HCoop / jackhill / guix / guix.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
gnu: Add java-jboss-jms-api-spec.
[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)
ac257f12 50 #:use-module (gnu packages check)
4e10a221 51 #:use-module (gnu packages compression)
82c370de 52 #:use-module (gnu packages cpio)
7cb61550 53 #:use-module (gnu packages cran)
1baee943 54 #:use-module (gnu packages curl)
99828fa7 55 #:use-module (gnu packages documentation)
94820951 56 #:use-module (gnu packages databases)
d29150b5 57 #:use-module (gnu packages datastructures)
75dd2424 58 #:use-module (gnu packages file)
99268755 59 #:use-module (gnu packages flex)
02f35bb5 60 #:use-module (gnu packages gawk)
2409f37f 61 #:use-module (gnu packages gcc)
66e40e00 62 #:use-module (gnu packages gd)
97b9da68 63 #:use-module (gnu packages gtk)
b16728b0 64 #:use-module (gnu packages glib)
18f5d2a7 65 #:use-module (gnu packages graph)
db7a3444 66 #:use-module (gnu packages groff)
50937297 67 #:use-module (gnu packages guile)
89984be4 68 #:use-module (gnu packages haskell)
66e40e00 69 #:use-module (gnu packages image)
97b9da68 70 #:use-module (gnu packages imagemagick)
15a3c3d4 71 #:use-module (gnu packages java)
5ded35d8 72 #:use-module (gnu packages ldc)
51c64999 73 #:use-module (gnu packages linux)
ec946638 74 #:use-module (gnu packages logging)
36742f43 75 #:use-module (gnu packages machine-learning)
db7a3444 76 #:use-module (gnu packages man)
c833ab55 77 #:use-module (gnu packages maths)
6c2b26e2 78 #:use-module (gnu packages mpi)
4e10a221 79 #:use-module (gnu packages ncurses)
c033f5d6 80 #:use-module (gnu packages ocaml)
81f3e0c1 81 #:use-module (gnu packages pcre)
ceb62d54 82 #:use-module (gnu packages parallel)
66e40e00 83 #:use-module (gnu packages pdf)
4e10a221 84 #:use-module (gnu packages perl)
5ccde207 85 #:use-module (gnu packages perl-check)
4e10a221 86 #:use-module (gnu packages pkg-config)
bfe3c685 87 #:use-module (gnu packages popt)
e4e5a4d8 88 #:use-module (gnu packages protobuf)
346a829a 89 #:use-module (gnu packages python)
589e3f4e 90 #:use-module (gnu packages python-web)
ec946638 91 #:use-module (gnu packages readline)
9c38b540 92 #:use-module (gnu packages ruby)
84be3b99 93 #:use-module (gnu packages serialization)
94820951 94 #:use-module (gnu packages shells)
c833ab55 95 #:use-module (gnu packages statistics)
aa163424 96 #:use-module (gnu packages swig)
d7678942 97 #:use-module (gnu packages tbb)
97b9da68 98 #:use-module (gnu packages tex)
db7a3444 99 #:use-module (gnu packages texinfo)
2127cedb 100 #:use-module (gnu packages textutils)
43c565d2 101 #:use-module (gnu packages time)
a2950fa4 102 #:use-module (gnu packages tls)
ce7155d5 103 #:use-module (gnu packages vim)
365c8153 104 #:use-module (gnu packages web)
c833ab55 105 #:use-module (gnu packages xml)
66e40e00 106 #:use-module (gnu packages xorg)
2c9232ae 107 #:use-module (srfi srfi-1)
ce7e361f 108 #:use-module (ice-9 match))
4e10a221 109
036cd0cb
RJPB		 110(define-public r-ape
111 (package
112 (name "r-ape")
b12c6a66 113 (version "5.0")
036cd0cb
RJPB		 114 (source
115 (origin
116 (method url-fetch)
117 (uri (cran-uri "ape" version))
118 (sha256
119 (base32
b12c6a66 120 "0q59pmxawz498cb9mv5m49lhiwxib8ak94yyydz7qg8b6lpd4bn3"))))
036cd0cb 121 (build-system r-build-system)
aeb64f3c
RW		 122 (propagated-inputs
123 `(("r-lattice" ,r-lattice)
b12c6a66
RW		 124 ("r-nlme" ,r-nlme)
125 ("r-rcpp" ,r-rcpp)))
036cd0cb
RJPB		 126 (home-page "http://ape-package.ird.fr/")
127 (synopsis "Analyses of phylogenetics and evolution")
128 (description
129 "This package provides functions for reading, writing, plotting, and
130manipulating phylogenetic trees, analyses of comparative data in a
131phylogenetic framework, ancestral character analyses, analyses of
132diversification and macroevolution, computing distances from DNA sequences,
133and several other tools.")
134 (license license:gpl2+)))
135
8dc797fa
BW		 136(define-public aragorn
137 (package
138 (name "aragorn")
e990c81d 139 (version "1.2.38")
8dc797fa
BW		 140 (source (origin
141 (method url-fetch)
142 (uri (string-append
143 "http://mbio-serv2.mbioekol.lu.se/ARAGORN/Downloads/aragorn"
144 version ".tgz"))
145 (sha256
146 (base32
e990c81d 147 "09i1rg716smlbnixfm7q1ml2mfpaa2fpn3hwjg625ysmfwwy712b"))))
8dc797fa
BW		 148 (build-system gnu-build-system)
149 (arguments
150 `(#:tests? #f ; there are no tests
151 #:phases
152 (modify-phases %standard-phases
153 (delete 'configure)
154 (replace 'build
155 (lambda _
156 (zero? (system* "gcc"
157 "-O3"
158 "-ffast-math"
159 "-finline-functions"
160 "-o"
161 "aragorn"
162 (string-append "aragorn" ,version ".c")))))
163 (replace 'install
164 (lambda* (#:key outputs #:allow-other-keys)
165 (let* ((out (assoc-ref outputs "out"))
166 (bin (string-append out "/bin"))
167 (man (string-append out "/share/man/man1")))
168 (mkdir-p bin)
f3860753 169 (install-file "aragorn" bin)
8dc797fa 170 (mkdir-p man)
f3860753 171 (install-file "aragorn.1" man))
8dc797fa
BW		 172 #t)))))
173 (home-page "http://mbio-serv2.mbioekol.lu.se/ARAGORN")
174 (synopsis "Detect tRNA, mtRNA and tmRNA genes in nucleotide sequences")
175 (description
176 "Aragorn identifies transfer RNA, mitochondrial RNA and
177transfer-messenger RNA from nucleotide sequences, based on homology to known
178tRNA consensus sequences and RNA structure. It also outputs the secondary
179structure of the predicted RNA.")
180 (license license:gpl2)))
181
a12ba6e8
BW		 182(define-public bamm
183 (package
184 (name "bamm")
4b6da268 185 (version "1.7.3")
a12ba6e8
BW		 186 (source (origin
187 (method url-fetch)
188 ;; BamM is not available on pypi.
189 (uri (string-append
4b6da268 190 "https://github.com/Ecogenomics/BamM/archive/"
a12ba6e8
BW		 191 version ".tar.gz"))
192 (file-name (string-append name "-" version ".tar.gz"))
193 (sha256
194 (base32
4b6da268 195 "1f35yxp4pc8aadsvbpg6r4kg2jh4fkjci0iby4iyljm6980sac0s"))
a12ba6e8
BW		 196 (modules '((guix build utils)))
197 (snippet
198 `(begin
199 ;; Delete bundled htslib.
200 (delete-file-recursively "c/htslib-1.3.1")
201 #t))))
202 (build-system python-build-system)
203 (arguments
204 `(#:python ,python-2 ; BamM is Python 2 only.
205 ;; Do not use bundled libhts. Do use the bundled libcfu because it has
206 ;; been modified from its original form.
207 #:configure-flags
208 (let ((htslib (assoc-ref %build-inputs "htslib")))
209 (list "--with-libhts-lib" (string-append htslib "/lib")
210 "--with-libhts-inc" (string-append htslib "/include/htslib")))
211 #:phases
212 (modify-phases %standard-phases
213 (add-after 'unpack 'autogen
214 (lambda _
215 (with-directory-excursion "c"
216 (let ((sh (which "sh")))
217 ;; Use autogen so that 'configure' works.
218 (substitute* "autogen.sh" (("/bin/sh") sh))
219 (setenv "CONFIG_SHELL" sh)
220 (substitute* "configure" (("/bin/sh") sh))
221 (zero? (system* "./autogen.sh"))))))
222 (delete 'build)
223 ;; Run tests after installation so compilation only happens once.
224 (delete 'check)
225 (add-after 'install 'wrap-executable
226 (lambda* (#:key outputs #:allow-other-keys)
227 (let* ((out (assoc-ref outputs "out"))
228 (path (getenv "PATH")))
229 (wrap-program (string-append out "/bin/bamm")
230 `("PATH" ":" prefix (,path))))
231 #t))
232 (add-after 'wrap-executable 'post-install-check
233 (lambda* (#:key inputs outputs #:allow-other-keys)
234 (setenv "PATH"
235 (string-append (assoc-ref outputs "out")
236 "/bin:"
237 (getenv "PATH")))
238 (setenv "PYTHONPATH"
239 (string-append
240 (assoc-ref outputs "out")
241 "/lib/python"
242 (string-take (string-take-right
243 (assoc-ref inputs "python") 5) 3)
244 "/site-packages:"
245 (getenv "PYTHONPATH")))
246 ;; There are 2 errors printed, but they are safe to ignore:
247 ;; 1) [E::hts_open_format] fail to open file ...
248 ;; 2) samtools view: failed to open ...
249 (zero? (system* "nosetests")))))))
250 (native-inputs
251 `(("autoconf" ,autoconf)
252 ("automake" ,automake)
253 ("libtool" ,libtool)
254 ("zlib" ,zlib)
255 ("python-nose" ,python2-nose)
f3b98f4f 256 ("python-pysam" ,python2-pysam)))
a12ba6e8 257 (inputs
bca2c576 258 `(("htslib" ,htslib-1.3) ; At least one test fails on htslib-1.4+.
a12ba6e8
BW		 259 ("samtools" ,samtools)
260 ("bwa" ,bwa)
261 ("grep" ,grep)
262 ("sed" ,sed)
263 ("coreutils" ,coreutils)))
264 (propagated-inputs
265 `(("python-numpy" ,python2-numpy)))
266 (home-page "http://ecogenomics.github.io/BamM/")
267 (synopsis "Metagenomics-focused BAM file manipulator")
268 (description
269 "BamM is a C library, wrapped in python, to efficiently generate and
270parse BAM files, specifically for the analysis of metagenomic data. For
271instance, it implements several methods to assess contig-wise read coverage.")
272 (license license:lgpl3+)))
273
9794180d
RW		 274(define-public bamtools
275 (package
276 (name "bamtools")
48d66a9c 277 (version "2.4.1")
9794180d
RW		 278 (source (origin
279 (method url-fetch)
280 (uri (string-append
281 "https://github.com/pezmaster31/bamtools/archive/v"
282 version ".tar.gz"))
283 (file-name (string-append name "-" version ".tar.gz"))
284 (sha256
285 (base32
48d66a9c 286 "0jr024kcrhjb82cm69i7p5fcg5375zlc1h3qh2n1v368hcd0qflk"))))
9794180d 287 (build-system cmake-build-system)
4702cec2
RW		 288 (arguments
289 `(#:tests? #f ;no "check" target
290 #:phases
291 (modify-phases %standard-phases
292 (add-before
293 'configure 'set-ldflags
294 (lambda* (#:key outputs #:allow-other-keys)
295 (setenv "LDFLAGS"
296 (string-append
297 "-Wl,-rpath="
298 (assoc-ref outputs "out") "/lib/bamtools")))))))
9794180d
RW		 299 (inputs `(("zlib" ,zlib)))
300 (home-page "https://github.com/pezmaster31/bamtools")
301 (synopsis "C++ API and command-line toolkit for working with BAM data")
302 (description
303 "BamTools provides both a C++ API and a command-line toolkit for handling
304BAM files.")
305 (license license:expat)))
306
bdc7be59
MB		 307(define-public bcftools
308 (package
309 (name "bcftools")
0620387a 310 (version "1.5")
bdc7be59
MB		 311 (source (origin
312 (method url-fetch)
313 (uri (string-append
314 "https://github.com/samtools/bcftools/releases/download/"
315 version "/bcftools-" version ".tar.bz2"))
316 (sha256
317 (base32
0620387a 318 "0093hkkvxmbwfaa7905s6185jymynvg42kq6sxv7fili11l5mxwz"))
3837108e 319 (patches (search-patches "bcftools-regidx-unsigned-char.patch"))
bdc7be59
MB		 320 (modules '((guix build utils)))
321 (snippet
322 ;; Delete bundled htslib.
0620387a 323 '(delete-file-recursively "htslib-1.5"))))
bdc7be59
MB		 324 (build-system gnu-build-system)
325 (arguments
326 `(#:test-target "test"
0620387a 327 #:configure-flags (list "--with-htslib=system")
bdc7be59
MB		 328 #:make-flags
329 (list
330 "USE_GPL=1"
0620387a 331 "LIBS=-lgsl -lgslcblas"
bdc7be59
MB		 332 (string-append "prefix=" (assoc-ref %outputs "out"))
333 (string-append "HTSDIR=" (assoc-ref %build-inputs "htslib") "/include")
98593f9f 334 (string-append "HTSLIB=" (assoc-ref %build-inputs "htslib") "/lib/libhts.so")
bdc7be59 335 (string-append "BGZIP=" (assoc-ref %build-inputs "htslib") "/bin/bgzip")
98593f9f
BW		 336 (string-append "TABIX=" (assoc-ref %build-inputs "htslib") "/bin/tabix")
337 (string-append "PACKAGE_VERSION=" ,version))
bdc7be59
MB		 338 #:phases
339 (modify-phases %standard-phases
bdc7be59
MB		 340 (add-before 'check 'patch-tests
341 (lambda _
342 (substitute* "test/test.pl"
343 (("/bin/bash") (which "bash")))
344 #t)))))
345 (native-inputs
346 `(("htslib" ,htslib)
347 ("perl" ,perl)))
348 (inputs
349 `(("gsl" ,gsl)
350 ("zlib" ,zlib)))
351 (home-page "https://samtools.github.io/bcftools/")
352 (synopsis "Utilities for variant calling and manipulating VCFs and BCFs")
353 (description
354 "BCFtools is a set of utilities that manipulate variant calls in the
355Variant Call Format (VCF) and its binary counterpart BCF. All commands work
356transparently with both VCFs and BCFs, both uncompressed and BGZF-compressed.")
357 ;; The sources are dual MIT/GPL, but becomes GPL-only when USE_GPL=1.
358 (license (list license:gpl3+ license:expat))))
359
8dd4ff11
RW		 360(define-public bedops
361 (package
362 (name "bedops")
1bbc3b1d 363 (version "2.4.14")
8dd4ff11
RW		 364 (source (origin
365 (method url-fetch)
366 (uri (string-append "https://github.com/bedops/bedops/archive/v"
367 version ".tar.gz"))
f586c877 368 (file-name (string-append name "-" version ".tar.gz"))
8dd4ff11
RW		 369 (sha256
370 (base32
1bbc3b1d 371 "1kqbac547wyqma81cyky9n7mkgikjpsfd3nnmcm6hpqwanqgh10v"))))
8dd4ff11
RW		 372 (build-system gnu-build-system)
373 (arguments
374 '(#:tests? #f
375 #:make-flags (list (string-append "BINDIR=" %output "/bin"))
376 #:phases
dc1d3cde
KK		 377 (modify-phases %standard-phases
378 (add-after 'unpack 'unpack-tarballs
379 (lambda _
380 ;; FIXME: Bedops includes tarballs of minimally patched upstream
381 ;; libraries jansson, zlib, and bzip2. We cannot just use stock
382 ;; libraries because at least one of the libraries (zlib) is
383 ;; patched to add a C++ function definition (deflateInit2cpp).
