bbmisc" ,r-bbmisc)
9811 ("r-biocgenerics" ,r-biocgenerics)
9812 ("r-bit" ,r-bit)
9813 ("r-doparallel" ,r-doparallel)
9814 ("r-ff" ,r-ff)
9815 ("r-ffbase" ,r-ffbase)
9816 ("r-foreach" ,r-foreach)
9817 ("r-genomicfiles" ,r-genomicfiles)
9818 ("r-genomicranges" ,r-genomicranges)
9819 ("r-rtracklayer" ,r-rtracklayer)
9820 ("r-s4vectors" ,r-s4vectors)
9821 ("r-summarizedexperiment" ,r-summarizedexperiment)))
9822 (home-page "https://bioconductor.org/packages/gQTLBase")
9823 (synopsis "Infrastructure for eQTL, mQTL and similar studies")
9824 (description
9825 "The purpose of this package is to simplify the storage and interrogation
9826of @dfn{quantitative trait loci} (QTL) archives, such as eQTL, mQTL, dsQTL,
9827and more.")
9828 (license license:artistic2.0)))
9829
dd5fc8b7
RW		 9830(define-public r-snpstats
9831 (package
9832 (name "r-snpstats")
9833 (version "1.28.0")
9834 (source
9835 (origin
9836 (method url-fetch)
9837 (uri (bioconductor-uri "snpStats" version))
9838 (sha256
9839 (base32
9840 "1x9qwynh2hwl24vq02naf4mchpch7xi2pkdrlgw896k28kx0lvir"))))
9841 (properties `((upstream-name . "snpStats")))
9842 (build-system r-build-system)
9843 (inputs `(("zlib" ,zlib)))
9844 (propagated-inputs
9845 `(("r-biocgenerics" ,r-biocgenerics)
9846 ("r-matrix" ,r-matrix)
9847 ("r-survival" ,r-survival)
9848 ("r-zlibbioc" ,r-zlibbioc)))
9849 (home-page "https://bioconductor.org/packages/snpStats")
9850 (synopsis "Methods for SNP association studies")
9851 (description
9852 "This package provides classes and statistical methods for large
9853@dfn{single-nucleotide polymorphism} (SNP) association studies. This extends
9854the earlier snpMatrix package, allowing for uncertainty in genotypes.")
9855 (license license:gpl3)))
9856
b52a866d
RW		 9857(define-public r-org-hs-eg-db
9858 (package
9859 (name "r-org-hs-eg-db")
9860 (version "3.4.2")
9861 (source (origin
9862 (method url-fetch)
9863 ;; We cannot use bioconductor-uri here because this tarball is
9864 ;; located under "data/annotation/" instead of "bioc/".
9865 (uri (string-append "http://www.bioconductor.org/packages/"
9866 "release/data/annotation/src/contrib/"
9867 "org.Hs.eg.db_"
9868 version ".tar.gz"))
9869 (sha256
9870 (base32
9871 "0izzmas99j64sc4x8pxi09xw0ra1941kjrsl8sjxl0cjw1d2b32z"))))
9872 (properties
9873 `((upstream-name . "org.Hs.eg.db")))
9874 (build-system r-build-system)
9875 (propagated-inputs
9876 `(("r-annotationdbi" ,r-annotationdbi)))
9877 (home-page "https://bioconductor.org/packages/org.Hs.eg.db/")
9878 (synopsis "Genome-wide annotation for Human")
9879 (description
9880 "This package contains genome-wide annotations for Human, primarily based
9881on mapping using Entrez Gene identifiers.")
9882 (license license:artistic2.0)))
9883
421c68e3
RW		 9884(define-public r-homo-sapiens
9885 (package
9886 (name "r-homo-sapiens")
9887 (version "1.3.1")
9888 (source (origin
9889 (method url-fetch)
9890 ;; We cannot use bioconductor-uri here because this tarball is
9891 ;; located under "data/annotation/" instead of "bioc/".
9892 (uri (string-append "http://www.bioconductor.org/packages/"
9893 "release/data/annotation/src/contrib/"
9894 "Homo.sapiens_"
9895 version ".tar.gz"))
9896 (sha256
9897 (base32
9898 "151vj7h5p1c8yd5swrchk46z469p135wk50hvkl0nhgndvy0jj01"))))
9899 (properties
9900 `((upstream-name . "Homo.sapiens")))
9901 (build-system r-build-system)
9902 (propagated-inputs
9903 `(("r-genomicfeatures" ,r-genomicfeatures)
9904 ("r-go-db" ,r-go-db)
9905 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
9906 ("r-txdb-hsapiens-ucsc-hg19-knowngene" ,r-txdb-hsapiens-ucsc-hg19-knowngene)
9907 ("r-organismdbi" ,r-organismdbi)
9908 ("r-annotationdbi" ,r-annotationdbi)))
9909 (home-page "https://bioconductor.org/packages/Homo.sapiens/")
9910 (synopsis "Annotation package for the Homo.sapiens object")
9911 (description
9912 "This package contains the Homo.sapiens object to access data from
9913several related annotation packages.")
9914 (license license:artistic2.0)))
9915
15f98b0d
RW		 9916(define-public r-erma
9917 (package
9918 (name "r-erma")
9919 (version "0.10.0")
9920 (source
9921 (origin
9922 (method url-fetch)
9923 (uri (bioconductor-uri "erma" version))
9924 (sha256
9925 (base32
9926 "0gcfs9g8vvdv5vmq9b21kd8sq5mizjj49nfzd4in9zvp4b9v7x1g"))))
9927 (build-system r-build-system)
9928 (propagated-inputs
9929 `(("r-annotationdbi" ,r-annotationdbi)
9930 ("r-biobase" ,r-biobase)
9931 ("r-biocgenerics" ,r-biocgenerics)
9932 ("r-foreach" ,r-foreach)
9933 ("r-genomicfiles" ,r-genomicfiles)
9934 ("r-genomicranges" ,r-genomicranges)
9935 ("r-ggplot2" ,r-ggplot2)
9936 ("r-homo-sapiens" ,r-homo-sapiens)
9937 ("r-rtracklayer" ,r-rtracklayer)
9938 ("r-s4vectors" ,r-s4vectors)
9939 ("r-shiny" ,r-shiny)
9940 ("r-summarizedexperiment" ,r-summarizedexperiment)))
9941 (home-page "https://bioconductor.org/packages/erma")
9942 (synopsis "Epigenomic road map adventures")
9943 (description
9944 "The epigenomics road map describes locations of epigenetic marks in DNA
9945from a variety of cell types. Of interest are locations of histone
9946modifications, sites of DNA methylation, and regions of accessible chromatin.