384 ;; Until the Bedops developers offer a way to link against system
385 ;; libraries we have to build the in-tree copies of these three
386 ;; libraries.
387
388 ;; See upstream discussion:
389 ;; https://github.com/bedops/bedops/issues/124
390
391 ;; Unpack the tarballs to benefit from shebang patching.
392 (with-directory-excursion "third-party"
393 (and (zero? (system* "tar" "xvf" "jansson-2.6.tar.bz2"))
394 (zero? (system* "tar" "xvf" "zlib-1.2.7.tar.bz2"))
395 (zero? (system* "tar" "xvf" "bzip2-1.0.6.tar.bz2"))))
396 ;; Disable unpacking of tarballs in Makefile.
397 (substitute* "system.mk/Makefile.linux"
398 (("^\tbzcat .*") "\t@echo \"not unpacking\"\n")
399 (("\\./configure") "CONFIG_SHELL=bash ./configure"))
400 (substitute* "third-party/zlib-1.2.7/Makefile.in"
401 (("^SHELL=.*$") "SHELL=bash\n"))
402 #t))
403 (delete 'configure))))
8dd4ff11
RW		 404 (home-page "https://github.com/bedops/bedops")
405 (synopsis "Tools for high-performance genomic feature operations")
406 (description
407 "BEDOPS is a suite of tools to address common questions raised in genomic
408studies---mostly with regard to overlap and proximity relationships between
409data sets. It aims to be scalable and flexible, facilitating the efficient
410and accurate analysis and management of large-scale genomic data.
411
412BEDOPS provides tools that perform highly efficient and scalable Boolean and
413other set operations, statistical calculations, archiving, conversion and
414other management of genomic data of arbitrary scale. Tasks can be easily
415split by chromosome for distributing whole-genome analyses across a
416computational cluster.")
417 (license license:gpl2+)))
418
81de5647
RW		 419(define-public bedtools
420 (package
421 (name "bedtools")
d285657e 422 (version "2.26.0")
81de5647
RW		 423 (source (origin
424 (method url-fetch)
425 (uri (string-append "https://github.com/arq5x/bedtools2/archive/v"
426 version ".tar.gz"))
f586c877 427 (file-name (string-append name "-" version ".tar.gz"))
81de5647
RW		 428 (sha256
429 (base32
d285657e 430 "0xvri5hnp2iim1cx6mcd5d9f102p5ql41x69rd6106x1c17pinqm"))))
81de5647
RW		 431 (build-system gnu-build-system)
432 (native-inputs `(("python" ,python-2)))
433 (inputs `(("samtools" ,samtools)
434 ("zlib" ,zlib)))
435 (arguments
436 '(#:test-target "test"
437 #:phases
6573ac82 438 (modify-phases %standard-phases
6573ac82
BW		 439 (delete 'configure)
440 (replace 'install
441 (lambda* (#:key outputs #:allow-other-keys)
442 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
443 (for-each (lambda (file)
444 (install-file file bin))
445 (find-files "bin" ".*")))
446 #t)))))
81de5647
RW		 447 (home-page "https://github.com/arq5x/bedtools2")
448 (synopsis "Tools for genome analysis and arithmetic")
449 (description
450 "Collectively, the bedtools utilities are a swiss-army knife of tools for
451a wide-range of genomics analysis tasks. The most widely-used tools enable
452genome arithmetic: that is, set theory on the genome. For example, bedtools
453allows one to intersect, merge, count, complement, and shuffle genomic
454intervals from multiple files in widely-used genomic file formats such as BAM,
455BED, GFF/GTF, VCF.")
456 (license license:gpl2)))
457
9a8f309c
RW		 458;; Later releases of bedtools produce files with more columns than
459;; what Ribotaper expects.
460(define-public bedtools-2.18
461 (package (inherit bedtools)
462 (name "bedtools")
463 (version "2.18.0")
464 (source (origin
465 (method url-fetch)
466 (uri (string-append "https://github.com/arq5x/bedtools2/"
467 "archive/v" version ".tar.gz"))
468 (file-name (string-append name "-" version ".tar.gz"))
469 (sha256
470 (base32
471 "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))))
472
17dc32a4
RW		 473(define-public ribotaper
474 (package
475 (name "ribotaper")
476 (version "1.3.1")
477 (source (origin
478 (method url-fetch)
479 (uri (string-append "https://ohlerlab.mdc-berlin.de/"
480 "files/RiboTaper/RiboTaper_Version_"
481 version ".tar.gz"))
482 (sha256
483 (base32
484 "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
485 (build-system gnu-build-system)
486 (inputs
487 `(("bedtools" ,bedtools-2.18)
488 ("samtools" ,samtools-0.1)
2d7c4ae3 489 ("r-minimal" ,r-minimal)
17dc32a4
RW		 490 ("r-foreach" ,r-foreach)
491 ("r-xnomial" ,r-xnomial)
492 ("r-domc" ,r-domc)
493 ("r-multitaper" ,r-multitaper)
494 ("r-seqinr" ,r-seqinr)))
495 (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
496 (synopsis "Define translated ORFs using ribosome profiling data")
497 (description
498 "Ribotaper is a method for defining translated @dfn{open reading
499frames} (ORFs) using ribosome profiling (ribo-seq) data. This package
500provides the Ribotaper pipeline.")
501 (license license:gpl3+)))
502
769fc6bb
RW		 503(define-public ribodiff
504 (package
505 (name "ribodiff")
506 (version "0.2.2")
507 (source
508 (origin
509 (method url-fetch)
510 (uri (string-append "https://github.com/ratschlab/RiboDiff/"
511 "archive/v" version ".tar.gz"))
512 (file-name (string-append name "-" version ".tar.gz"))
513 (sha256
514 (base32
515 "0wpbwmfv05wdjxv7ikm664f7s7p7cqr8jnw99zrda0q67rl50aaj"))))
516 (build-system python-build-system)
517 (arguments
518 `(#:python ,python-2
519 #:phases
520 (modify-phases %standard-phases
521 ;; Generate an installable executable script wrapper.
522 (add-after 'unpack 'patch-setup.py
523 (lambda _
524 (substitute* "setup.py"
525 (("^(.*)packages=.*" line prefix)
526 (string-append line "\n"
527 prefix "scripts=['scripts/TE.py'],\n")))
528 #t)))))
529 (inputs
530 `(("python-numpy" ,python2-numpy)
531 ("python-matplotlib" ,python2-matplotlib)
532 ("python-scipy" ,python2-scipy)
533 ("python-statsmodels" ,python2-statsmodels)))
2bb12f5a
MB		 534 (native-inputs
535 `(("python-mock" ,python2-mock)
536 ("python-nose" ,python2-nose)))
769fc6bb
RW		 537 (home-page "http://public.bmi.inf.ethz.ch/user/zhongy/RiboDiff/")
538 (synopsis "Detect translation efficiency changes from ribosome footprints")
539 (description "RiboDiff is a statistical tool that detects the protein
540translational efficiency change from Ribo-Seq (ribosome footprinting) and
541RNA-Seq data. It uses a generalized linear model to detect genes showing
542difference in translational profile taking mRNA abundance into account. It
543facilitates us to decipher the translational regulation that behave
544independently with transcriptional regulation.")
545 (license license:gpl3+)))
546
a0a71439
RJ		 547(define-public bioawk
548 (package
549 (name "bioawk")
550 (version "1.0")
551 (source (origin
552 (method url-fetch)
553 (uri (string-append "https://github.com/lh3/bioawk/archive/v"
554 version ".tar.gz"))
555 (file-name (string-append name "-" version ".tar.gz"))
556 (sha256
557 (base32 "1daizxsk17ahi9n58fj8vpgwyhzrzh54bzqhanjanp88kgrz7gjw"))))
558 (build-system gnu-build-system)
559 (inputs
560 `(("zlib" ,zlib)))
561 (native-inputs
562 `(("bison" ,bison)))
563 (arguments
564 `(#:tests? #f ; There are no tests to run.
565 ;; Bison must generate files, before other targets can build.
566 #:parallel-build? #f
567 #:phases
568 (modify-phases %standard-phases
569 (delete 'configure) ; There is no configure phase.
570 (replace 'install
571 (lambda* (#:key outputs #:allow-other-keys)
572 (let* ((out (assoc-ref outputs "out"))
573 (bin (string-append out "/bin"))
574 (man (string-append out "/share/man/man1")))
575 (mkdir-p man)
576 (copy-file "awk.1" (string-append man "/bioawk.1"))
577 (install-file "bioawk" bin)))))))
578 (home-page "https://github.com/lh3/bioawk")
579 (synopsis "AWK with bioinformatics extensions")
580 (description "Bioawk is an extension to Brian Kernighan's awk, adding the
581support of several common biological data formats, including optionally gzip'ed
582BED, GFF, SAM, VCF, FASTA/Q and TAB-delimited formats with column names. It
583also adds a few built-in functions and a command line option to use TAB as the
584input/output delimiter. When the new functionality is not used, bioawk is
585intended to behave exactly the same as the original BWK awk.")
586 (license license:x11)))
587
a2fb1492
RW		 588(define-public python2-pybedtools
589 (package
590 (name "python2-pybedtools")
591 (version "0.6.9")
592 (source (origin
593 (method url-fetch)
594 (uri (string-append
595 "https://pypi.python.org/packages/source/p/pybedtools/pybedtools-"
596 version ".tar.gz"))
597 (sha256
598 (base32
599 "1ldzdxw1p4y3g2ignmggsdypvqkcwqwzhdha4rbgpih048z5p4an"))))
600 (build-system python-build-system)
601 (arguments `(#:python ,python-2)) ; no Python 3 support
602 (inputs
f2516de2 603 `(("python-matplotlib" ,python2-matplotlib)))
a2fb1492
RW		 604 (propagated-inputs
605 `(("bedtools" ,bedtools)
606 ("samtools" ,samtools)))
607 (native-inputs
f2516de2
HG		 608 `(("python-cython" ,python2-cython)
609 ("python-pyyaml" ,python2-pyyaml)
f3b98f4f 610 ("python-nose" ,python2-nose)))
a2fb1492
RW		 611 (home-page "https://pythonhosted.org/pybedtools/")
612 (synopsis "Python wrapper for BEDtools programs")
613 (description
614 "pybedtools is a Python wrapper for Aaron Quinlan's BEDtools programs,
615which are widely used for genomic interval manipulation or \"genome algebra\".
616pybedtools extends BEDTools by offering feature-level manipulations from with
617Python.")
618 (license license:gpl2+)))
619
9e12eba8
BW		 620(define-public python-biom-format
621 (package
622 (name "python-biom-format")
abc08cba 623 (version "2.1.6")
9e12eba8
BW		 624 (source
625 (origin
626 (method url-fetch)
627 ;; Use GitHub as source because PyPI distribution does not contain
628 ;; test data: https://github.com/biocore/biom-format/issues/693
629 (uri (string-append "https://github.com/biocore/biom-format/archive/"
630 version ".tar.gz"))
631 (file-name (string-append name "-" version ".tar.gz"))
632 (sha256
633 (base32
abc08cba 634 "08cr7wpahk6zb31h4bs7jmzpvxcqv9s13xz40h6y2h656jvdvnpj"))))
9e12eba8 635 (build-system python-build-system)
de96ea28 636 (propagated-inputs
9e12eba8
BW		 637 `(("python-numpy" ,python-numpy)
638 ("python-scipy" ,python-scipy)
639 ("python-future" ,python-future)
640 ("python-click" ,python-click)
abc08cba
BW		 641 ("python-h5py" ,python-h5py)
642 ("python-pandas" ,python-pandas)))
da5ebd10
MB		 643 (native-inputs
644 `(("python-nose" ,python-nose)))
9e12eba8
BW		 645 (home-page "http://www.biom-format.org")
646 (synopsis "Biological Observation Matrix (BIOM) format utilities")
647 (description
648 "The BIOM file format is designed to be a general-use format for
649representing counts of observations e.g. operational taxonomic units, KEGG
650orthology groups or lipid types, in one or more biological samples
651e.g. microbiome samples, genomes, metagenomes.")
652 (license license:bsd-3)
653 (properties `((python2-variant . ,(delay python2-biom-format))))))
654
655(define-public python2-biom-format
656 (let ((base (package-with-python2 (strip-python2-variant python-biom-format))))
657 (package
658 (inherit base)
659 (arguments
660 `(#:phases
661 (modify-phases %standard-phases
662 ;; Do not require the unmaintained pyqi library.
663 (add-after 'unpack 'remove-pyqi
664 (lambda _
665 (substitute* "setup.py"
666 (("install_requires.append\\(\"pyqi\"\\)") "pass"))
667 #t)))
00e10c6e 668 ,@(package-arguments base))))))
9e12eba8 669
f7283db3
RW		 670(define-public bioperl-minimal
671 (let* ((inputs `(("perl-module-build" ,perl-module-build)
672 ("perl-data-stag" ,perl-data-stag)
673 ("perl-libwww" ,perl-libwww)
674 ("perl-uri" ,perl-uri)))
675 (transitive-inputs
676 (map (compose package-name cadr)
677 (delete-duplicates
678 (concatenate
679 (map (compose package-transitive-target-inputs cadr) inputs))))))
680 (package
681 (name "bioperl-minimal")
c70271ec 682 (version "1.7.0")
f7283db3
RW		 683 (source
684 (origin
685 (method url-fetch)
c70271ec
RW		 686 (uri (string-append "https://github.com/bioperl/bioperl-live/"
687 "archive/release-"
688 (string-map (lambda (c)
689 (if (char=? c #\.)
690 #\- c)) version)
691 ".tar.gz"))
f7283db3
RW		 692 (sha256
693 (base32
c70271ec 694 "12phgpxwgkqflkwfb9dcqg7a31dpjlfhar8wcgv0aj5ln4akfz06"))))
f7283db3
RW		 695 (build-system perl-build-system)
696 (arguments
697 `(#:phases
698 (modify-phases %standard-phases
699 (add-after
700 'install 'wrap-programs
701 (lambda* (#:key outputs #:allow-other-keys)
702 ;; Make sure all executables in "bin" find the required Perl
703 ;; modules at runtime. As the PERL5LIB variable contains also
704 ;; the paths of native inputs, we pick the transitive target
705 ;; inputs from %build-inputs.
706 (let* ((out (assoc-ref outputs "out"))
707 (bin (string-append out "/bin/"))
708 (path (string-join
709 (cons (string-append out "/lib/perl5/site_perl")
710 (map (lambda (name)
711 (assoc-ref %build-inputs name))
712 ',transitive-inputs))
713 ":")))
714 (for-each (lambda (file)
715 (wrap-program file
716 `("PERL5LIB" ":" prefix (,path))))
717 (find-files bin "\\.pl$"))
718 #t))))))
719 (inputs inputs)
720 (native-inputs
721 `(("perl-test-most" ,perl-test-most)))
722 (home-page "http://search.cpan.org/dist/BioPerl")
723 (synopsis "Bioinformatics toolkit")
724 (description
725 "BioPerl is the product of a community effort to produce Perl code which
726is useful in biology. Examples include Sequence objects, Alignment objects
727and database searching objects. These objects not only do what they are
728advertised to do in the documentation, but they also interact - Alignment
729objects are made from the Sequence objects, Sequence objects have access to
730Annotation and SeqFeature objects and databases, Blast objects can be
731converted to Alignment objects, and so on. This means that the objects
732provide a coordinated and extensible framework to do computational biology.")
2f3108ad 733 (license license:perl-license))))
f7283db3 734
85c37e29
RW		 735(define-public python-biopython
736 (package
737 (name "python-biopython")
af6ce610 738 (version "1.70")
85c37e29
RW		 739 (source (origin
740 (method url-fetch)
e815c094
BW		 741 ;; use PyPi rather than biopython.org to ease updating
742 (uri (pypi-uri "biopython" version))
85c37e29
RW		 743 (sha256
744 (base32
af6ce610 745 "0nz4n9d2y2dg849gn1z0vjlkwcpzzkzy3fij7x94a6ixy2c54z2a"))))
85c37e29 746 (build-system python-build-system)
4ce60305
BW		 747 (arguments
748 `(#:phases
749 (modify-phases %standard-phases
750 (add-before 'check 'set-home
751 ;; Some tests require a home directory to be set.
752 (lambda _ (setenv "HOME" "/tmp") #t)))))
f22efa01 753 (propagated-inputs
85c37e29 754 `(("python-numpy" ,python-numpy)))
85c37e29
RW		 755 (home-page "http://biopython.org/")
756 (synopsis "Tools for biological computation in Python")
757 (description
758 "Biopython is a set of tools for biological computation including parsers
759for bioinformatics files into Python data structures; interfaces to common
760bioinformatics programs; a standard sequence class and tools for performing
761common operations on them; code to perform data classification; code for
762dealing with alignments; code making it easy to split up parallelizable tasks
763into separate processes; and more.")
5c31f4aa 764 (license (license:non-copyleft "http://www.biopython.org/DIST/LICENSE"))))
85c37e29
RW		 765
766(define-public python2-biopython
5c31f4aa 767 (package-with-python2 python-biopython))
85c37e29 768
4b1a1528
BW		 769;; An outdated version of biopython is required for seqmagick, see
770;; https://github.com/fhcrc/seqmagick/issues/59
771;; When that issue has been resolved this package should be removed.
772(define python2-biopython-1.66
773 (package
774 (inherit python2-biopython)
775 (version "1.66")
776 (source (origin
777 (method url-fetch)
778 (uri (pypi-uri "biopython" version))
779 (sha256
780 (base32
781 "1gdv92593klimg22icf5j9by7xiq86jnwzkpz4abaa05ylkdf6hp"))))))
782
985d8411
BW		 783(define-public bpp-core
784 ;; The last release was in 2014 and the recommended way to install from source
785 ;; is to clone the git repository, so we do this.
786 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
787 (let ((commit "7d8bced0d1a87291ea8dd7046b7fb5ff9c35c582"))
788 (package
789 (name "bpp-core")
790 (version (string-append "2.2.0-1." (string-take commit 7)))
791 (source (origin
792 (method git-fetch)
793 (uri (git-reference
794 (url "http://biopp.univ-montp2.fr/git/bpp-core")
795 (commit commit)))
796 (file-name (string-append name "-" version "-checkout"))
797 (sha256
798 (base32
799 "10djsq5vlnkilv436gnmh4irpk49v29pa69r6xiryg32xmvn909j"))))
800 (build-system cmake-build-system)
801 (arguments
802 `(#:parallel-build? #f))
803 (inputs
804 `(("gcc" ,gcc-5))) ; Compilation of bpp-phyl fails with GCC 4.9 so we
805 ; compile all of the bpp packages with GCC 5.
806 (home-page "http://biopp.univ-montp2.fr")
807 (synopsis "C++ libraries for Bioinformatics")
808 (description
809 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
810analysis, phylogenetics, molecular evolution and population genetics. It is
811Object Oriented and is designed to be both easy to use and computer efficient.
812Bio++ intends to help programmers to write computer expensive programs, by
813providing them a set of re-usable tools.")
814 (license license:cecill-c))))
815
8b5f4d57
BW		 816(define-public bpp-phyl
817 ;; The last release was in 2014 and the recommended way to install from source
818 ;; is to clone the git repository, so we do this.
819 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
820 (let ((commit "0c07167b629f68b569bf274d1ad0c4af83276ae2"))
821 (package
822 (name "bpp-phyl")
823 (version (string-append "2.2.0-1." (string-take commit 7)))
824 (source (origin
825 (method git-fetch)
826 (uri (git-reference
827 (url "http://biopp.univ-montp2.fr/git/bpp-phyl")
828 (commit commit)))
829 (file-name (string-append name "-" version "-checkout"))
830 (sha256
831 (base32
832 "1ssjgchzwj3iai26kyly7gwkdv8sk59nqhkb1wpap3sf5m6kyllh"))))
833 (build-system cmake-build-system)
834 (arguments
835 `(#:parallel-build? #f
836 ;; If out-of-source, test data is not copied into the build directory
837 ;; so the tests fail.
838 #:out-of-source? #f))
839 (inputs
840 `(("bpp-core" ,bpp-core)
841 ("bpp-seq" ,bpp-seq)
842 ;; GCC 4.8 fails due to an 'internal compiler error', so we use a more
843 ;; modern GCC.
844 ("gcc" ,gcc-5)))
845 (home-page "http://biopp.univ-montp2.fr")
846 (synopsis "Bio++ phylogenetic Library")
847 (description
848 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
849analysis, phylogenetics, molecular evolution and population genetics. This
850library provides phylogenetics-related modules.")
851 (license license:cecill-c))))
852
159a7016
BW		 853(define-public bpp-popgen
854 ;; The last release was in 2014 and the recommended way to install from source
855 ;; is to clone the git repository, so we do this.
856 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
857 (let ((commit "e472bac9b1a148803895d747cd6d0c5904f85d9f"))
858 (package
859 (name "bpp-popgen")
860 (version (string-append "2.2.0-1." (string-take commit 7)))
861 (source (origin
862 (method git-fetch)
863 (uri (git-reference
864 (url "http://biopp.univ-montp2.fr/git/bpp-popgen")
865 (commit commit)))
866 (file-name (string-append name "-" version "-checkout"))
867 (sha256
868 (base32
869 "0yn82dzn1n5629nzja68xfrhi655709rjanyryb36vzkmymy6dw5"))))
870 (build-system cmake-build-system)
871 (arguments
872 `(#:parallel-build? #f
873 #:tests? #f)) ; There are no tests.
874 (inputs
875 `(("bpp-core" ,bpp-core)
876 ("bpp-seq" ,bpp-seq)
877 ("gcc" ,gcc-5)))
878 (home-page "http://biopp.univ-montp2.fr")
879 (synopsis "Bio++ population genetics library")
880 (description
881 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
882analysis, phylogenetics, molecular evolution and population genetics. This
883library provides population genetics-related modules.")
884 (license license:cecill-c))))
885
70f1bc05
BW		 886(define-public bpp-seq
887 ;; The last release was in 2014 and the recommended way to install from source
888 ;; is to clone the git repository, so we do this.
889 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
890 (let ((commit "6cfa07965ce152e5598a89df2fa80a75973bfa33"))
891 (package
892 (name "bpp-seq")
893 (version (string-append "2.2.0-1." (string-take commit 7)))
894 (source (origin
895 (method git-fetch)
896 (uri (git-reference
897 (url "http://biopp.univ-montp2.fr/git/bpp-seq")
898 (commit commit)))
899 (file-name (string-append name "-" version "-checkout"))
900 (sha256
901 (base32
902 "1nys5jq7jqvdg40d91wsmj3q2yzy4276cp7sp44n67p468f27zf2"))))
903 (build-system cmake-build-system)
904 (arguments
905 `(#:parallel-build? #f
906 ;; If out-of-source, test data is not copied into the build directory
907 ;; so the tests fail.
908 #:out-of-source? #f))
909 (inputs
910 `(("bpp-core" ,bpp-core)
911 ("gcc" ,gcc-5))) ; Use GCC 5 as per 'bpp-core'.
912 (home-page "http://biopp.univ-montp2.fr")
913 (synopsis "Bio++ sequence library")
914 (description
915 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
916analysis, phylogenetics, molecular evolution and population genetics. This
917library provides sequence-related modules.")
918 (license license:cecill-c))))
919
db7a3444
BW		 920(define-public bppsuite
921 ;; The last release was in 2014 and the recommended way to install from source
922 ;; is to clone the git repository, so we do this.
923 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
924 (let ((commit "c516147f57aa50961121cd505bed52cd7603698b"))
925 (package
926 (name "bppsuite")
927 (version (string-append "2.2.0-1." (string-take commit 7)))
928 (source (origin
929 (method git-fetch)
930 (uri (git-reference
931 (url "http://biopp.univ-montp2.fr/git/bppsuite")
932 (commit commit)))
933 (file-name (string-append name "-" version "-checkout"))
934 (sha256
935 (base32
936 "1y87pxvw0jxjizhq2dr9g2r91md45k1p9ih2sl1yy1y3p934l2kb"))))
937 (build-system cmake-build-system)
938 (arguments
939 `(#:parallel-build? #f
940 #:tests? #f)) ; There are no tests.
941 (native-inputs
942 `(("groff" ,groff)
943 ("man-db" ,man-db)
944 ("texinfo" ,texinfo)))
945 (inputs
946 `(("bpp-core" ,bpp-core)
947 ("bpp-seq" ,bpp-seq)
948 ("bpp-phyl" ,bpp-phyl)
949 ("bpp-phyl" ,bpp-popgen)
950 ("gcc" ,gcc-5)))
951 (home-page "http://biopp.univ-montp2.fr")
952 (synopsis "Bioinformatics tools written with the Bio++ libraries")
953 (description
954 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
955analysis, phylogenetics, molecular evolution and population genetics. This
956package provides command line tools using the Bio++ library.")
957 (license license:cecill-c))))
958
82c370de
RW		 959(define-public blast+
960 (package
961 (name "blast+")
4732e6ee 962 (version "2.6.0")
82c370de
RW		 963 (source (origin
964 (method url-fetch)
965 (uri (string-append
966 "ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/"
967 version "/ncbi-blast-" version "+-src.tar.gz"))
968 (sha256
969 (base32
4732e6ee
BW		 970 "15n937pw5aqmyfjb6l387d18grqbb96l63d5xj4l7yyh0zbf2405"))
971 (patches (search-patches "blast+-fix-makefile.patch"))
82c370de
RW		 972 (modules '((guix build utils)))
973 (snippet
974 '(begin
4732e6ee 975 ;; Remove bundled bzip2, zlib and pcre.
82c370de
RW		 976 (delete-file-recursively "c++/src/util/compress/bzip2")
977 (delete-file-recursively "c++/src/util/compress/zlib")
4732e6ee 978 (delete-file-recursively "c++/src/util/regexp")
82c370de
RW		 979 (substitute* "c++/src/util/compress/Makefile.in"
980 (("bzip2 zlib api") "api"))
981 ;; Remove useless msbuild directory
982 (delete-file-recursively
983 "c++/src/build-system/project_tree_builder/msbuild")
984 #t))))
985 (build-system gnu-build-system)
986 (arguments
4732e6ee 987 `(;; There are two(!) tests for this massive library, and both fail with
82c370de 988 ;; "unparsable timing stats".
82c370de
RW		 989 ;; ERR [127] -- [serial/datatool] datatool.sh (unparsable timing stats)
990 ;; ERR [127] -- [serial/datatool] datatool_xml.sh (unparsable timing stats)
991 #:tests? #f
992 #:out-of-source? #t
993 #:parallel-build? #f ; not supported
994 #:phases
995 (modify-phases %standard-phases
996 (add-before
997 'configure 'set-HOME
998 ;; $HOME needs to be set at some point during the configure phase
999 (lambda _ (setenv "HOME" "/tmp") #t))
1000 (add-after
1001 'unpack 'enter-dir
1002 (lambda _ (chdir "c++") #t))
1003 (add-after
1004 'enter-dir 'fix-build-system
1005 (lambda _
1006 (define (which* cmd)
1007 (cond ((string=? cmd "date")
1008 ;; make call to "date" deterministic
1009 "date -d @0")
1010 ((which cmd)
1011 => identity)
1012 (else
1013 (format (current-error-port)
1014 "WARNING: Unable to find absolute path for ~s~%"
1015 cmd)
1016 #f)))
1017
1018 ;; Rewrite hardcoded paths to various tools
1019 (substitute* (append '("src/build-system/configure.ac"
1020 "src/build-system/configure"
4732e6ee 1021 "src/build-system/helpers/run_with_lock.c"
82c370de
RW		 1022 "scripts/common/impl/if_diff.sh"
1023 "scripts/common/impl/run_with_lock.sh"
1024 "src/build-system/Makefile.configurables.real"
1025 "src/build-system/Makefile.in.top"
1026 "src/build-system/Makefile.meta.gmake=no"
1027 "src/build-system/Makefile.meta.in"
1028 "src/build-system/Makefile.meta_l"
1029 "src/build-system/Makefile.meta_p"
1030 "src/build-system/Makefile.meta_r"
1031 "src/build-system/Makefile.mk.in"
1032 "src/build-system/Makefile.requirements"
1033 "src/build-system/Makefile.rules_with_autodep.in")
1034 (find-files "scripts/common/check" "\\.sh$"))
1035 (("(/usr/bin/|/bin/)([a-z][-_.a-z]*)" all dir cmd)
1036 (or (which* cmd) all)))
1037
1038 (substitute* (find-files "src/build-system" "^config.*")
1039 (("LN_S=/bin/\\$LN_S") (string-append "LN_S=" (which "ln")))
1040 (("^PATH=.*") ""))
1041
1042 ;; rewrite "/var/tmp" in check script
1043 (substitute* "scripts/common/check/check_make_unix.sh"
1044 (("/var/tmp") "/tmp"))
1045
1046 ;; do not reset PATH
1047 (substitute* (find-files "scripts/common/impl/" "\\.sh$")
1048 (("^ *PATH=.*") "")
1049 (("action=/bin/") "action=")
1050 (("export PATH") ":"))
1051 #t))
1052 (replace
1053 'configure
1054 (lambda* (#:key inputs outputs #:allow-other-keys)
1055 (let ((out (assoc-ref outputs "out"))
1056 (lib (string-append (assoc-ref outputs "lib") "/lib"))
1057 (include (string-append (assoc-ref outputs "include")
1058 "/include/ncbi-tools++")))
1059 ;; The 'configure' script doesn't recognize things like
1060 ;; '--enable-fast-install'.
1061 (zero? (system* "./configure.orig"
1062 (string-append "--with-build-root=" (getcwd) "/build")
1063 (string-append "--prefix=" out)
1064 (string-append "--libdir=" lib)
1065 (string-append "--includedir=" include)
1066 (string-append "--with-bz2="
1067 (assoc-ref inputs "bzip2"))
1068 (string-append "--with-z="
1069 (assoc-ref inputs "zlib"))
4732e6ee
BW		 1070 (string-append "--with-pcre="
1071 (assoc-ref inputs "pcre"))
82c370de
RW		 1072 ;; Each library is built twice by default, once
1073 ;; with "-static" in its name, and again
1074 ;; without.