9947This package presents a selection of elements of the road map including
9948metadata and outputs of the ChromImpute procedure applied to ENCODE cell lines
9949by Ernst and Kellis.")
9950 (license license:artistic2.0)))
9951
66c39102
RW		 9952(define-public r-ldblock
9953 (package
9954 (name "r-ldblock")
9955 (version "1.8.0")
9956 (source
9957 (origin
9958 (method url-fetch)
9959 (uri (bioconductor-uri "ldblock" version))
9960 (sha256
9961 (base32
9962 "18nfsixh6d2wfrb9laqsgly5w1frzihhak683k0p8fdf51h4aqba"))))
9963 (build-system r-build-system)
9964 (propagated-inputs
9965 `(("r-erma" ,r-erma)
9966 ("r-genomeinfodb" ,r-genomeinfodb)
9967 ("r-genomicfiles" ,r-genomicfiles)
9968 ("r-go-db" ,r-go-db)
9969 ("r-homo-sapiens" ,r-homo-sapiens)
9970 ("r-matrix" ,r-matrix)
9971 ("r-rsamtools" ,r-rsamtools)
9972 ("r-snpstats" ,r-snpstats)
9973 ("r-variantannotation" ,r-variantannotation)))
9974 (home-page "https://bioconductor.org/packages/ldblock")
9975 (synopsis "Data structures for linkage disequilibrium measures in populations")
9976 (description
9977 "This package defines data structures for @dfn{linkage
9978disequilibrium} (LD) measures in populations. Its purpose is to simplify
9979handling of existing population-level data for the purpose of flexibly
9980defining LD blocks.")
9981 (license license:artistic2.0)))
9982
794ff347
RW		 9983(define-public r-gqtlstats
9984 (package
9985 (name "r-gqtlstats")
9986 (version "1.10.0")
9987 (source
9988 (origin
9989 (method url-fetch)
9990 (uri (bioconductor-uri "gQTLstats" version))
9991 (sha256
9992 (base32
9993 "1cbdqawxzgna8rrgj3siph5sw4d2pb57qc0gn6ibfkhyk45f8gdv"))))
9994 (properties `((upstream-name . "gQTLstats")))
9995 (build-system r-build-system)
9996 (propagated-inputs
9997 `(("r-annotationdbi" ,r-annotationdbi)
9998 ("r-batchjobs" ,r-batchjobs)
9999 ("r-bbmisc" ,r-bbmisc)
10000 ("r-beeswarm" ,r-beeswarm)
10001 ("r-biobase" ,r-biobase)
10002 ("r-biocgenerics" ,r-biocgenerics)
10003 ("r-doparallel" ,r-doparallel)
10004 ("r-dplyr" ,r-dplyr)
10005 ("r-erma" ,r-erma)
10006 ("r-ffbase" ,r-ffbase)
10007 ("r-foreach" ,r-foreach)
10008 ("r-genomeinfodb" ,r-genomeinfodb)
10009 ("r-genomicfeatures" ,r-genomicfeatures)
10010 ("r-genomicfiles" ,r-genomicfiles)
10011 ("r-genomicranges" ,r-genomicranges)
10012 ("r-ggbeeswarm" ,r-ggbeeswarm)
10013 ("r-ggplot2" ,r-ggplot2)
10014 ("r-gqtlbase" ,r-gqtlbase)
10015 ("r-hardyweinberg" ,r-hardyweinberg)
10016 ("r-iranges" ,r-iranges)
10017 ("r-ldblock" ,r-ldblock)
10018 ("r-limma" ,r-limma)
10019 ("r-mgcv" ,r-mgcv)
10020 ("r-plotly" ,r-plotly)
10021 ("r-reshape2" ,r-reshape2)
10022 ("r-s4vectors" ,r-s4vectors)
10023 ("r-shiny" ,r-shiny)
10024 ("r-snpstats" ,r-snpstats)
10025 ("r-summarizedexperiment" ,r-summarizedexperiment)
10026 ("r-variantannotation" ,r-variantannotation)))
10027 (home-page "https://bioconductor.org/packages/gQTLstats")
10028 (synopsis "Computationally efficient analysis for eQTL and allied studies")
10029 (description
10030 "This package provides tools for the computationally efficient analysis
10031of @dfn{quantitative trait loci} (QTL) data, including eQTL, mQTL, dsQTL, etc.
10032The software in this package aims to support refinements and functional
10033interpretation of members of a collection of association statistics on a
10034family of feature/genome hypotheses.")
10035 (license license:artistic2.0)))
10036
42720006
RW		 10037(define-public r-gviz
10038 (package
10039 (name "r-gviz")
10040 (version "1.22.0")
10041 (source
10042 (origin
10043 (method url-fetch)
10044 (uri (bioconductor-uri "Gviz" version))
10045 (sha256
10046 (base32
10047 "1lrw65a8426wpxw975wjcaiacpp6fqa00nif1yxigyankbfs23c8"))))
10048 (properties `((upstream-name . "Gviz")))
10049 (build-system r-build-system)
10050 (propagated-inputs
10051 `(("r-annotationdbi" ,r-annotationdbi)
10052 ("r-biobase" ,r-biobase)
10053 ("r-biocgenerics" ,r-biocgenerics)
10054 ("r-biomart" ,r-biomart)
10055 ("r-biostrings" ,r-biostrings)
10056 ("r-biovizbase" ,r-biovizbase)
10057 ("r-bsgenome" ,r-bsgenome)
10058 ("r-digest" ,r-digest)
10059 ("r-genomeinfodb" ,r-genomeinfodb)
10060 ("r-genomicalignments" ,r-genomicalignments)
10061 ("r-genomicfeatures" ,r-genomicfeatures)
10062 ("r-genomicranges" ,r-genomicranges)
10063 ("r-iranges" ,r-iranges)
10064 ("r-lattice" ,r-lattice)
10065 ("r-latticeextra" ,r-latticeextra)
10066 ("r-matrixstats" ,r-matrixstats)
10067 ("r-rcolorbrewer" ,r-rcolorbrewer)
10068 ("r-rsamtools" ,r-rsamtools)
10069 ("r-rtracklayer" ,r-rtracklayer)
10070 ("r-s4vectors" ,r-s4vectors)
10071 ("r-xvector" ,r-xvector)))
10072 (home-page "https://bioconductor.org/packages/Gviz")
10073 (synopsis "Plotting data and annotation information along genomic coordinates")
10074 (description
10075 "Genomic data analyses requires integrated visualization of known genomic
10076information and new experimental data. Gviz uses the biomaRt and the
10077rtracklayer packages to perform live annotation queries to Ensembl and UCSC
10078and translates this to e.g. gene/transcript structures in viewports of the
10079grid graphics package. This results in genomic information plotted together
10080with your data.")