1075 "--without-static"
1076 "--with-dll"))))))))
4732e6ee
BW		 1077 (outputs '("out" ; 21 MB
1078 "lib" ; 226 MB
1079 "include")) ; 33 MB
82c370de
RW		 1080 (inputs
1081 `(("bzip2" ,bzip2)
4732e6ee
BW		 1082 ("zlib" ,zlib)
1083 ("pcre" ,pcre)
1084 ("perl" ,perl)
1085 ("python" ,python-wrapper)))
82c370de
RW		 1086 (native-inputs
1087 `(("cpio" ,cpio)))
1088 (home-page "http://blast.ncbi.nlm.nih.gov")
1089 (synopsis "Basic local alignment search tool")
1090 (description
1091 "BLAST is a popular method of performing a DNA or protein sequence
1092similarity search, using heuristics to produce results quickly. It also
1093calculates an “expect value” that estimates how many matches would have
1094occurred at a given score by chance, which can aid a user in judging how much
1095confidence to have in an alignment.")
1096 ;; Most of the sources are in the public domain, with the following
1097 ;; exceptions:
1098 ;; * Expat:
1099 ;; * ./c++/include/util/bitset/
1100 ;; * ./c++/src/html/ncbi_menu*.js
1101 ;; * Boost license:
1102 ;; * ./c++/include/util/impl/floating_point_comparison.hpp
1103 ;; * LGPL 2+:
1104 ;; * ./c++/include/dbapi/driver/odbc/unix_odbc/
1105 ;; * ASL 2.0:
1106 ;; * ./c++/src/corelib/teamcity_*
1107 (license (list license:public-domain
1108 license:expat
1109 license:boost1.0
1110 license:lgpl2.0+
1111 license:asl2.0))))
1112
6c2b26e2
RW		 1113(define-public bless
1114 (package
1115 (name "bless")
1116 (version "1p02")
1117 (source (origin
1118 (method url-fetch)
1119 (uri (string-append "mirror://sourceforge/bless-ec/bless.v"
1120 version ".tgz"))
1121 (sha256
1122 (base32
4d75e03a
RW		 1123 "0rm0gw2s18dqwzzpl3c2x1z05ni2v0xz5dmfk3d33j6g4cgrlrdd"))
1124 (modules '((guix build utils)))
6c2b26e2
RW		 1125 (snippet
1126 `(begin
1127 ;; Remove bundled boost, pigz, zlib, and .git directory
953c1223
RW		 1128 ;; FIXME: also remove bundled sources for murmurhash3 and
1129 ;; kmc once packaged.
6c2b26e2
RW		 1130 (delete-file-recursively "boost")
1131 (delete-file-recursively "pigz")
953c1223 1132 (delete-file-recursively "google-sparsehash")
6c2b26e2
RW		 1133 (delete-file-recursively "zlib")
1134 (delete-file-recursively ".git")
1135 #t))))
1136 (build-system gnu-build-system)
1137 (arguments
1138 '(#:tests? #f ;no "check" target
1139 #:make-flags
1140 (list (string-append "ZLIB="
1141 (assoc-ref %build-inputs "zlib")
1142 "/lib/libz.a")
1143 (string-append "LDFLAGS="
1144 (string-join '("-lboost_filesystem"
1145 "-lboost_system"
1146 "-lboost_iostreams"
1147 "-lz"
1148 "-fopenmp"
1149 "-std=c++11"))))
1150 #:phases
1151 (modify-phases %standard-phases
1152 (add-after 'unpack 'do-not-build-bundled-pigz
1153 (lambda* (#:key inputs outputs #:allow-other-keys)
1154 (substitute* "Makefile"
1155 (("cd pigz/pigz-2.3.3; make") ""))
1156 #t))
1157 (add-after 'unpack 'patch-paths-to-executables
1158 (lambda* (#:key inputs outputs #:allow-other-keys)
1159 (substitute* "parse_args.cpp"
1160 (("kmc_binary = .*")
1161 (string-append "kmc_binary = \""
1162 (assoc-ref outputs "out")
1163 "/bin/kmc\";"))
1164 (("pigz_binary = .*")
1165 (string-append "pigz_binary = \""
1166 (assoc-ref inputs "pigz")
1167 "/bin/pigz\";")))
1168 #t))
1169 (replace 'install
1170 (lambda* (#:key outputs #:allow-other-keys)
1171 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
1172 (for-each (lambda (file)
1173 (install-file file bin))
1174 '("bless" "kmc/bin/kmc"))
1175 #t)))
1176 (delete 'configure))))
1177 (native-inputs
1178 `(("perl" ,perl)))
1179 (inputs
1180 `(("openmpi" ,openmpi)
1181 ("boost" ,boost)
953c1223 1182 ("sparsehash" ,sparsehash)
6c2b26e2
RW		 1183 ("pigz" ,pigz)
1184 ("zlib" ,zlib)))
9641a899 1185 (supported-systems '("x86_64-linux"))
3b3b60d0 1186 (home-page "https://sourceforge.net/p/bless-ec/wiki/Home/")
6c2b26e2
RW		 1187 (synopsis "Bloom-filter-based error correction tool for NGS reads")
1188 (description
1189 "@dfn{Bloom-filter-based error correction solution for high-throughput
1190sequencing reads} (BLESS) uses a single minimum-sized bloom filter is a
1191correction tool for genomic reads produced by @dfn{Next-generation
1192sequencing} (NGS). BLESS produces accurate correction results with much less
1193memory compared with previous solutions and is also able to tolerate a higher
1194false-positive rate. BLESS can extend reads like DNA assemblers to correct
1195errors at the end of reads.")
1196 (license license:gpl3+)))
1197
2c7ee167
RW		 1198(define-public bowtie
1199 (package
1200 (name "bowtie")
d6e63cf3 1201 (version "2.3.2")
2c7ee167
RW		 1202 (source (origin
1203 (method url-fetch)
1204 (uri (string-append "https://github.com/BenLangmead/bowtie2/archive/v"
1205 version ".tar.gz"))
f586c877 1206 (file-name (string-append name "-" version ".tar.gz"))
2c7ee167
RW		 1207 (sha256
1208 (base32
d6e63cf3 1209 "0hwa5r9qbglppb7sz5z79rlmmddr3n51n468jb3wh8rwjgn3yr90"))
2c7ee167
RW		 1210 (modules '((guix build utils)))
1211 (snippet
1212 '(substitute* "Makefile"
2c7ee167
RW		 1213 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
1214 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
0047d26a 1215 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\"")))))
2c7ee167 1216 (build-system gnu-build-system)
d6e63cf3
BW		 1217 (inputs
1218 `(("perl" ,perl)
1219 ("perl-clone" ,perl-clone)
1220 ("perl-test-deep" ,perl-test-deep)
1221 ("perl-test-simple" ,perl-test-simple)
1222 ("python" ,python-2)
1223 ("tbb" ,tbb)
1224 ("zlib" ,zlib)))
2c7ee167 1225 (arguments
0047d26a
RW		 1226 '(#:make-flags
1227 (list "allall"
1228 "WITH_TBB=1"
1229 (string-append "prefix=" (assoc-ref %outputs "out")))
2c7ee167 1230 #:phases
06e37236
BW		 1231 (modify-phases %standard-phases
1232 (delete 'configure)
1233 (replace 'check
1234 (lambda* (#:key outputs #:allow-other-keys)
1235 (zero? (system* "perl"
1236 "scripts/test/simple_tests.pl"
1237 "--bowtie2=./bowtie2"
1238 "--bowtie2-build=./bowtie2-build")))))))
2c7ee167
RW		 1239 (home-page "http://bowtie-bio.sourceforge.net/bowtie2/index.shtml")
1240 (synopsis "Fast and sensitive nucleotide sequence read aligner")
1241 (description
1242 "Bowtie 2 is a fast and memory-efficient tool for aligning sequencing
1243reads to long reference sequences. It is particularly good at aligning reads
1244of about 50 up to 100s or 1,000s of characters, and particularly good at
1245aligning to relatively long (e.g. mammalian) genomes. Bowtie 2 indexes the
1246genome with an FM Index to keep its memory footprint small: for the human
1247genome, its memory footprint is typically around 3.2 GB. Bowtie 2 supports
1248gapped, local, and paired-end alignment modes.")
241e1221 1249 (supported-systems '("x86_64-linux"))
2c7ee167
RW		 1250 (license license:gpl3+)))
1251
94ce537e
RW		 1252(define-public tophat
1253 (package
1254 (name "tophat")
1255 (version "2.1.0")
1256 (source (origin
1257 (method url-fetch)
1258 (uri (string-append
1259 "http://ccb.jhu.edu/software/tophat/downloads/tophat-"
1260 version ".tar.gz"))
1261 (sha256
1262 (base32
1263 "168zlzykq622zbgkh90a90f1bdgsxkscq2zxzbj8brq80hbjpyp7"))
fc1adab1 1264 (patches (search-patches "tophat-build-with-later-seqan.patch"))
94ce537e
RW		 1265 (modules '((guix build utils)))
1266 (snippet
1267 '(begin
1268 ;; Remove bundled SeqAn and samtools
1269 (delete-file-recursively "src/SeqAn-1.3")
1270 (delete-file-recursively "src/samtools-0.1.18")
1271 #t))))
1272 (build-system gnu-build-system)
1273 (arguments
1274 '(#:parallel-build? #f ; not supported
1275 #:phases
1276 (modify-phases %standard-phases
1277 (add-after 'unpack 'use-system-samtools
1278 (lambda* (#:key inputs #:allow-other-keys)
1279 (substitute* "src/Makefile.in"
1280 (("(noinst_LIBRARIES = )\\$\\(SAMLIB\\)" _ prefix) prefix)
1281 (("\\$\\(SAMPROG\\): \\$\\(SAMLIB\\)") "")
1282 (("SAMPROG = samtools_0\\.1\\.18") "")
1283 (("\\$\\(samtools_0_1_18_SOURCES\\)") "")
1284 (("am__EXEEXT_1 = samtools_0\\.1\\.18\\$\\(EXEEXT\\)") ""))
1285 (substitute* '("src/common.cpp"
1286 "src/tophat.py")
1287 (("samtools_0.1.18") (which "samtools")))
1288 (substitute* '("src/common.h"
1289 "src/bam2fastx.cpp")
1290 (("#include \"bam.h\"") "#include <samtools/bam.h>")
1291 (("#include \"sam.h\"") "#include <samtools/sam.h>"))
1292 (substitute* '("src/bwt_map.h"
1293 "src/map2gtf.h"
1294 "src/align_status.h")
1295 (("#include <bam.h>") "#include <samtools/bam.h>")
1296 (("#include <sam.h>") "#include <samtools/sam.h>"))
1297 #t)))))
1298 (inputs
1299 `(("boost" ,boost)
1300 ("bowtie" ,bowtie)
1301 ("samtools" ,samtools-0.1)
1302 ("ncurses" ,ncurses)
1303 ("python" ,python-2)
1304 ("perl" ,perl)
1305 ("zlib" ,zlib)
1306 ("seqan" ,seqan)))
1307 (home-page "http://ccb.jhu.edu/software/tophat/index.shtml")
1308 (synopsis "Spliced read mapper for RNA-Seq data")
1309 (description
1310 "TopHat is a fast splice junction mapper for nucleotide sequence
1311reads produced by the RNA-Seq method. It aligns RNA-Seq reads to
1312mammalian-sized genomes using the ultra high-throughput short read
1313aligner Bowtie, and then analyzes the mapping results to identify
1314splice junctions between exons.")
1315 ;; TopHat is released under the Boost Software License, Version 1.0
1316 ;; See https://github.com/infphilo/tophat/issues/11#issuecomment-121589893
1317 (license license:boost1.0)))
1318
9a8336d8
RW		 1319(define-public bwa
1320 (package
1321 (name "bwa")
6f141eff 1322 (version "0.7.17")
9a8336d8
RW		 1323 (source (origin
1324 (method url-fetch)
ae6e00f6
BW		 1325 (uri (string-append
1326 "https://github.com/lh3/bwa/releases/download/v"
1327 version "/bwa-" version ".tar.bz2"))
9a8336d8
RW		 1328 (sha256
1329 (base32
6f141eff 1330 "1zfhv2zg9v1icdlq4p9ssc8k01mca5d1bd87w71py2swfi74s6yy"))))
9a8336d8
RW		 1331 (build-system gnu-build-system)
1332 (arguments
1333 '(#:tests? #f ;no "check" target
1334 #:phases
dc1d3cde
KK		 1335 (modify-phases %standard-phases
1336 (replace 'install
1337 (lambda* (#:key outputs #:allow-other-keys)
1338 (let ((bin (string-append
1339 (assoc-ref outputs "out") "/bin"))
1340 (doc (string-append
1341 (assoc-ref outputs "out") "/share/doc/bwa"))
1342 (man (string-append
1343 (assoc-ref outputs "out") "/share/man/man1")))
1344 (install-file "bwa" bin)
1345 (install-file "README.md" doc)
1346 (install-file "bwa.1" man))
1347 #t))
1348 ;; no "configure" script
1349 (delete 'configure))))
9a8336d8 1350 (inputs `(("zlib" ,zlib)))
db94f8c7
RW		 1351 ;; Non-portable SSE instructions are used so building fails on platforms
1352 ;; other than x86_64.
1353 (supported-systems '("x86_64-linux"))
9a8336d8
RW		 1354 (home-page "http://bio-bwa.sourceforge.net/")
1355 (synopsis "Burrows-Wheeler sequence aligner")
1356 (description
1357 "BWA is a software package for mapping low-divergent sequences against a
1358large reference genome, such as the human genome. It consists of three
1359algorithms: BWA-backtrack, BWA-SW and BWA-MEM. The first algorithm is
1360designed for Illumina sequence reads up to 100bp, while the rest two for
1361longer sequences ranged from 70bp to 1Mbp. BWA-MEM and BWA-SW share similar
1362features such as long-read support and split alignment, but BWA-MEM, which is
1363the latest, is generally recommended for high-quality queries as it is faster
1364and more accurate. BWA-MEM also has better performance than BWA-backtrack for
136570-100bp Illumina reads.")
1366 (license license:gpl3+)))
1367
d29150b5
RW		 1368(define-public bwa-pssm
1369 (package (inherit bwa)
1370 (name "bwa-pssm")
1371 (version "0.5.11")
1372 (source (origin
1373 (method url-fetch)
1374 (uri (string-append "https://github.com/pkerpedjiev/bwa-pssm/"
1375 "archive/" version ".tar.gz"))
1376 (file-name (string-append name "-" version ".tar.gz"))
1377 (sha256
1378 (base32
1379 "02p7mpbs4mlxmn84g2x4ghak638vbj4lqix2ipx5g84pz9bhdavg"))))
1380 (build-system gnu-build-system)
1381 (inputs
1382 `(("gdsl" ,gdsl)
1383 ("zlib" ,zlib)
1384 ("perl" ,perl)))
1385 (home-page "http://bwa-pssm.binf.ku.dk/")
1386 (synopsis "Burrows-Wheeler transform-based probabilistic short read mapper")
1387 (description
1388 "BWA-PSSM is a probabilistic short genomic sequence read aligner based on
1389the use of @dfn{position specific scoring matrices} (PSSM). Like many of the
1390existing aligners it is fast and sensitive. Unlike most other aligners,
1391however, it is also adaptible in the sense that one can direct the alignment
1392based on known biases within the data set. It is coded as a modification of
1393the original BWA alignment program and shares the genome index structure as
1394well as many of the command line options.")
1395 (license license:gpl3+)))
1396
ad641d53
RW		 1397(define-public python2-bx-python
1398 (package
1399 (name "python2-bx-python")
c1dfe8c3 1400 (version "0.7.3")
ad641d53
RW		 1401 (source (origin
1402 (method url-fetch)
c1dfe8c3 1403 (uri (pypi-uri "bx-python" version))
ad641d53
RW		 1404 (sha256
1405 (base32
c1dfe8c3 1406 "15z2w3bvnc0n4qmb9bd6d8ylc2h2nj883x2w9iixf4x3vki9b22i"))
ad641d53
RW		 1407 (modules '((guix build utils)))
1408 (snippet
1409 '(substitute* "setup.py"
1410 ;; remove dependency on outdated "distribute" module
1411 (("^from distribute_setup import use_setuptools") "")
1412 (("^use_setuptools\\(\\)") "")))))
1413 (build-system python-build-system)
1414 (arguments
1415 `(#:tests? #f ;tests fail because test data are not included
1416 #:python ,python-2))
1417 (inputs
1418 `(("python-numpy" ,python2-numpy)
1419 ("zlib" ,zlib)))
1420 (native-inputs
f3b98f4f 1421 `(("python-nose" ,python2-nose)))
ad641d53
RW		 1422 (home-page "http://bitbucket.org/james_taylor/bx-python/")
1423 (synopsis "Tools for manipulating biological data")
1424 (description
1425 "bx-python provides tools for manipulating biological data, particularly
1426multiple sequence alignments.")
1427 (license license:expat)))
1428
55a9a8c2
RW		 1429(define-public python-pysam
1430 (package
1431 (name "python-pysam")
60739965 1432 (version "0.11.2.2")
d454640c
RW		 1433 (source (origin
1434 (method url-fetch)
f536dce5
MB		 1435 ;; Test data is missing on PyPi.
1436 (uri (string-append
1437 "https://github.com/pysam-developers/pysam/archive/v"
1438 version ".tar.gz"))
1439 (file-name (string-append name "-" version ".tar.gz"))
d454640c
RW		 1440 (sha256
1441 (base32
60739965 1442 "1cfqdxsqs3xhacns9n0271ck6wkc76px66ddjm91wfw2jxxfklvc"))
dff26b23
MB		 1443 (modules '((guix build utils)))
1444 (snippet
1445 ;; Drop bundled htslib. TODO: Also remove samtools and bcftools.
1446 '(delete-file-recursively "htslib"))))
55a9a8c2
RW		 1447 (build-system python-build-system)
1448 (arguments
71dbf592
RW		 1449 `(#:modules ((ice-9 ftw)
1450 (srfi srfi-26)
1451 (guix build python-build-system)
1452 (guix build utils))
1453 #:phases
397d463a
MB		 1454 (modify-phases %standard-phases
1455 (add-before 'build 'set-flags
dff26b23
MB		 1456 (lambda* (#:key inputs #:allow-other-keys)
1457 (setenv "HTSLIB_MODE" "external")
1458 (setenv "HTSLIB_LIBRARY_DIR"
1459 (string-append (assoc-ref inputs "htslib") "/lib"))
1460 (setenv "HTSLIB_INCLUDE_DIR"
1461 (string-append (assoc-ref inputs "htslib") "/include"))
397d463a
MB		 1462 (setenv "LDFLAGS" "-lncurses")
1463 (setenv "CFLAGS" "-D_CURSES_LIB=1")
f536dce5 1464 #t))
71dbf592 1465 (replace 'check
f536dce5 1466 (lambda* (#:key inputs outputs #:allow-other-keys)
71dbf592 1467 ;; Add first subdirectory of "build" directory to PYTHONPATH.
f536dce5
MB		 1468 (setenv "PYTHONPATH"
1469 (string-append
1470 (getenv "PYTHONPATH")
71dbf592
RW		 1471 ":" (getcwd) "/build/"
1472 (car (scandir "build"
e1f02f92 1473 (negate (cut string-prefix? "." <>))))))
f536dce5 1474 ;; Step out of source dir so python does not import from CWD.
71dbf592
RW		 1475 (with-directory-excursion "tests"
1476 (setenv "HOME" "/tmp")
1477 (and (zero? (system* "make" "-C" "pysam_data"))
1478 (zero? (system* "make" "-C" "cbcf_data"))
b2955e22
RW		 1479 ;; Running nosetests without explicitly asking for a
1480 ;; single process leads to a crash. Running with multiple
1481 ;; processes fails because the tests are not designed to
1482 ;; run in parallel.
31c374e0
RW		 1483
1484 ;; FIXME: tests keep timing out on some systems.
1485 ;; (zero? (system* "nosetests" "-v"
1486 ;; "--processes" "1"))
1487 )))))))
dff26b23
MB		 1488 (propagated-inputs
1489 `(("htslib" ,htslib))) ; Included from installed header files.
55a9a8c2 1490 (inputs
649e9b3b 1491 `(("ncurses" ,ncurses)
55a9a8c2 1492 ("zlib" ,zlib)))
649e9b3b
RW		 1493 (native-inputs
1494 `(("python-cython" ,python-cython)
f536dce5
MB		 1495 ;; Dependencies below are are for tests only.
1496 ("samtools" ,samtools)
1497 ("bcftools" ,bcftools)
1498 ("python-nose" ,python-nose)))
55a9a8c2
RW		 1499 (home-page "https://github.com/pysam-developers/pysam")
1500 (synopsis "Python bindings to the SAMtools C API")
1501 (description
1502 "Pysam is a Python module for reading and manipulating files in the
1503SAM/BAM format. Pysam is a lightweight wrapper of the SAMtools C API. It
1504also includes an interface for tabix.")
1505 (license license:expat)))
1506
1507(define-public python2-pysam
1508 (package-with-python2 python-pysam))
1509
4db9433a
RW		 1510(define-public python-twobitreader
1511 (package
1512 (name "python-twobitreader")
044ac8d2 1513 (version "3.1.4")
4db9433a
RW		 1514 (source (origin
1515 (method url-fetch)
1516 (uri (pypi-uri "twobitreader" version))
1517 (sha256
1518 (base32
044ac8d2 1519 "1q8wnj2kga9nz1lwc4w7qv52smfm536hp6mc8w6s53lhyj0mpi22"))))
4db9433a 1520 (build-system python-build-system)
900fb8d0
LF		 1521 (arguments
1522 '(;; Tests are not distributed in the PyPi release.
1523 ;; TODO Try building from the Git repo or asking the upstream maintainer
1524 ;; to distribute the tests on PyPi.
1525 #:tests? #f))
4db9433a
RW		 1526 (native-inputs
1527 `(("python-sphinx" ,python-sphinx)))
1528 (home-page "https://github.com/benjschiller/twobitreader")
1529 (synopsis "Python library for reading .2bit files")
1530 (description
1531 "twobitreader is a Python library for reading .2bit files as used by the
1532UCSC genome browser.")
1533 (license license:artistic2.0)))
1534
1535(define-public python2-twobitreader
5c31f4aa 1536 (package-with-python2 python-twobitreader))
4db9433a 1537
f94bf198
RW		 1538(define-public python-plastid
1539 (package
1540 (name "python-plastid")
897ab082 1541 (version "0.4.8")
f94bf198
RW		 1542 (source (origin
1543 (method url-fetch)
1544 (uri (pypi-uri "plastid" version))
1545 (sha256
1546 (base32
897ab082 1547 "0l24dd3q66if8yj042m4s0g95n6acn7im1imqd3p6h8ns43kxhj8"))))
f94bf198
RW		 1548 (build-system python-build-system)
1549 (arguments
1550 ;; Some test files are not included.
1551 `(#:tests? #f))
1552 (propagated-inputs
1553 `(("python-numpy" ,python-numpy)
1554 ("python-scipy" ,python-scipy)
1555 ("python-pandas" ,python-pandas)
1556 ("python-pysam" ,python-pysam)
1557 ("python-matplotlib" ,python-matplotlib)
1558 ("python-biopython" ,python-biopython)
99caa6f7
BW		 1559 ("python-twobitreader" ,python-twobitreader)
1560 ("python-termcolor" ,python-termcolor)))
f94bf198
RW		 1561 (native-inputs
1562 `(("python-cython" ,python-cython)
1563 ("python-nose" ,python-nose)))
1564 (home-page "https://github.com/joshuagryphon/plastid")
1565 (synopsis "Python library for genomic analysis")
1566 (description
1567 "plastid is a Python library for genomic analysis – in particular,
1568high-throughput sequencing data – with an emphasis on simplicity.")
1569 (license license:bsd-3)))
1570
1571(define-public python2-plastid
5c31f4aa 1572 (package-with-python2 python-plastid))
f94bf198 1573
6c1305f9
RW		 1574(define-public cd-hit
1575 (package
1576 (name "cd-hit")
ba773f65 1577 (version "4.6.8")
6c1305f9
RW		 1578 (source (origin
1579 (method url-fetch)
1580 (uri (string-append "https://github.com/weizhongli/cdhit"
1581 "/releases/download/V" version
ba773f65
BW		 1582 "/cd-hit-v" version
1583 "-2017-0621-source.tar.gz"))
6c1305f9
RW		 1584 (sha256
1585 (base32
d4735e8c 1586 "1b4mwm2520ixjbw57sil20f9iixzw4bkdqqwgg1fc3pzm6rz4zmn"))))
6c1305f9
RW		 1587 (build-system gnu-build-system)
1588 (arguments
1589 `(#:tests? #f ; there are no tests
1590 #:make-flags
1591 ;; Executables are copied directly to the PREFIX.
1592 (list (string-append "PREFIX=" (assoc-ref %outputs "out") "/bin"))
1593 #:phases
1594 (modify-phases %standard-phases
1595 ;; No "configure" script
1596 (delete 'configure)
1597 ;; Remove sources of non-determinism
1598 (add-after 'unpack 'be-timeless
1599 (lambda _
1600 (substitute* "cdhit-utility.c++"
1601 ((" \\(built on \" __DATE__ \"\\)") ""))
1602 (substitute* "cdhit-common.c++"
1603 (("__DATE__") "\"0\"")
1604 (("\", %s, \" __TIME__ \"\\\\n\", date") ""))
1605 #t))
ba773f65 1606 ;; The "install" target does not create the target directory.
6c1305f9
RW		 1607 (add-before 'install 'create-target-dir
1608 (lambda* (#:key outputs #:allow-other-keys)
1609 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
1610 #t)))))
1611 (inputs
1612 `(("perl" ,perl)))
1613 (home-page "http://weizhongli-lab.org/cd-hit/")
1614 (synopsis "Cluster and compare protein or nucleotide sequences")
1615 (description
1616 "CD-HIT is a program for clustering and comparing protein or nucleotide
1617sequences. CD-HIT is designed to be fast and handle extremely large
1618databases.")
1619 ;; The manual says: "It can be copied under the GNU General Public License
1620 ;; version 2 (GPLv2)."
1621 (license license:gpl2)))
1622
810cff85
RW		 1623(define-public clipper
1624 (package
1625 (name "clipper")
433530a5 1626 (version "1.1")
810cff85
RW		 1627 (source (origin
1628 (method url-fetch)
1629 (uri (string-append
1630 "https://github.com/YeoLab/clipper/archive/"
1631 version ".tar.gz"))
9ab5ea44 1632 (file-name (string-append name "-" version ".tar.gz"))
810cff85
RW		 1633 (sha256
1634 (base32
433530a5 1635 "0pflmsvhbf8izbgwhbhj1i7349sw1f55qpqj8ljmapp16hb0p0qi"))
810cff85
RW		 1636 (modules '((guix build utils)))
1637 (snippet
433530a5
RW		 1638 '(begin
1639 ;; remove unnecessary setup dependency
1640 (substitute* "setup.py"
1641 (("setup_requires = .*") ""))
1642 (for-each delete-file
1643 '("clipper/src/peaks.so"
1644 "clipper/src/readsToWiggle.so"))
1645 (delete-file-recursively "dist/")
1646 #t))))
810cff85
RW		 1647 (build-system python-build-system)
1648 (arguments `(#:python ,python-2)) ; only Python 2 is supported
1649 (inputs
92971d68 1650 `(("htseq" ,python2-htseq)
810cff85
RW		 1651 ("python-pybedtools" ,python2-pybedtools)
1652 ("python-cython" ,python2-cython)
1653 ("python-scikit-learn" ,python2-scikit-learn)
1654 ("python-matplotlib" ,python2-matplotlib)
433530a5 1655 ("python-pandas" ,python2-pandas)
810cff85
RW		 1656 ("python-pysam" ,python2-pysam)
1657 ("python-numpy" ,python2-numpy)
1658 ("python-scipy" ,python2-scipy)))
1659 (native-inputs
f3b98f4f 1660 `(("python-mock" ,python2-mock) ; for tests
d281be18 1661 ("python-nose" ,python2-nose) ; for tests
f3b98f4f 1662 ("python-pytz" ,python2-pytz))) ; for tests
810cff85
RW		 1663 (home-page "https://github.com/YeoLab/clipper")
1664 (synopsis "CLIP peak enrichment recognition")
1665 (description
1666 "CLIPper is a tool to define peaks in CLIP-seq datasets.")
1667 (license license:gpl2)))
1668
6a35566d
RS		 1669(define-public codingquarry
1670 (package
1671 (name "codingquarry")
1672 (version "2.0")
1673 (source (origin
1674 (method url-fetch)
1675 (uri (string-append
1676 "mirror://sourceforge/codingquarry/CodingQuarry_v"
1677 version ".tar.gz"))
1678 (sha256
1679 (base32
1680 "0115hkjflsnfzn36xppwf9h9avfxlavr43djqmshkkzbgjzsz60i"))))
1681 (build-system gnu-build-system)
1682 (arguments
1683 '(#:tests? #f ; no "check" target
1684 #:phases
1685 (modify-phases %standard-phases
1686 (delete 'configure)
1687 (replace 'install
1688 (lambda* (#:key outputs #:allow-other-keys)
1689 (let* ((out (assoc-ref outputs "out"))
1690 (bin (string-append out "/bin"))
1691 (doc (string-append out "/share/doc/codingquarry")))
1692 (install-file "INSTRUCTIONS.pdf" doc)
1693 (copy-recursively "QuarryFiles"
1694 (string-append out "/QuarryFiles"))
1695 (install-file "CodingQuarry" bin)
1696 (install-file "CufflinksGTF_to_CodingQuarryGFF3.py" bin)))))))
1697 (inputs `(("openmpi" ,openmpi)))
1698 (native-search-paths
1699 (list (search-path-specification
1700 (variable "QUARRY_PATH")
1701 (files '("QuarryFiles")))))
1702 (native-inputs `(("python" ,python-2))) ; Only Python 2 is supported
1703 (synopsis "Fungal gene predictor")
1704 (description "CodingQuarry is a highly accurate, self-training GHMM fungal
1705gene predictor designed to work with assembled, aligned RNA-seq transcripts.")
1706 (home-page "https://sourceforge.net/projects/codingquarry/")
1707 (license license:gpl3+)))
1708
36742f43
RW		 1709(define-public couger
1710 (package
1711 (name "couger")
1712 (version "1.8.2")
1713 (source (origin
1714 (method url-fetch)
1715 (uri (string-append
1716 "http://couger.oit.duke.edu/static/assets/COUGER"
1717 version ".zip"))
1718 (sha256
1719 (base32
1720 "04p2b14nmhzxw5h72mpzdhalv21bx4w9b87z0wpw0xzxpysyncmq"))))
1721 (build-system gnu-build-system)
1722 (arguments
1723 `(#:tests? #f
1724 #:phases
1725 (modify-phases %standard-phases
1726 (delete 'configure)
1727 (delete 'build)
1728 (replace
1729 'install
1730 (lambda* (#:key outputs #:allow-other-keys)
f3860753
TGR		 1731 (let* ((out (assoc-ref outputs "out"))
1732 (bin (string-append out "/bin")))
36742f43 1733 (copy-recursively "src" (string-append out "/src"))
f3860753 1734 (mkdir bin)
36742f43
RW		 1735 ;; Add "src" directory to module lookup path.