10081 (license license:artistic2.0)))
10082
2e4ce60e
RW		 10083(define-public r-gwascat
10084 (package
10085 (name "r-gwascat")
10086 (version "2.10.0")
10087 (source
10088 (origin
10089 (method url-fetch)
10090 (uri (bioconductor-uri "gwascat" version))
10091 (sha256
10092 (base32
10093 "0n5x5i5v6a8wpn5mxmlpkl34b4kyypmymiwww6g61zch7xqrgywi"))))
10094 (build-system r-build-system)
10095 (propagated-inputs
10096 `(("r-annotationdbi" ,r-annotationdbi)
10097 ("r-annotationhub" ,r-annotationhub)
10098 ("r-biocgenerics" ,r-biocgenerics)
10099 ("r-biostrings" ,r-biostrings)
10100 ("r-genomeinfodb" ,r-genomeinfodb)
10101 ("r-genomicfeatures" ,r-genomicfeatures)
10102 ("r-genomicranges" ,r-genomicranges)
10103 ("r-ggbio" ,r-ggbio)
10104 ("r-ggplot2" ,r-ggplot2)
10105 ("r-gqtlstats" ,r-gqtlstats)
10106 ("r-graph" ,r-graph)
10107 ("r-gviz" ,r-gviz)
10108 ("r-homo-sapiens" ,r-homo-sapiens)
10109 ("r-iranges" ,r-iranges)
10110 ("r-rsamtools" ,r-rsamtools)
10111 ("r-rtracklayer" ,r-rtracklayer)
10112 ("r-s4vectors" ,r-s4vectors)
10113 ("r-snpstats" ,r-snpstats)
10114 ("r-summarizedexperiment" ,r-summarizedexperiment)
10115 ("r-variantannotation" ,r-variantannotation)))
10116 (home-page "https://bioconductor.org/packages/gwascat")
10117 (synopsis "Tools for data in the EMBL-EBI GWAS catalog")
10118 (description
10119 "This package provides tools for representing and modeling data in the
10120EMBL-EBI GWAS catalog.")
10121 (license license:artistic2.0)))
10122
fbf54406
RW		 10123(define-public r-sushi
10124 (package
10125 (name "r-sushi")
10126 (version "1.16.0")
10127 (source (origin
10128 (method url-fetch)
10129 (uri (bioconductor-uri "Sushi" version))
10130 (sha256
10131 (base32
10132 "0axaqm480z8d0b2ldgxwm0swava1p4irc62bpl17p2k8k78g687g"))))
10133 (properties `((upstream-name . "Sushi")))
10134 (build-system r-build-system)
10135 (propagated-inputs
10136 `(("r-biomart" ,r-biomart)
10137 ("r-zoo" ,r-zoo)))
10138 (home-page "https://bioconductor.org/packages/Sushi")
10139 (synopsis "Tools for visualizing genomics data")
10140 (description
10141 "This package provides flexible, quantitative, and integrative genomic
10142visualizations for publication-quality multi-panel figures.")
10143 (license license:gpl2+)))
10144
30eb4e37
RW		 10145(define-public r-fithic
10146 (package
10147 (name "r-fithic")
10148 (version "1.4.0")
10149 (source (origin
10150 (method url-fetch)
10151 (uri (bioconductor-uri "FitHiC" version))
10152 (sha256
10153 (base32
10154 "12ylhrppi051m7nqsgq95kzd9g9wmp34i0zzfi55cjqawlpx7c6n"))))
10155 (properties `((upstream-name . "FitHiC")))
10156 (build-system r-build-system)
10157 (propagated-inputs
10158 `(("r-data-table" ,r-data-table)
10159 ("r-fdrtool" ,r-fdrtool)
10160 ("r-rcpp" ,r-rcpp)))
10161 (home-page "https://bioconductor.org/packages/FitHiC")
10162 (synopsis "Confidence estimation for intra-chromosomal contact maps")
10163 (description
10164 "Fit-Hi-C is a tool for assigning statistical confidence estimates to
10165intra-chromosomal contact maps produced by genome-wide genome architecture
10166assays such as Hi-C.")
10167 (license license:gpl2+)))
10168
c49102ec
RW		 10169(define-public r-hitc
10170 (package
10171 (name "r-hitc")
10172 (version "1.22.0")
10173 (source (origin
10174 (method url-fetch)
10175 (uri (bioconductor-uri "HiTC" version))
10176 (sha256
10177 (base32
10178 "0288xa1jy6nzvz2ha07csmp6dirjw5r7p9vy69q2wsbyzr02ymkp"))))
10179 (properties `((upstream-name . "HiTC")))
10180 (build-system r-build-system)
10181 (propagated-inputs
10182 `(("r-biostrings" ,r-biostrings)
10183 ("r-genomeinfodb" ,r-genomeinfodb)
10184 ("r-genomicranges" ,r-genomicranges)
10185 ("r-iranges" ,r-iranges)
10186 ("r-matrix" ,r-matrix)
10187 ("r-rcolorbrewer" ,r-rcolorbrewer)
10188 ("r-rtracklayer" ,r-rtracklayer)))
10189 (home-page "https://bioconductor.org/packages/HiTC")
10190 (synopsis "High throughput chromosome conformation capture analysis")
10191 (description
10192 "The HiTC package was developed to explore high-throughput \"C\" data
10193such as 5C or Hi-C. Dedicated R classes as well as standard methods for
10194quality controls, normalization, visualization, and further analysis are also
10195provided.")