1736 (substitute* "couger"
1737 (("from argparse")
1738 (string-append "import sys\nsys.path.append(\""
1739 out "\")\nfrom argparse")))
f3860753 1740 (install-file "couger" bin))
36742f43
RW		 1741 #t))
1742 (add-after
1743 'install 'wrap-program
1744 (lambda* (#:key inputs outputs #:allow-other-keys)
1745 ;; Make sure 'couger' runs with the correct PYTHONPATH.
1746 (let* ((out (assoc-ref outputs "out"))
1747 (path (getenv "PYTHONPATH")))
1748 (wrap-program (string-append out "/bin/couger")
1749 `("PYTHONPATH" ":" prefix (,path))))
1750 #t)))))
1751 (inputs
1752 `(("python" ,python-2)
1753 ("python2-pillow" ,python2-pillow)
1754 ("python2-numpy" ,python2-numpy)
1755 ("python2-scipy" ,python2-scipy)
1756 ("python2-matplotlib" ,python2-matplotlib)))
1757 (propagated-inputs
2d7c4ae3 1758 `(("r-minimal" ,r-minimal)
36742f43
RW		 1759 ("libsvm" ,libsvm)
1760 ("randomjungle" ,randomjungle)))
1761 (native-inputs
1762 `(("unzip" ,unzip)))
1763 (home-page "http://couger.oit.duke.edu")
1764 (synopsis "Identify co-factors in sets of genomic regions")
1765 (description
1766 "COUGER can be applied to any two sets of genomic regions bound by
1767paralogous TFs (e.g., regions derived from ChIP-seq experiments) to identify
1768putative co-factors that provide specificity to each TF. The framework
1769determines the genomic targets uniquely-bound by each TF, and identifies a
1770small set of co-factors that best explain the in vivo binding differences
1771between the two TFs.
1772
1773COUGER uses classification algorithms (support vector machines and random
1774forests) with features that reflect the DNA binding specificities of putative
1775co-factors. The features are generated either from high-throughput TF-DNA
1776binding data (from protein binding microarray experiments), or from large
1777collections of DNA motifs.")
1778 (license license:gpl3+)))
1779
bfe3c685
RW		 1780(define-public clustal-omega
1781 (package
1782 (name "clustal-omega")
b3936f35 1783 (version "1.2.4")
bfe3c685
RW		 1784 (source (origin
1785 (method url-fetch)
b3936f35
RW		 1786 (uri (string-append "http://www.clustal.org/omega/clustal-omega-"
1787 version ".tar.gz"))
bfe3c685
RW		 1788 (sha256
1789 (base32
b3936f35 1790 "1vm30mzncwdv881vrcwg11vzvrsmwy4wg80j5i0lcfk6dlld50w6"))))
bfe3c685
RW		 1791 (build-system gnu-build-system)
1792 (inputs
1793 `(("argtable" ,argtable)))
1794 (home-page "http://www.clustal.org/omega/")
1795 (synopsis "Multiple sequence aligner for protein and DNA/RNA")
1796 (description
1797 "Clustal-Omega is a general purpose multiple sequence alignment (MSA)
1798program for protein and DNA/RNA. It produces high quality MSAs and is capable
1799of handling data-sets of hundreds of thousands of sequences in reasonable
1800time.")
1801 (license license:gpl2+)))
1802
191c7101
RW		 1803(define-public crossmap
1804 (package
1805 (name "crossmap")
61d5fd03 1806 (version "0.2.1")
191c7101
RW		 1807 (source (origin
1808 (method url-fetch)
1809 (uri (string-append "mirror://sourceforge/crossmap/CrossMap-"
1810 version ".tar.gz"))
1811 (sha256
1812 (base32
61d5fd03
RW		 1813 "07y179f63d7qnzdvkqcziwk9bs3k4zhp81q392fp1hwszjdvy22f"))
1814 ;; This patch has been sent upstream already and is available
1815 ;; for download from Sourceforge, but it has not been merged.
fc1adab1 1816 (patches (search-patches "crossmap-allow-system-pysam.patch"))
191c7101
RW		 1817 (modules '((guix build utils)))
1818 ;; remove bundled copy of pysam
1819 (snippet
1820 '(delete-file-recursively "lib/pysam"))))
1821 (build-system python-build-system)
1822 (arguments
1823 `(#:python ,python-2
1824 #:phases
dc1d3cde
KK		 1825 (modify-phases %standard-phases
1826 (add-after 'unpack 'set-env
1827 (lambda _ (setenv "CROSSMAP_USE_SYSTEM_PYSAM" "1") #t)))))
191c7101
RW		 1828 (inputs
1829 `(("python-numpy" ,python2-numpy)
1830 ("python-pysam" ,python2-pysam)
1831 ("zlib" ,zlib)))
1832 (native-inputs
1833 `(("python-cython" ,python2-cython)
f3b98f4f 1834 ("python-nose" ,python2-nose)))
191c7101
RW		 1835 (home-page "http://crossmap.sourceforge.net/")
1836 (synopsis "Convert genome coordinates between assemblies")
1837 (description
1838 "CrossMap is a program for conversion of genome coordinates or annotation
1839files between different genome assemblies. It supports most commonly used
1840file formats including SAM/BAM, Wiggle/BigWig, BED, GFF/GTF, VCF.")
1841 (license license:gpl2+)))
1842
8e913213
RW		 1843(define-public cutadapt
1844 (package
1845 (name "cutadapt")
0af4df7e 1846 (version "1.14")
8e913213
RW		 1847 (source (origin
1848 (method url-fetch)
1849 (uri (string-append
1850 "https://github.com/marcelm/cutadapt/archive/v"
1851 version ".tar.gz"))
1852 (file-name (string-append name "-" version ".tar.gz"))
1853 (sha256
1854 (base32
0af4df7e 1855 "16gbpiwy4m48vq2h5wqar3i8vr6vcj9gcl2qvqim19x6ya9dp8kd"))))
8e913213
RW		 1856 (build-system python-build-system)
1857 (arguments
33d5b246
TGR		 1858 `(#:phases
1859 (modify-phases %standard-phases
1860 ;; The tests must be run after installation.
1861 (delete 'check)
1862 (add-after 'install 'check
1863 (lambda* (#:key inputs outputs #:allow-other-keys)
1864 (setenv "PYTHONPATH"
1865 (string-append
1866 (getenv "PYTHONPATH")
1867 ":" (assoc-ref outputs "out")
1868 "/lib/python"
1869 (string-take (string-take-right
1870 (assoc-ref inputs "python") 5) 3)
1871 "/site-packages"))
1872 (zero? (system* "nosetests" "-P" "tests")))))))
1f94bff2
TGR		 1873 (inputs
1874 `(("python-xopen" ,python-xopen)))
8e913213
RW		 1875 (native-inputs
1876 `(("python-cython" ,python-cython)
f3b98f4f 1877 ("python-nose" ,python-nose)))
0c6c9c00 1878 (home-page "https://cutadapt.readthedocs.io/en/stable/")
8e913213
RW		 1879 (synopsis "Remove adapter sequences from nucleotide sequencing reads")
1880 (description
1881 "Cutadapt finds and removes adapter sequences, primers, poly-A tails and
1882other types of unwanted sequence from high-throughput sequencing reads.")
1883 (license license:expat)))
1884
1baee943
RW		 1885(define-public libbigwig
1886 (package
1887 (name "libbigwig")
1888 (version "0.1.4")
1889 (source (origin
1890 (method url-fetch)
1891 (uri (string-append "https://github.com/dpryan79/libBigWig/"
1892 "archive/" version ".tar.gz"))
1893 (file-name (string-append name "-" version ".tar.gz"))
1894 (sha256
1895 (base32
1896 "098rjh35pi4a9q83n8wiwvyzykjqj6l8q189p1xgfw4ghywdlvw1"))))
1897 (build-system gnu-build-system)
1898 (arguments
1899 `(#:test-target "test"
1900 #:make-flags
1901 (list "CC=gcc"
1902 (string-append "prefix=" (assoc-ref %outputs "out")))
1903 #:phases
1904 (modify-phases %standard-phases
1905 (delete 'configure)
1906 (add-before 'check 'disable-curl-test
1907 (lambda _
1908 (substitute* "Makefile"
1909 (("./test/testRemote.*") ""))
1910 #t))
1911 ;; This has been fixed with the upstream commit 4ff6959cd8a0, but
1912 ;; there has not yet been a release containing this change.
1913 (add-before 'install 'create-target-dirs
1914 (lambda* (#:key outputs #:allow-other-keys)
1915 (let ((out (assoc-ref outputs "out")))
1916 (mkdir-p (string-append out "/lib"))
1917 (mkdir-p (string-append out "/include"))
1918 #t))))))
1919 (inputs
1920 `(("zlib" ,zlib)
1921 ("curl" ,curl)))
1922 (native-inputs
1923 `(("doxygen" ,doxygen)))
1924 (home-page "https://github.com/dpryan79/libBigWig")
1925 (synopsis "C library for handling bigWig files")
1926 (description
1927 "This package provides a C library for parsing local and remote BigWig
1928files.")
1929 (license license:expat)))
1930
69e0e03c
RW		 1931(define-public python-pybigwig
1932 (package
1933 (name "python-pybigwig")
1934 (version "0.2.5")
1935 (source (origin
1936 (method url-fetch)
1937 (uri (pypi-uri "pyBigWig" version))
1938 (sha256
1939 (base32
1940 "0yrpdxg3y0sny25x4w22lv1k47jzccqjmg7j4bp0hywklvp0hg7d"))
1941 (modules '((guix build utils)))
1942 (snippet
1943 '(begin
1944 ;; Delete bundled libBigWig sources
1945 (delete-file-recursively "libBigWig")))))
1946 (build-system python-build-system)
1947 (arguments
1948 `(#:phases
1949 (modify-phases %standard-phases
1950 (add-after 'unpack 'link-with-libBigWig
1951 (lambda* (#:key inputs #:allow-other-keys)
1952 (substitute* "setup.py"
1953 (("libs=\\[") "libs=[\"BigWig\", "))
1954 #t)))))
1955 (inputs
1956 `(("libbigwig" ,libbigwig)
1957 ("zlib" ,zlib)
1958 ("curl" ,curl)))
1959 (home-page "https://github.com/dpryan79/pyBigWig")
1960 (synopsis "Access bigWig files in Python using libBigWig")
1961 (description
1962 "This package provides Python bindings to the libBigWig library for
1963accessing bigWig files.")
1964 (license license:expat)))
1965
1966(define-public python2-pybigwig
5c31f4aa 1967 (package-with-python2 python-pybigwig))
69e0e03c 1968
ec2a67de
BW		 1969(define-public python-dendropy
1970 (package
1971 (name "python-dendropy")
25d84d31 1972 (version "4.2.0")
ec2a67de
BW		 1973 (source
1974 (origin
1975 (method url-fetch)
1976 (uri (pypi-uri "DendroPy" version))
1977 (sha256
1978 (base32
1885bb0c
RW		 1979 "15c7s3d5gf19ljsxvq5advaa752wfi7pwrdjyhzmg85hccyvp47p"))
1980 (patches (search-patches "python-dendropy-fix-tests.patch"))))
ec2a67de
BW		 1981 (build-system python-build-system)
1982 (home-page "http://packages.python.org/DendroPy/")
1983 (synopsis "Library for phylogenetics and phylogenetic computing")
1984 (description
1985 "DendroPy is a library for phylogenetics and phylogenetic computing: reading,
1986writing, simulation, processing and manipulation of phylogenetic
1987trees (phylogenies) and characters.")
1988 (license license:bsd-3)
1989 (properties `((python2-variant . ,(delay python2-dendropy))))))
1990
1991(define-public python2-dendropy
1992 (let ((base (package-with-python2 (strip-python2-variant python-dendropy))))
1993 (package
1994 (inherit base)
9602e3cc
BW		 1995 (arguments
1996 `(#:python ,python-2
1997 #:phases
1998 (modify-phases %standard-phases
1999 (replace 'check
2000 ;; There is currently a test failure that only happens on some
2001 ;; systems, and only using "setup.py test"
2002 (lambda _ (zero? (system* "nosetests")))))))
f3b98f4f 2003 (native-inputs `(("python2-nose" ,python2-nose)
ec2a67de
BW		 2004 ,@(package-native-inputs base))))))
2005
eb2200f3
RW		 2006(define-public python-py2bit
2007 (package
2008 (name "python-py2bit")
2009 (version "0.2.1")
2010 (source
2011 (origin
2012 (method url-fetch)
2013 (uri (pypi-uri "py2bit" version))
2014 (sha256
2015 (base32
2016 "1cdf4qlmgwsh1f4k0wdv2sr8x9qn4366p0k3614vbd0fpqiarxrl"))))
2017 (build-system python-build-system)
2018 (home-page "https://github.com/dpryan79/py2bit")
2019 (synopsis "Access 2bit files using lib2bit")
2020 (description
2021 "This package provides Python bindings for lib2bit to access 2bit files
2022with Python.")
2023 (license license:expat)))
ec2a67de 2024
1921b1de
RW		 2025(define-public deeptools
2026 (package
2027 (name "deeptools")
fed72008 2028 (version "2.5.1")
1921b1de
RW		 2029 (source (origin
2030 (method url-fetch)
3acb8c85
RW		 2031 (uri (string-append "https://github.com/fidelram/deepTools/"
2032 "archive/" version ".tar.gz"))
1921b1de
RW		 2033 (file-name (string-append name "-" version ".tar.gz"))
2034 (sha256
2035 (base32
fed72008 2036 "1q8i12l2gvk4n2s8lhyzwhh9g4qbc8lrk5l7maz00yvd5g6z5540"))))
1921b1de 2037 (build-system python-build-system)
14bda1ff 2038 (inputs
fed72008
RW		 2039 `(("python-scipy" ,python-scipy)
2040 ("python-numpy" ,python-numpy)
2041 ("python-numpydoc" ,python-numpydoc)
2042 ("python-matplotlib" ,python-matplotlib)
2043 ("python-pysam" ,python-pysam)
2044 ("python-py2bit" ,python-py2bit)
2045 ("python-pybigwig" ,python-pybigwig)))
1921b1de 2046 (native-inputs
fed72008
RW		 2047 `(("python-mock" ,python-mock) ;for tests
2048 ("python-nose" ,python-nose) ;for tests
2049 ("python-pytz" ,python-pytz))) ;for tests
1921b1de
RW		 2050 (home-page "https://github.com/fidelram/deepTools")
2051 (synopsis "Tools for normalizing and visualizing deep-sequencing data")
2052 (description
2053 "DeepTools addresses the challenge of handling the large amounts of data
2054that are now routinely generated from DNA sequencing centers. To do so,
2055deepTools contains useful modules to process the mapped reads data to create
2056coverage files in standard bedGraph and bigWig file formats. By doing so,
2057deepTools allows the creation of normalized coverage files or the comparison
2058between two files (for example, treatment and control). Finally, using such
2059normalized and standardized files, multiple visualizations can be created to
2060identify enrichments with functional annotations of the genome.")
2061 (license license:gpl3+)))
2062
684bf7c7
BW		 2063(define-public diamond
2064 (package
2065 (name "diamond")
a68e5860 2066 (version "0.9.13")
684bf7c7
BW		 2067 (source (origin
2068 (method url-fetch)
2069 (uri (string-append
2070 "https://github.com/bbuchfink/diamond/archive/v"
2071 version ".tar.gz"))
2072 (file-name (string-append name "-" version ".tar.gz"))
2073 (sha256
2074 (base32
a68e5860 2075 "1pi5ncqwmynqpmmp3j3lhnqrjhj34sr6wpmsgrpkv3wyxx22fv86"))))
122395f9 2076 (build-system cmake-build-system)
684bf7c7 2077 (arguments
7c544991
BW		 2078 '(#:tests? #f ; no "check" target
2079 #:phases
2080 (modify-phases %standard-phases
2081 (add-after 'unpack 'remove-native-compilation
2082 (lambda _
2083 (substitute* "CMakeLists.txt" (("-march=native") ""))
2084 #t)))))
684bf7c7 2085 (inputs
122395f9 2086 `(("zlib" ,zlib)))
684bf7c7
BW		 2087 (home-page "https://github.com/bbuchfink/diamond")
2088 (synopsis "Accelerated BLAST compatible local sequence aligner")
2089 (description
2090 "DIAMOND is a BLAST-compatible local aligner for mapping protein and
2091translated DNA query sequences against a protein reference database (BLASTP
2092and BLASTX alignment mode). The speedup over BLAST is up to 20,000 on short
2093reads at a typical sensitivity of 90-99% relative to BLAST depending on the
2094data and settings.")
ef81341f 2095 (license license:agpl3+)))
684bf7c7 2096
97b9da68
RW		 2097(define-public discrover
2098 (package
2099 (name "discrover")
2100 (version "1.6.0")
2101 (source
2102 (origin
2103 (method url-fetch)
2104 (uri (string-append "https://github.com/maaskola/discrover/archive/"
2105 version ".tar.gz"))
2106 (file-name (string-append name "-" version ".tar.gz"))
2107 (sha256
2108 (base32
2109 "0rah9ja4m0rl5mldd6vag9rwrivw1zrqxssfq8qx64m7961fp68k"))))
2110 (build-system cmake-build-system)
fa702e1a
RW		 2111 (arguments
2112 `(#:tests? #f ; there are no tests
2113 #:phases
2114 (modify-phases %standard-phases
2115 (add-after 'unpack 'add-missing-includes
2116 (lambda _
2117 (substitute* "src/executioninformation.hpp"
2118 (("#define EXECUTIONINFORMATION_HPP" line)
2119 (string-append line "\n#include <random>")))
2120 (substitute* "src/plasma/fasta.hpp"
2121 (("#define FASTA_HPP" line)
2122 (string-append line "\n#include <random>")))
2123 #t)))))
97b9da68
RW		 2124 (inputs
2125 `(("boost" ,boost)
2126 ("cairo" ,cairo)))
2127 (native-inputs
2128 `(("texlive" ,texlive)
2129 ("imagemagick" ,imagemagick)))
2130 (home-page "http://dorina.mdc-berlin.de/public/rajewsky/discrover/")
2131 (synopsis "Discover discriminative nucleotide sequence motifs")
2132 (description "Discrover is a motif discovery method to find binding sites
2133of nucleic acid binding proteins.")
2134 (license license:gpl3+)))
2135
6619f9c7
RW		 2136(define-public eigensoft
2137 (let ((revision "1")
2138 (commit "b14d1e202e21e532536ff8004f0419cd5e259dc7"))
2139 (package
2140 (name "eigensoft")
2141 (version (string-append "6.1.2-"
2142 revision "."
2143 (string-take commit 9)))
2144 (source
2145 (origin
2146 (method git-fetch)
2147 (uri (git-reference
2148 (url "https://github.com/DReichLab/EIG.git")
2149 (commit commit)))
2150 (file-name (string-append "eigensoft-" commit "-checkout"))
2151 (sha256
2152 (base32
2153 "0f5m6k2j5c16xc3xbywcs989xyc26ncy1zfzp9j9n55n9r4xcaiq"))
2154 (modules '((guix build utils)))
2155 ;; Remove pre-built binaries.
2156 (snippet '(begin
2157 (delete-file-recursively "bin")
2158 (mkdir "bin")
2159 #t))))
2160 (build-system gnu-build-system)
2161 (arguments
2162 `(#:tests? #f ; There are no tests.
2163 #:make-flags '("CC=gcc")
2164 #:phases
2165 (modify-phases %standard-phases
2166 ;; There is no configure phase, but the Makefile is in a
2167 ;; sub-directory.
2168 (replace 'configure
2169 (lambda _
2170 (chdir "src")
2171 ;; The link flags are incomplete.
2172 (substitute* "Makefile"
2173 (("-lgsl") "-lgsl -lm -llapack -llapacke -lpthread"))
2174 #t))
2175 ;; The provided install target only copies executables to
2176 ;; the "bin" directory in the build root.
2177 (add-after 'install 'actually-install
2178 (lambda* (#:key outputs #:allow-other-keys)
2179 (let* ((out (assoc-ref outputs "out"))
2180 (bin (string-append out "/bin")))
6619f9c7
RW		 2181 (for-each (lambda (file)
2182 (install-file file bin))
2183 (find-files "../bin" ".*"))
2184 #t))))))
2185 (inputs
2186 `(("gsl" ,gsl)
2187 ("lapack" ,lapack)
6619f9c7
RW		 2188 ("openblas" ,openblas)
2189 ("perl" ,perl)
2190 ("gfortran" ,gfortran "lib")))
2191 (home-page "https://github.com/DReichLab/EIG")
2192 (synopsis "Tools for population genetics")
2193 (description "The EIGENSOFT package provides tools for population
2194genetics and stratification correction. EIGENSOFT implements methods commonly
2195used in population genetics analyses such as PCA, computation of Tracy-Widom
2196statistics, and finding related individuals in structured populations. It
2197comes with a built-in plotting script and supports multiple file formats and
2198quantitative phenotypes.")
2199 ;; The license of the eigensoft tools is Expat, but since it's
2200 ;; linking with the GNU Scientific Library (GSL) the effective
2201 ;; license is the GPL.
2202 (license license:gpl3+))))
2203
365c8153
RW		 2204(define-public edirect
2205 (package
2206 (name "edirect")
83b84fa8 2207 (version "4.10")
365c8153
RW		 2208 (source (origin
2209 (method url-fetch)
83b84fa8
RW		 2210 (uri (string-append "ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/"
2211 "versions/2016-05-03/edirect.tar.gz"))
365c8153
RW		 2212 (sha256
2213 (base32
83b84fa8 2214 "15zsprak5yh8c1yrz4r1knmb5s8qcmdid4xdhkh3lqcv64l60hli"))))
365c8153
RW		 2215 (build-system perl-build-system)
2216 (arguments
2217 `(#:tests? #f ;no "check" target
2218 #:phases
2219 (modify-phases %standard-phases
2220 (delete 'configure)
2221 (delete 'build)
2222 (replace 'install
2223 (lambda* (#:key outputs #:allow-other-keys)
2224 (let ((target (string-append (assoc-ref outputs "out")
2225 "/bin")))
2226 (mkdir-p target)
f3860753 2227 (install-file "edirect.pl" target)
365c8153
RW		 2228 #t)))
2229 (add-after
2230 'install 'wrap-program
2231 (lambda* (#:key inputs outputs #:allow-other-keys)
2232 ;; Make sure 'edirect.pl' finds all perl inputs at runtime.
2233 (let* ((out (assoc-ref outputs "out"))
2234 (path (getenv "PERL5LIB")))
2235 (wrap-program (string-append out "/bin/edirect.pl")
2236 `("PERL5LIB" ":" prefix (,path)))))))))
2237 (inputs
2238 `(("perl-html-parser" ,perl-html-parser)
2239 ("perl-encode-locale" ,perl-encode-locale)
2240 ("perl-file-listing" ,perl-file-listing)
2241 ("perl-html-tagset" ,perl-html-tagset)
2242 ("perl-html-tree" ,perl-html-tree)
2243 ("perl-http-cookies" ,perl-http-cookies)
2244 ("perl-http-date" ,perl-http-date)
2245 ("perl-http-message" ,perl-http-message)
2246 ("perl-http-negotiate" ,perl-http-negotiate)
2247 ("perl-lwp-mediatypes" ,perl-lwp-mediatypes)
2248 ("perl-lwp-protocol-https" ,perl-lwp-protocol-https)
2249 ("perl-net-http" ,perl-net-http)
2250 ("perl-uri" ,perl-uri)
2251 ("perl-www-robotrules" ,perl-www-robotrules)
2252 ("perl" ,perl)))
3d51ec91 2253 (home-page "http://www.ncbi.nlm.nih.gov/books/NBK179288/")
365c8153
RW		 2254 (synopsis "Tools for accessing the NCBI's set of databases")
2255 (description
2256 "Entrez Direct (EDirect) is a method for accessing the National Center
2257for Biotechnology Information's (NCBI) set of interconnected
2258databases (publication, sequence, structure, gene, variation, expression,
2259etc.) from a terminal. Functions take search terms from command-line
2260arguments. Individual operations are combined to build multi-step queries.
2261Record retrieval and formatting normally complete the process.
2262
2263EDirect also provides an argument-driven function that simplifies the
2264extraction of data from document summaries or other results that are returned
2265in structured XML format. This can eliminate the need for writing custom
2266software to answer ad hoc questions.")
2267 (license license:public-domain)))
2268
b16728b0
BW		 2269(define-public exonerate
2270 (package
2271 (name "exonerate")
2272 (version "2.4.0")
2273 (source
2274 (origin
2275 (method url-fetch)
2276 (uri
2277 (string-append
2278 "http://ftp.ebi.ac.uk/pub/software/vertebrategenomics/exonerate/"
2279 "exonerate-" version ".tar.gz"))
2280 (sha256
2281 (base32
2282 "0hj0m9xygiqsdxvbg79wq579kbrx1mdrabi2bzqz2zn9qwfjcjgq"))))
2283 (build-system gnu-build-system)
2284 (arguments
2285 `(#:parallel-build? #f)) ; Building in parallel fails on some machines.
2286 (native-inputs
2287 `(("pkg-config" ,pkg-config)))
2288 (inputs
2289 `(("glib" ,glib)))
2290 (home-page
2291 "https://www.ebi.ac.uk/about/vertebrate-genomics/software/exonerate")
2292 (synopsis "Generic tool for biological sequence alignment")
2293 (description
2294 "Exonerate is a generic tool for pairwise sequence comparison. It allows
2295the alignment of sequences using a many alignment models, either exhaustive
2296dynamic programming or a variety of heuristics.")
2297 (license license:gpl3)))
2298
e4e5a4d8
RW		 2299(define-public express
2300 (package
2301 (name "express")
2302 (version "1.5.1")
2303 (source (origin
2304 (method url-fetch)
2305 (uri
2306 (string-append
2307 "http://bio.math.berkeley.edu/eXpress/downloads/express-"
2308 version "/express-" version "-src.tgz"))
2309 (sha256
2310 (base32
2311 "03rczxd0gjp2l1jxcmjfmf5j94j77zqyxa6x063zsc585nj40n0c"))))
2312 (build-system cmake-build-system)
2313 (arguments
2314 `(#:tests? #f ;no "check" target
2315 #:phases
dc1d3cde
KK		 2316 (modify-phases %standard-phases
2317 (add-after 'unpack 'use-shared-boost-libs-and-set-bamtools-paths
2318 (lambda* (#:key inputs #:allow-other-keys)
2319 (substitute* "CMakeLists.txt"
2320 (("set\\(Boost_USE_STATIC_LIBS ON\\)")
2321 "set(Boost_USE_STATIC_LIBS OFF)")
2322 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/bamtools/include")
2323 (string-append (assoc-ref inputs "bamtools") "/include/bamtools")))
2324 (substitute* "src/CMakeLists.txt"
2325 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/\\.\\./bamtools/lib")
2326 (string-append (assoc-ref inputs "bamtools") "/lib/bamtools")))
2327 #t)))))
e4e5a4d8
RW		 2328 (inputs
2329 `(("boost" ,boost)
2330 ("bamtools" ,bamtools)
2331 ("protobuf" ,protobuf)
2332 ("zlib" ,zlib)))
2333 (home-page "http://bio.math.berkeley.edu/eXpress")
2334 (synopsis "Streaming quantification for high-throughput genomic sequencing")
2335 (description
2336 "eXpress is a streaming tool for quantifying the abundances of a set of
2337target sequences from sampled subsequences. Example applications include
2338transcript-level RNA-Seq quantification, allele-specific/haplotype expression
2339analysis (from RNA-Seq), transcription factor binding quantification in
2340ChIP-Seq, and analysis of metagenomic data.")
2341 (license license:artistic2.0)))
2342
f3674b1c
BW		 2343(define-public express-beta-diversity
2344 (package
2345 (name "express-beta-diversity")
2346 (version "1.0.7")
2347 (source (origin
2348 (method url-fetch)
2349 (uri
2350 (string-append
2351 "https://github.com/dparks1134/ExpressBetaDiversity/archive/v"
2352 version ".tar.gz"))
2353 (file-name (string-append name "-" version ".tar.gz"))
2354 (sha256
2355 (base32
2356 "1djvdlmqvjf6h0zq7w36y8cl5cli6rgj86x65znl48agnwmzxfxr"))))
2357 (build-system gnu-build-system)
2358 (arguments
2359 `(#:phases
2360 (modify-phases %standard-phases
2361 (delete 'configure)
2362 (add-before 'build 'enter-source (lambda _ (chdir "source") #t))
2363 (replace 'check
2364 (lambda _ (zero? (system* "../bin/ExpressBetaDiversity"
2365 "-u"))))
2366 (add-after 'check 'exit-source (lambda _ (chdir "..") #t))
2367 (replace 'install
2368 (lambda* (#:key outputs #:allow-other-keys)
2369 (let ((bin (string-append (assoc-ref outputs "out")
2370 "/bin")))
2371 (mkdir-p bin)
f3860753
TGR		 2372 (install-file "scripts/convertToEBD.py" bin)
2373 (install-file "bin/ExpressBetaDiversity" bin)
f3674b1c
BW		 2374 #t))))))
2375 (inputs
2376 `(("python" ,python-2)))
2377 (home-page "http://kiwi.cs.dal.ca/Software/ExpressBetaDiversity")
2378 (synopsis "Taxon- and phylogenetic-based beta diversity measures")
2379 (description
2380 "Express Beta Diversity (EBD) calculates ecological beta diversity
2381(dissimilarity) measures between biological communities. EBD implements a
2382variety of diversity measures including those that make use of phylogenetic
2383similarity of community members.")
2384 (license license:gpl3+)))
2385
12b04cbe
BW		 2386(define-public fasttree
2387 (package
2388 (name "fasttree")
88682c9a 2389 (version "2.1.10")
12b04cbe
BW		 2390 (source (origin
2391 (method url-fetch)
2392 (uri (string-append
2393 "http://www.microbesonline.org/fasttree/FastTree-"
2394 version ".c"))
2395 (sha256
2396 (base32
88682c9a 2397 "0vcjdvy1j4m702vmak4svbfkrpcw63k7wymfksjp9a982zy8kjsl"))))
12b04cbe
BW		 2398 (build-system gnu-build-system)
2399 (arguments
2400 `(#:tests? #f ; no "check" target
2401 #:phases
2402 (modify-phases %standard-phases
2403 (delete 'unpack)
2404 (delete 'configure)
2405 (replace 'build
e03a5153
BW		 2406 (lambda* (#:key source #:allow-other-keys)
2407 (and (zero? (system* "gcc"
2408 "-O3"
2409 "-finline-functions"
2410 "-funroll-loops"
2411 "-Wall"
2412 "-o"
2413 "FastTree"
2414 source
2415 "-lm"))
2416 (zero? (system* "gcc"
2417 "-DOPENMP"
2418 "-fopenmp"
2419 "-O3"
2420 "-finline-functions"
2421 "-funroll-loops"
2422 "-Wall"
2423 "-o"
2424 "FastTreeMP"
2425 source
2426 "-lm")))))
12b04cbe 2427 (replace 'install
e03a5153
BW		 2428 (lambda* (#:key outputs #:allow-other-keys)
2429 (let ((bin (string-append (assoc-ref outputs "out")
2430 "/bin")))
2431 (mkdir-p bin)
f3860753
TGR		 2432 (install-file "FastTree" bin)
2433 (install-file "FastTreeMP" bin)
e03a5153 2434 #t))))))
12b04cbe
BW		 2435 (home-page "http://www.microbesonline.org/fasttree")
2436 (synopsis "Infers approximately-maximum-likelihood phylogenetic trees")
2437 (description
2438 "FastTree can handle alignments with up to a million of sequences in a
2439reasonable amount of time and memory. For large alignments, FastTree is
2440100-1,000 times faster than PhyML 3.0 or RAxML 7.")