10196 (license license:artistic2.0)))
10197
212f6120
RW		 10198(define-public r-qvalue
10199 (package
10200 (name "r-qvalue")
98791c25 10201 (version "2.10.0")
212f6120
RW		 10202 (source
10203 (origin
10204 (method url-fetch)
10205 (uri (bioconductor-uri "qvalue" version))
10206 (sha256
10207 (base32
98791c25 10208 "1rd9rnf16kh8wc076kahd9hsb9rfwsbzmz3kjmp0pj6rbiq0051i"))))
212f6120
RW		 10209 (build-system r-build-system)
10210 (propagated-inputs
10211 `(("r-ggplot2" ,r-ggplot2)
10212 ("r-reshape2" ,r-reshape2)))
10213 (home-page "http://github.com/jdstorey/qvalue")
10214 (synopsis "Q-value estimation for false discovery rate control")
10215 (description
10216 "This package takes a list of p-values resulting from the simultaneous
10217testing of many hypotheses and estimates their q-values and local @dfn{false
10218discovery rate} (FDR) values. The q-value of a test measures the proportion
10219of false positives incurred when that particular test is called significant.
10220The local FDR measures the posterior probability the null hypothesis is true
10221given the test's p-value. Various plots are automatically generated, allowing
10222one to make sensible significance cut-offs. The software can be applied to
10223problems in genomics, brain imaging, astrophysics, and data mining.")
10224 ;; Any version of the LGPL.
10225 (license license:lgpl3+)))
10226
5ded35d8
RW		 10227(define htslib-for-sambamba
10228 (let ((commit "2f3c3ea7b301f9b45737a793c0b2dcf0240e5ee5"))
10229 (package
10230 (inherit htslib)
10231 (name "htslib-for-sambamba")
10232 (version (string-append "1.3.1-1." (string-take commit 9)))
10233 (source
10234 (origin
10235 (method git-fetch)
10236 (uri (git-reference
10237 (url "https://github.com/lomereiter/htslib.git")
10238 (commit commit)))
10239 (file-name (string-append "htslib-" version "-checkout"))
10240 (sha256
10241 (base32
10242 "0g38g8s3npr0gjm9fahlbhiskyfws9l5i0x1ml3rakzj7az5l9c9"))))
10243 (arguments
10244 (substitute-keyword-arguments (package-arguments htslib)
10245 ((#:phases phases)
10246 `(modify-phases ,phases
d10092b8 10247 (add-after 'unpack 'bootstrap
5ded35d8
RW		 10248 (lambda _
10249 (zero? (system* "autoreconf" "-vif"))))))))
10250 (native-inputs
10251 `(("autoconf" ,autoconf)
10252 ("automake" ,automake)
10253 ,@(package-native-inputs htslib))))))
10254
10255(define-public sambamba
10256 (package
10257 (name "sambamba")
10258 (version "0.6.5")
10259 (source
10260 (origin
10261 (method url-fetch)
10262 (uri (string-append "https://github.com/lomereiter/sambamba/"
10263 "archive/v" version ".tar.gz"))
10264 (file-name (string-append name "-" version ".tar.gz"))
10265 (sha256
10266 (base32
10267 "17076gijd65a3f07zns2gvbgahiz5lriwsa6dq353ss3jl85d8vy"))))
10268 (build-system gnu-build-system)
10269 (arguments
10270 `(#:tests? #f ; there is no test target
10271 #:make-flags
10272 '("D_COMPILER=ldc2"
10273 ;; Override "--compiler" flag only.
10274 "D_FLAGS=--compiler=ldc2 -IBioD -g -d"
10275 "sambamba-ldmd2-64")
10276 #:phases
10277 (modify-phases %standard-phases
10278 (delete 'configure)
10279 (add-after 'unpack 'place-biod
10280 (lambda* (#:key inputs #:allow-other-keys)
10281 (copy-recursively (assoc-ref inputs "biod") "BioD")
10282 #t))
10283 (add-after 'unpack 'unbundle-prerequisites
10284 (lambda _
10285 (substitute* "Makefile"
10286 ((" htslib-static lz4-static") ""))
10287 #t))
10288 (replace 'install
10289 (lambda* (#:key outputs #:allow-other-keys)
10290 (let* ((out (assoc-ref outputs "out"))
10291 (bin (string-append out "/bin")))
10292 (mkdir-p bin)
10293 (install-file "build/sambamba" bin)
10294 #t))))))
10295 (native-inputs
10296 `(("ldc" ,ldc)
10297 ("rdmd" ,rdmd)
10298 ("biod"
10299 ,(let ((commit "1248586b54af4bd4dfb28ebfebfc6bf012e7a587"))
10300 (origin
10301 (method git-fetch)
10302 (uri (git-reference
10303 (url "https://github.com/biod/BioD.git")
10304 (commit commit)))
10305 (file-name (string-append "biod-"
10306 (string-take commit 9)
10307 "-checkout"))
10308 (sha256
10309 (base32
10310 "1m8hi1n7x0ri4l6s9i0x6jg4z4v94xrfdzp7mbizdipfag0m17g3")))))))
10311 (inputs
10312 `(("lz4" ,lz4)
10313 ("htslib" ,htslib-for-sambamba)))
10314 (home-page "http://lomereiter.github.io/sambamba")
10315 (synopsis "Tools for working with SAM/BAM data")
10316 (description "Sambamba is a high performance modern robust and
10317fast tool (and library), written in the D programming language, for
10318working with SAM and BAM files. Current parallelised functionality is
10319an important subset of samtools functionality, including view, index,
10320sort, markdup, and depth.")
10321 (license license:gpl2+)))
b8fffd3a
RW		 10322
10323(define-public ritornello
10324 (package
10325 (name "ritornello")
10326 (version "1.0.0")
10327 (source (origin
10328 (method url-fetch)
10329 (uri (string-append "https://github.com/KlugerLab/"
10330 "Ritornello/archive/v"
10331 version ".tar.gz"))
10332 (file-name (string-append name "-" version ".tar.gz"))
10333 (sha256
10334 (base32
10335 "02nik86gq9ljjriv6pamwlmqnfky3ads1fpklx6mc3hx6k40pg38"))))
10336 (build-system gnu-build-system)
10337 (arguments
10338 `(#:tests? #f ; there are no tests
10339 #:phases
10340 (modify-phases %standard-phases
10341 (add-after 'unpack 'patch-samtools-references
10342 (lambda* (#:key inputs #:allow-other-keys)
10343 (substitute* '("src/SamStream.h"
10344 "src/BufferedGenomeReader.h")
10345 (("<sam.h>") "<samtools/sam.h>"))
10346 #t))
10347 (delete 'configure)
10348 (replace 'install
10349 (lambda* (#:key inputs outputs #:allow-other-keys)
10350 (let* ((out (assoc-ref outputs "out"))
10351 (bin (string-append out "/bin/")))
10352 (mkdir-p bin)
10353 (install-file "bin/Ritornello" bin)
10354 #t))))))
10355 (inputs
10356 `(("samtools" ,samtools-0.1)
10357 ("fftw" ,fftw)
10358 ("boost" ,boost)
10359 ("zlib" ,zlib)))
10360 (home-page "https://github.com/KlugerLab/Ritornello")
10361 (synopsis "Control-free peak caller for ChIP-seq data")
10362 (description "Ritornello is a ChIP-seq peak calling algorithm based on
10363signal processing that can accurately call binding events without the need to
10364do a pair total DNA input or IgG control sample. It has been tested for use
10365with narrow binding events such as transcription factor ChIP-seq.")
10366 (license license:gpl3+)))
44f6c889
RW		 10367
10368(define-public trim-galore
10369 (package
10370 (name "trim-galore")
10371 (version "0.4.2")
10372 (source
10373 (origin
10374 (method url-fetch)
10375 (uri (string-append "http://www.bioinformatics.babraham.ac.uk/"
10376 "projects/trim_galore/trim_galore_v"
10377 version ".zip"))
10378 (sha256
10379 (base32
10380 "0b9qdxi4521gsrjvbhgky8g7kry9b5nx3byzaxkgxz7p4k8bn1mn"))))
10381 (build-system gnu-build-system)
10382 (arguments
10383 `(#:tests? #f ; no tests
10384 #:phases
10385 (modify-phases %standard-phases
10386 ;; The archive contains plain files.
10387 (replace 'unpack
10388 (lambda* (#:key source #:allow-other-keys)
10389 (zero? (system* "unzip" source))))
10390 (delete 'configure)
10391 (delete 'build)
10392 (add-after 'unpack 'hardcode-tool-references
10393 (lambda* (#:key inputs #:allow-other-keys)
10394 (substitute* "trim_galore"
10395 (("\\$path_to_cutadapt = 'cutadapt'")
10396 (string-append "$path_to_cutadapt = '"
10397 (assoc-ref inputs "cutadapt")
10398 "/bin/cutadapt'"))
10399 (("\\| gzip")
10400 (string-append "| "
10401 (assoc-ref inputs "gzip")
10402 "/bin/gzip"))
10403 (("\"gunzip")
10404 (string-append "\""
10405 (assoc-ref inputs "gzip")
10406 "/bin/gunzip")))
10407 #t))
10408 (replace 'install
10409 (lambda* (#:key outputs #:allow-other-keys)
10410 (let ((bin (string-append (assoc-ref outputs "out")
10411 "/bin")))
10412 (mkdir-p bin)
10413 (install-file "trim_galore" bin)
10414 #t))))))
10415 (inputs
10416 `(("gzip" ,gzip)
10417 ("perl" ,perl)
10418 ("cutadapt" ,cutadapt)))
10419 (native-inputs
10420 `(("unzip" ,unzip)))
10421 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/")
10422 (synopsis "Wrapper around Cutadapt and FastQC")
10423 (description "Trim Galore! is a wrapper script to automate quality and
10424adapter trimming as well as quality control, with some added functionality to
10425remove biased methylation positions for RRBS sequence files.")
10426 (license license:gpl3+)))
3420c905
RW		 10427
10428(define-public gess
10429 (package
10430 (name "gess")
10431 (version "1.0")
10432 (source (origin
10433 (method url-fetch)
10434 (uri (string-append "http://compbio.uthscsa.edu/"
10435 "GESS_Web/files/"
10436 "gess-" version ".src.tar.gz"))
10437 (sha256
10438 (base32
10439 "0hyk403kxscclzfs24pvdgiv0wm03kjcziqdrp5w46cb049gz0d7"))))
10440 (build-system gnu-build-system)
10441 (arguments
10442 `(#:tests? #f ; no tests
10443 #:phases
10444 (modify-phases %standard-phases
10445 (delete 'configure)
10446 (delete 'build)
10447 (replace 'install
10448 (lambda* (#:key inputs outputs #:allow-other-keys)
10449 (let* ((python (assoc-ref inputs "python"))
10450 (out (assoc-ref outputs "out"))
10451 (bin (string-append out "/bin/"))
10452 (target (string-append
10453 out "/lib/python2.7/site-packages/gess/")))
10454 (mkdir-p target)
10455 (copy-recursively "." target)
10456 ;; Make GESS.py executable
10457 (chmod (string-append target "GESS.py") #o555)
10458 ;; Add Python shebang to the top and make Matplotlib
10459 ;; usable.
10460 (substitute* (string-append target "GESS.py")
10461 (("\"\"\"Description:" line)
10462 (string-append "#!" (which "python") "
10463import matplotlib
10464matplotlib.use('Agg')
10465" line)))
10466 ;; Make sure GESS has all modules in its path
10467 (wrap-program (string-append target "GESS.py")
10468 `("PYTHONPATH" ":" prefix (,target ,(getenv "PYTHONPATH"))))
10469 (mkdir-p bin)
10470 (symlink (string-append target "GESS.py")
10471 (string-append bin "GESS.py"))
10472 #t))))))
10473 (inputs
10474 `(("python" ,python-2)
10475 ("python2-pysam" ,python2-pysam)
10476 ("python2-scipy" ,python2-scipy)
10477 ("python2-numpy" ,python2-numpy)
10478 ("python2-networkx" ,python2-networkx)
10479 ("python2-biopython" ,python2-biopython)))
10480 (home-page "http://compbio.uthscsa.edu/GESS_Web/")
10481 (synopsis "Detect exon-skipping events from raw RNA-seq data")
10482 (description
10483 "GESS is an implementation of a novel computational method to detect de
10484novo exon-skipping events directly from raw RNA-seq data without the prior
10485knowledge of gene annotation information. GESS stands for the graph-based
10486exon-skipping scanner detection scheme.")
10487 (license license:bsd-3)))
282b0151
RW		 10488
10489(define-public phylip
10490 (package
10491 (name "phylip")
10492 (version "3.696")
10493 (source
10494 (origin
10495 (method url-fetch)
10496 (uri (string-append "http://evolution.gs.washington.edu/phylip/"
10497 "download/phylip-" version ".tar.gz"))
10498 (sha256
10499 (base32
10500 "01jar1rayhr2gba2pgbw49m56rc5z4p5wn3ds0m188hrlln4a2nd"))))
10501 (build-system gnu-build-system)
10502 (arguments
10503 `(#:tests? #f ; no check target
10504 #:make-flags (list "-f" "Makefile.unx" "install")
10505 #:parallel-build? #f ; not supported
10506 #:phases
10507 (modify-phases %standard-phases
10508 (add-after 'unpack 'enter-dir
10509 (lambda _ (chdir "src") #t))
10510 (delete 'configure)
10511 (replace 'install
10512 (lambda* (#:key inputs outputs #:allow-other-keys)
10513 (let ((target (string-append (assoc-ref outputs "out")
10514 "/bin")))
10515 (mkdir-p target)
10516 (for-each (lambda (file)
10517 (install-file file target))
10518 (find-files "../exe" ".*")))
10519 #t)))))
10520 (home-page "http://evolution.genetics.washington.edu/phylip/")
10521 (synopsis "Tools for inferring phylogenies")
10522 (description "PHYLIP (the PHYLogeny Inference Package) is a package of
10523programs for inferring phylogenies (evolutionary trees).")
10524 (license license:bsd-2)))
aa163424
RW		 10525
10526(define-public imp
10527 (package
10528 (name "imp")
10529 (version "2.6.2")
10530 (source
10531 (origin
10532 (method url-fetch)
10533 (uri (string-append "https://integrativemodeling.org/"
10534 version "/download/imp-" version ".tar.gz"))
10535 (sha256
10536 (base32
10537 "0lxqx7vh79d771svr611dkilp6sn30qrbw8zvscbrm37v38d2j6h"))))
10538 (build-system cmake-build-system)
10539 (arguments
10540 `(;; FIXME: Some tests fail because they produce warnings, others fail
10541 ;; because the PYTHONPATH does not include the modeller's directory.
10542 #:tests? #f
10543 ;; Do not place libraries in an architecture-specific directory.
10544 #:configure-flags
10545 (list "-DCMAKE_INSTALL_LIBDIR=lib")))
10546 (inputs
10547 `(("boost" ,boost)
10548 ("gsl" ,gsl)
10549 ("swig" ,swig)
10550 ("hdf5" ,hdf5)
10551 ("fftw" ,fftw)
10552 ("python" ,python-2)))
10553 (propagated-inputs
10554 `(("python2-numpy" ,python2-numpy)
10555 ("python2-scipy" ,python2-scipy)
10556 ("python2-pandas" ,python2-pandas)
10557 ("python2-scikit-learn" ,python2-scikit-learn)
10558 ("python2-networkx" ,python2-networkx)))
10559 (home-page "https://integrativemodeling.org")
10560 (synopsis "Integrative modeling platform")
10561 (description "IMP's broad goal is to contribute to a comprehensive
10562structural characterization of biomolecules ranging in size and complexity
10563from small peptides to large macromolecular assemblies, by integrating data
10564from diverse biochemical and biophysical experiments. IMP provides a C++ and
10565Python toolbox for solving complex modeling problems, and a number of
10566applications for tackling some common problems in a user-friendly way.")
10567 ;; IMP is largely available under the GNU Lesser GPL; see the file
10568 ;; COPYING.LGPL for the full text of this license. Some IMP modules are
10569 ;; available under the GNU GPL (see the file COPYING.GPL).
10570 (license (list license:lgpl2.1+
10571 license:gpl3+))))
8befd094
RW		 10572
10573(define-public tadbit
10574 (package
10575 (name "tadbit")
10576 (version "0.2")
10577 (source (origin
10578 (method url-fetch)
10579 (uri (string-append "https://github.com/3DGenomes/TADbit/"
10580 "archive/v" version ".tar.gz"))
10581 (file-name (string-append name "-" version ".tar.gz"))
10582 (sha256
10583 (base32
10584 "1cnfqrl4685zar4nnw94j94nhvl2h29jm448nadqi1h05z6fdk4f"))))
10585 (build-system python-build-system)
10586 (arguments
10587 `(;; Tests are included and must be run after installation, but
10588 ;; they are incomplete and thus cannot be run.
10589 #:tests? #f
10590 #:python ,python-2
10591 #:phases
10592 (modify-phases %standard-phases
10593 (add-after 'unpack 'fix-problems-with-setup.py
10594 (lambda* (#:key outputs #:allow-other-keys)
10595 ;; setup.py opens these files for writing
10596 (chmod "_pytadbit/_version.py" #o664)
10597 (chmod "README.rst" #o664)
10598
10599 ;; Don't attempt to install the bash completions to
10600 ;; the home directory.
10601 (rename-file "extras/.bash_completion"
10602 "extras/tadbit")
10603 (substitute* "setup.py"
10604 (("\\(path.expanduser\\('~'\\)")
10605 (string-append "(\""
10606 (assoc-ref outputs "out")
10607 "/etc/bash_completion.d\""))
10608 (("extras/\\.bash_completion")
10609 "extras/tadbit"))
10610 #t)))))
10611 (inputs
10612 ;; TODO: add Chimera for visualization
10613 `(("imp" ,imp)
10614 ("mcl" ,mcl)
10615 ("python2-scipy" ,python2-scipy)
10616 ("python2-numpy" ,python2-numpy)
10617 ("python2-matplotlib" ,python2-matplotlib)
10618 ("python2-pysam" ,python2-pysam)))
10619 (home-page "http://3dgenomes.github.io/TADbit/")
10620 (synopsis "Analyze, model, and explore 3C-based data")
10621 (description
10622 "TADbit is a complete Python library to deal with all steps to analyze,
10623model, and explore 3C-based data. With TADbit the user can map FASTQ files to
10624obtain raw interaction binned matrices (Hi-C like matrices), normalize and
a147fadd 10625correct interaction matrices, identify and compare the so-called
8befd094
RW		 10626@dfn{Topologically Associating Domains} (TADs), build 3D models from the
10627interaction matrices, and finally, extract structural properties from the
10628models. TADbit is complemented by TADkit for visualizing 3D models.")
10629 (license license:gpl3+)))
94820951
RW		 10630
10631(define-public kentutils
10632 (package
10633 (name "kentutils")
10634 ;; 302.1.0 is out, but the only difference is the inclusion of
10635 ;; pre-built binaries.
10636 (version "302.0.0")
10637 (source
10638 (origin
10639 (method url-fetch)
10640 (uri (string-append "https://github.com/ENCODE-DCC/kentUtils/"
10641 "archive/v" version ".tar.gz"))
10642 (file-name (string-append name "-" version ".tar.gz"))
10643 (sha256
10644 (base32
10645 "134aja3k1cj32kbk1nnw0q9gxjb2krr15q6sga8qldzvc0585rmm"))
10646 (modules '((guix build utils)
10647 (srfi srfi-26)
10648 (ice-9 ftw)))
10649 (snippet
10650 '(begin
10651 ;; Only the contents of the specified directories are free
10652 ;; for all uses, so we remove the rest. "hg/autoSql" and
10653 ;; "hg/autoXml" are nominally free, but they depend on a
10654 ;; library that is built from the sources in "hg/lib",
10655 ;; which is nonfree.
10656 (let ((free (list "." ".."
10657 "utils" "lib" "inc" "tagStorm"
10658 "parasol" "htslib"))
10659 (directory? (lambda (file)
10660 (eq? 'directory (stat:type (stat file))))))
10661 (for-each (lambda (file)
10662 (and (directory? file)
10663 (delete-file-recursively file)))
10664 (map (cut string-append "src/" <>)
10665 (scandir "src"
10666 (lambda (file)
10667 (not (member file free)))))))
10668 ;; Only make the utils target, not the userApps target,
10669 ;; because that requires libraries we won't build.
10670 (substitute* "Makefile"
10671 ((" userApps") " utils"))
10672 ;; Only build libraries that are free.
10673 (substitute* "src/makefile"
10674 (("DIRS =.*") "DIRS =\n")
10675 (("cd jkOwnLib.*") "")
10676 ((" hgLib") "")
10677 (("cd hg.*") ""))
10678 (substitute* "src/utils/makefile"
10679 ;; These tools depend on "jkhgap.a", which is part of the
10680 ;; nonfree "src/hg/lib" directory.
10681 (("raSqlQuery") "")
10682 (("pslLiftSubrangeBlat") "")
10683
10684 ;; Do not build UCSC tools, which may require nonfree
10685 ;; components.
10686 (("ALL_APPS =.*") "ALL_APPS = $(UTILS_APPLIST)\n"))
10687 #t))))
10688 (build-system gnu-build-system)
10689 (arguments
10690 `( ;; There is no global test target and the test target for
10691 ;; individual tools depends on input files that are not
10692 ;; included.
10693 #:tests? #f
10694 #:phases
10695 (modify-phases %standard-phases
10696 (add-after 'unpack 'fix-paths
10697 (lambda _
10698 (substitute* "Makefile"
10699 (("/bin/echo") (which "echo")))
10700 #t))
10701 (add-after 'unpack 'prepare-samtabix
10702 (lambda* (#:key inputs #:allow-other-keys)
10703 (copy-recursively (assoc-ref inputs "samtabix")
10704 "samtabix")
10705 #t))
10706 (delete 'configure)
10707 (replace 'install
10708 (lambda* (#:key outputs #:allow-other-keys)
10709 (let ((bin (string-append (assoc-ref outputs "out")
10710 "/bin")))
10711 (copy-recursively "bin" bin))
10712 #t)))))
10713 (native-inputs
10714 `(("samtabix"
10715 ,(origin
10716 (method git-fetch)
10717 (uri (git-reference
10718 (url "http://genome-source.cse.ucsc.edu/samtabix.git")
10719 (commit "10fd107909c1ac4d679299908be4262a012965ba")))
10720 (sha256
10721 (base32
10722 "0c1nj64l42v395sa84n7az43xiap4i6f9n9dfz4058aqiwkhkmma"))))))
10723 (inputs
10724 `(("zlib" ,zlib)
10725 ("tcsh" ,tcsh)
10726 ("perl" ,perl)
10727 ("libpng" ,libpng)
10728 ("mysql" ,mysql)
10729 ("openssl" ,openssl)))
10730 (home-page "http://genome.cse.ucsc.edu/index.html")
10731 (synopsis "Assorted bioinformatics utilities")
10732 (description "This package provides the kentUtils, a selection of
10733bioinformatics utilities used in combination with the UCSC genome
10734browser.")
10735 ;; Only a subset of the sources are released under a non-copyleft
10736 ;; free software license. All other sources are removed in a
10737 ;; snippet. See this bug report for an explanation of how the
10738 ;; license statements apply:
10739 ;; https://github.com/ENCODE-DCC/kentUtils/issues/12
10740 (license (license:non-copyleft
10741 "http://genome.ucsc.edu/license/"
10742 "The contents of this package are free for all uses."))))
7ceb0a83
RW		 10743
10744(define-public f-seq
10745 (let ((commit "6ccded34cff38cf432deed8503648b4a66953f9b")
10746 (revision "1"))
10747 (package
10748 (name "f-seq")
c6643f2d 10749 (version (string-append "1.1-" revision "." (string-take commit 7)))
7ceb0a83
RW		 10750 (source (origin
10751 (method git-fetch)
10752 (uri (git-reference
10753 (url "https://github.com/aboyle/F-seq.git")
10754 (commit commit)))
10755 (file-name (string-append name "-" version))
10756 (sha256
10757 (base32
10758 "1nk33k0yajg2id4g59bc4szr58r2q6pdq42vgcw054m8ip9wv26h"))
10759 (modules '((guix build utils)))
10760 ;; Remove bundled Java library archives.
10761 (snippet
10762 '(begin
10763 (for-each delete-file (find-files "lib" ".*"))
10764 #t))))
10765 (build-system ant-build-system)
10766 (arguments
10767 `(#:tests? #f ; no tests included
10768 #:phases
10769 (modify-phases %standard-phases
10770 (replace 'install
c0770fee 10771 (lambda* (#:key inputs outputs #:allow-other-keys)
7ceb0a83
RW		 10772 (let* ((target (assoc-ref outputs "out"))
10773 (doc (string-append target "/share/doc/f-seq/")))
10774 (mkdir-p target)
10775 (mkdir-p doc)
10776 (substitute* "bin/linux/fseq"
c0770fee
RW		 10777 (("java") (which "java"))
10778 (("\\$REALDIR/../lib/commons-cli-1.1.jar")
10779 (string-append (assoc-ref inputs "java-commons-cli")
10780 "/share/java/commons-cli.jar"))
10781 (("REALDIR=.*")
10782 (string-append "REALDIR=" target "/bin\n")))
7ceb0a83
RW		 10783 (install-file "README.txt" doc)
10784 (install-file "bin/linux/fseq" (string-append target "/bin"))
10785 (install-file "build~/fseq.jar" (string-append target "/lib"))
10786 (copy-recursively "lib" (string-append target "/lib"))
10787 #t))))))
10788 (inputs
10789 `(("perl" ,perl)
10790 ("java-commons-cli" ,java-commons-cli)))
10791 (home-page "http://fureylab.web.unc.edu/software/fseq/")
10792 (synopsis "Feature density estimator for high-throughput sequence tags")
10793 (description
10794 "F-Seq is a software package that generates a continuous tag sequence
10795density estimation allowing identification of biologically meaningful sites
10796such as transcription factor binding sites (ChIP-seq) or regions of open
10797chromatin (DNase-seq). Output can be displayed directly in the UCSC Genome
10798Browser.")
10799 (license license:gpl3+))))
96348048
RW		 10800
10801(define-public bismark
10802 (package
10803 (name "bismark")
10804 (version "0.16.3")
10805 (source
10806 (origin
10807 (method url-fetch)
10808 (uri (string-append "https://github.com/FelixKrueger/Bismark/"
10809 "archive/" version ".tar.gz"))
10810 (file-name (string-append name "-" version ".tar.gz"))
10811 (sha256
10812 (base32
10813 "1204i0pa02ll2jn5pnxypkclnskvv7a2nwh5nxhagmhxk9wfv9sq"))))
10814 (build-system perl-build-system)
10815 (arguments
10816 `(#:tests? #f ; there are no tests
10817 #:phases
10818 (modify-phases %standard-phases
10819 (delete 'configure)
10820 (delete 'build)
10821 (replace 'install
10822 (lambda* (#:key outputs #:allow-other-keys)
10823 (let ((bin (string-append (assoc-ref outputs "out")
10824 "/bin"))
10825 (docdir (string-append (assoc-ref outputs "out")
10826 "/share/doc/bismark"))
10827 (docs '("Bismark_User_Guide.pdf"
10828 "RELEASE_NOTES.txt"))
10829 (scripts '("bismark"
10830 "bismark_genome_preparation"
10831 "bismark_methylation_extractor"
10832 "bismark2bedGraph"
10833 "bismark2report"
10834 "coverage2cytosine"
10835 "deduplicate_bismark"
10836 "bismark_sitrep.tpl"
10837 "bam2nuc"
10838 "bismark2summary")))
10839 (mkdir-p docdir)
10840 (mkdir-p bin)
10841 (for-each (lambda (file) (install-file file bin))
10842 scripts)
10843 (for-each (lambda (file) (install-file file docdir))
10844 docs)
10845 #t))))))
10846 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/bismark/")
10847 (synopsis "Map bisulfite treated sequence reads and analyze methylation")
10848 (description "Bismark is a program to map bisulfite treated sequencing
10849reads to a genome of interest and perform methylation calls in a single step.
10850The output can be easily imported into a genome viewer, such as SeqMonk, and
10851enables a researcher to analyse the methylation levels of their samples
10852straight away. Its main features are:
10853
10854@itemize
10855@item Bisulfite mapping and methylation calling in one single step
10856@item Supports single-end and paired-end read alignments
10857@item Supports ungapped and gapped alignments
10858@item Alignment seed length, number of mismatches etc are adjustable
10859@item Output discriminates between cytosine methylation in CpG, CHG
10860 and CHH context
10861@end itemize\n")
10862 (license license:gpl3+)))
dc8ddbfb
RW		 10863
10864(define-public paml
10865 (package
10866 (name "paml")
10867 (version "4.9e")
10868 (source (origin
10869 (method url-fetch)
10870 (uri (string-append "http://abacus.gene.ucl.ac.uk/software/"
10871 "paml" version ".tgz"))
10872 (sha256
10873 (base32
10874 "13zf6h9fiqghwhch2h06x1zdr6s42plsnqahflp5g7myr3han3s6"))
10875 (modules '((guix build utils)))
10876 ;; Remove Windows binaries
10877 (snippet
10878 '(begin
10879 (for-each delete-file (find-files "." "\\.exe$"))
10880 #t))))
10881 (build-system gnu-build-system)
10882 (arguments
10883 `(#:tests? #f ; there are no tests
10884 #:make-flags '("CC=gcc")
10885 #:phases
10886 (modify-phases %standard-phases
10887 (replace 'configure
10888 (lambda _
10889 (substitute* "src/BFdriver.c"
10890 (("/bin/bash") (which "bash")))
10891 (chdir "src")
10892 #t))
10893 (replace 'install
10894 (lambda* (#:key outputs #:allow-other-keys)
10895 (let ((tools '("baseml" "basemlg" "codeml"
10896 "pamp" "evolver" "yn00" "chi2"))
10897 (bin (string-append (assoc-ref outputs "out") "/bin"))
10898 (docdir (string-append (assoc-ref outputs "out")
10899 "/share/doc/paml")))
10900 (mkdir-p bin)
10901 (for-each (lambda (file) (install-file file bin)) tools)
10902 (copy-recursively "../doc" docdir)
10903 #t))))))
10904 (home-page "http://abacus.gene.ucl.ac.uk/software/paml.html")
10905 (synopsis "Phylogentic analysis by maximum likelihood")
10906 (description "PAML (for Phylogentic Analysis by Maximum Likelihood)
10907contains a few programs for model fitting and phylogenetic tree reconstruction
10908using nucleotide or amino-acid sequence data.")
10909 ;; GPLv3 only
10910 (license license:gpl3)))
jackhill's copy of GNU Guix: https://guix.gnu.org/