2441 (license license:gpl2+)))
2442
2127cedb
RW		 2443(define-public fastx-toolkit
2444 (package
2445 (name "fastx-toolkit")
2446 (version "0.0.14")
2447 (source (origin
2448 (method url-fetch)
2449 (uri
2450 (string-append
2451 "https://github.com/agordon/fastx_toolkit/releases/download/"
2452 version "/fastx_toolkit-" version ".tar.bz2"))
2453 (sha256
2454 (base32
2455 "01jqzw386873sr0pjp1wr4rn8fsga2vxs1qfmicvx1pjr72007wy"))))
2456 (build-system gnu-build-system)
2457 (inputs
2458 `(("libgtextutils" ,libgtextutils)))
2459 (native-inputs
2460 `(("pkg-config" ,pkg-config)))
2461 (home-page "http://hannonlab.cshl.edu/fastx_toolkit/")
2462 (synopsis "Tools for FASTA/FASTQ file preprocessing")
2463 (description
2464 "The FASTX-Toolkit is a collection of command line tools for Short-Reads
2465FASTA/FASTQ files preprocessing.
2466
2467Next-Generation sequencing machines usually produce FASTA or FASTQ files,
2468containing multiple short-reads sequences. The main processing of such
2469FASTA/FASTQ files is mapping the sequences to reference genomes. However, it
2470is sometimes more productive to preprocess the files before mapping the
2471sequences to the genome---manipulating the sequences to produce better mapping
2472results. The FASTX-Toolkit tools perform some of these preprocessing tasks.")
2473 (license license:agpl3+)))
2474
d7678942
RW		 2475(define-public flexbar
2476 (package
2477 (name "flexbar")
2478 (version "2.5")
2479 (source (origin
2480 (method url-fetch)
2481 (uri
2482 (string-append "mirror://sourceforge/flexbar/"
2483 version "/flexbar_v" version "_src.tgz"))
2484 (sha256
2485 (base32
2486 "13jaykc3y1x8y5nn9j8ljnb79s5y51kyxz46hdmvvjj6qhyympmf"))))
2487 (build-system cmake-build-system)
2488 (arguments
4ca009c0 2489 `(#:configure-flags (list
d7678942
RW		 2490 (string-append "-DFLEXBAR_BINARY_DIR="
2491 (assoc-ref %outputs "out")
2492 "/bin/"))
2493 #:phases
dc1d3cde
KK		 2494 (modify-phases %standard-phases
2495 (replace 'check
2496 (lambda* (#:key outputs #:allow-other-keys)
2497 (setenv "PATH" (string-append
2498 (assoc-ref outputs "out") "/bin:"
2499 (getenv "PATH")))
2500 (chdir "../flexbar_v2.5_src/test")
2501 (zero? (system* "bash" "flexbar_validate.sh"))))
2502 (delete 'install))))
d7678942
RW		 2503 (inputs
2504 `(("tbb" ,tbb)
2505 ("zlib" ,zlib)))
2506 (native-inputs
2507 `(("pkg-config" ,pkg-config)
2508 ("seqan" ,seqan)))
2509 (home-page "http://flexbar.sourceforge.net")
2510 (synopsis "Barcode and adapter removal tool for sequencing platforms")
2511 (description
2512 "Flexbar preprocesses high-throughput nucleotide sequencing data
2513efficiently. It demultiplexes barcoded runs and removes adapter sequences.
2514Moreover, trimming and filtering features are provided. Flexbar increases
2515read mapping rates and improves genome and transcriptome assemblies. It
2516supports next-generation sequencing data in fasta/q and csfasta/q format from
2517Illumina, Roche 454, and the SOLiD platform.")
2518 (license license:gpl3)))
2519
19f4554c
BW		 2520(define-public fraggenescan
2521 (package
2522 (name "fraggenescan")
74297231 2523 (version "1.30")
19f4554c
BW		 2524 (source
2525 (origin
2526 (method url-fetch)
2527 (uri
2528 (string-append "mirror://sourceforge/fraggenescan/"
2529 "FragGeneScan" version ".tar.gz"))
2530 (sha256
74297231 2531 (base32 "158dcnwczgcyhwm4qlx19sanrwgdpzf6bn2y57mbpx55lkgz1mzj"))))
19f4554c
BW		 2532 (build-system gnu-build-system)
2533 (arguments
2534 `(#:phases
2535 (modify-phases %standard-phases
2536 (delete 'configure)
2537 (add-before 'build 'patch-paths
2538 (lambda* (#:key outputs #:allow-other-keys)
2539 (let* ((out (string-append (assoc-ref outputs "out")))
2540 (share (string-append out "/share/fraggenescan/")))
2541 (substitute* "run_FragGeneScan.pl"
2542 (("system\\(\"rm")
2543 (string-append "system(\"" (which "rm")))
2544 (("system\\(\"mv")
2545 (string-append "system(\"" (which "mv")))
74297231 2546 (("\\\"awk") (string-append "\"" (which "awk")))
19f4554c
BW		 2547 ;; This script and other programs expect the training files
2548 ;; to be in the non-standard location bin/train/XXX. Change
2549 ;; this to be share/fraggenescan/train/XXX instead.
2550 (("^\\$train.file = \\$dir.*")
2551 (string-append "$train_file = \""
2552 share
2553 "train/\".$FGS_train_file;")))
2554 (substitute* "run_hmm.c"
2555 (("^ strcat\\(train_dir, \\\"train/\\\"\\);")
74297231 2556 (string-append " strcpy(train_dir, \"" share "/train/\");"))))
19f4554c
BW		 2557 #t))
2558 (replace 'build
2559 (lambda _ (and (zero? (system* "make" "clean"))
2560 (zero? (system* "make" "fgs")))))
2561 (replace 'install
2562 (lambda* (#:key outputs #:allow-other-keys)
2563 (let* ((out (string-append (assoc-ref outputs "out")))
2564 (bin (string-append out "/bin/"))
2565 (share (string-append out "/share/fraggenescan/train")))
2566 (install-file "run_FragGeneScan.pl" bin)
2567 (install-file "FragGeneScan" bin)
19f4554c
BW		 2568 (copy-recursively "train" share))))
2569 (delete 'check)
2570 (add-after 'install 'post-install-check
2571 ;; In lieu of 'make check', run one of the examples and check the
2572 ;; output files gets created.
2573 (lambda* (#:key outputs #:allow-other-keys)
2574 (let* ((out (string-append (assoc-ref outputs "out")))
74297231
BW		 2575 (bin (string-append out "/bin/"))
2576 (frag (string-append bin "run_FragGeneScan.pl")))
2577 (and (zero? (system* frag ; Test complete genome.
19f4554c
BW		 2578 "-genome=./example/NC_000913.fna"
2579 "-out=./test2"
2580 "-complete=1"
2581 "-train=complete"))
2582 (file-exists? "test2.faa")
2583 (file-exists? "test2.ffn")
2584 (file-exists? "test2.gff")
74297231
BW		 2585 (file-exists? "test2.out")
2586 (zero? (system* ; Test incomplete sequences.
2587 frag
2588 "-genome=./example/NC_000913-fgs.ffn"
2589 "-out=out"
2590 "-complete=0"
2591 "-train=454_30")))))))))
19f4554c
BW		 2592 (inputs
2593 `(("perl" ,perl)
2594 ("python" ,python-2))) ;not compatible with python 3.
2595 (home-page "https://sourceforge.net/projects/fraggenescan/")
2596 (synopsis "Finds potentially fragmented genes in short reads")
2597 (description
2598 "FragGeneScan is a program for predicting bacterial and archaeal genes in
2599short and error-prone DNA sequencing reads. It can also be applied to predict
2600genes in incomplete assemblies or complete genomes.")
2601 ;; GPL3+ according to private correspondense with the authors.
2602 (license license:gpl3+)))
2603
81f3e0c1
BW		 2604(define-public fxtract
2605 (let ((util-commit "776ca85a18a47492af3794745efcb4a905113115"))
2606 (package
2607 (name "fxtract")
2608 (version "2.3")
2609 (source
2610 (origin
2611 (method url-fetch)
2612 (uri (string-append
2613 "https://github.com/ctSkennerton/fxtract/archive/"
2614 version ".tar.gz"))
2615 (file-name (string-append "ctstennerton-util-"
2616 (string-take util-commit 7)
2617 "-checkout"))
2618 (sha256
2619 (base32
2620 "0275cfdhis8517hm01is62062swmi06fxzifq7mr3knbbxjlaiwj"))))
2621 (build-system gnu-build-system)
2622 (arguments
2623 `(#:make-flags (list
2624 (string-append "PREFIX=" (assoc-ref %outputs "out"))
2625 "CC=gcc")
2626 #:test-target "fxtract_test"
2627 #:phases
2628 (modify-phases %standard-phases
2629 (delete 'configure)
2630 (add-before 'build 'copy-util
2631 (lambda* (#:key inputs #:allow-other-keys)
2632 (rmdir "util")
2633 (copy-recursively (assoc-ref inputs "ctskennerton-util") "util")
2634 #t))
2635 ;; Do not use make install as this requires additional dependencies.
2636 (replace 'install
2637 (lambda* (#:key outputs #:allow-other-keys)
2638 (let* ((out (assoc-ref outputs "out"))
2639 (bin (string-append out"/bin")))
2640 (install-file "fxtract" bin)
2641 #t))))))
2642 (inputs
2643 `(("pcre" ,pcre)
2644 ("zlib" ,zlib)))
2645 (native-inputs
2646 ;; ctskennerton-util is licensed under GPL2.
2647 `(("ctskennerton-util"
2648 ,(origin
2649 (method git-fetch)
2650 (uri (git-reference
2651 (url "https://github.com/ctSkennerton/util.git")
2652 (commit util-commit)))
2653 (file-name (string-append
2654 "ctstennerton-util-" util-commit "-checkout"))
2655 (sha256
2656 (base32
2657 "0cls1hd4vgj3f36fpzzg4xc77d6f3hpc60cbpfmn2gdr7ykzzad7"))))))
2658 (home-page "https://github.com/ctSkennerton/fxtract")
2659 (synopsis "Extract sequences from FASTA and FASTQ files")
2660 (description
2661 "Fxtract extracts sequences from a protein or nucleotide fastx (FASTA
2662or FASTQ) file given a subsequence. It uses a simple substring search for
2663basic tasks but can change to using POSIX regular expressions, PCRE, hash
2664lookups or multi-pattern searching as required. By default fxtract looks in
2665the sequence of each record but can also be told to look in the header,
2666comment or quality sections.")
afde1a26
BW		 2667 ;; 'util' requires SSE instructions.
2668 (supported-systems '("x86_64-linux"))
81f3e0c1
BW		 2669 (license license:expat))))
2670
2b18ad05
PP		 2671(define-public gemma
2672 (package
2673 (name "gemma")
2674 (version "0.96")
2675 (source (origin
2676 (method url-fetch)
2677 (uri (string-append "https://github.com/xiangzhou/GEMMA/archive/v"
2678 version ".tar.gz"))
2679 (file-name (string-append name "-" version ".tar.gz"))
2680 (sha256
2681 (base32
ce7e361f
EF		 2682 "055ynn16gd12pf78n4vr2a9jlwsbwzajpdnf2y2yilg1krfff222"))
2683 (patches (search-patches "gemma-intel-compat.patch"))))
2b18ad05
PP		 2684 (inputs
2685 `(("gsl" ,gsl)
2686 ("lapack" ,lapack)
2687 ("zlib" ,zlib)))
2688 (build-system gnu-build-system)
2689 (arguments
2c9232ae 2690 `(#:make-flags
ce7e361f
EF		 2691 '(,@(match (%current-system)
2692 ("x86_64-linux"
2693 '("FORCE_DYNAMIC=1"))
2694 ("i686-linux"
2695 '("FORCE_DYNAMIC=1" "FORCE_32BIT=1"))
2696 (_
2697 '("FORCE_DYNAMIC=1" "NO_INTEL_COMPAT=1"))))
2b18ad05
PP		 2698 #:phases
2699 (modify-phases %standard-phases
2700 (delete 'configure)
2701 (add-before 'build 'bin-mkdir
07bf6929
EF		 2702 (lambda _
2703 (mkdir-p "bin")
2704 #t))
2b18ad05 2705 (replace 'install
07bf6929
EF		 2706 (lambda* (#:key outputs #:allow-other-keys)
2707 (let ((out (assoc-ref outputs "out")))
2708 (install-file "bin/gemma"
2709 (string-append
2710 out "/bin")))
2711 #t)))
2b18ad05
PP		 2712 #:tests? #f)) ; no tests included yet
2713 (home-page "https://github.com/xiangzhou/GEMMA")
2714 (synopsis "Tool for genome-wide efficient mixed model association")
2715 (description
2716 "Genome-wide Efficient Mixed Model Association (GEMMA) provides a
2717standard linear mixed model resolver with application in genome-wide
2718association studies (GWAS).")
2719 (license license:gpl3)))
2720
5854f685
RW		 2721(define-public grit
2722 (package
2723 (name "grit")
2724 (version "2.0.2")
2725 (source (origin
2726 (method url-fetch)
2727 (uri (string-append
2728 "https://github.com/nboley/grit/archive/"
2729 version ".tar.gz"))
2730 (file-name (string-append name "-" version ".tar.gz"))
2731 (sha256
2732 (base32
2733 "157in84dj70wimbind3x7sy1whs3h57qfgcnj2s6lrd38fbrb7mj"))))
2734 (build-system python-build-system)
2735 (arguments
2736 `(#:python ,python-2
2737 #:phases
dc1d3cde
KK		 2738 (modify-phases %standard-phases
2739 (add-after 'unpack 'generate-from-cython-sources
2740 (lambda* (#:key inputs outputs #:allow-other-keys)
2741 ;; Delete these C files to force fresh generation from pyx sources.
2742 (delete-file "grit/sparsify_support_fns.c")
2743 (delete-file "grit/call_peaks_support_fns.c")
2744 (substitute* "setup.py"
2745 (("Cython.Setup") "Cython.Build")
2746 ;; Add numpy include path to fix compilation
2747 (("pyx\", \\]")
2748 (string-append "pyx\", ], include_dirs = ['"
2749 (assoc-ref inputs "python-numpy")
2750 "/lib/python2.7/site-packages/numpy/core/include/"
2751 "']")))
2752 #t)))))
5854f685
RW		 2753 (inputs
2754 `(("python-scipy" ,python2-scipy)
2755 ("python-numpy" ,python2-numpy)
2756 ("python-pysam" ,python2-pysam)
2757 ("python-networkx" ,python2-networkx)))
2758 (native-inputs
f3b98f4f 2759 `(("python-cython" ,python2-cython)))
5854f685
RW		 2760 (home-page "http://grit-bio.org")
2761 (synopsis "Tool for integrative analysis of RNA-seq type assays")
2762 (description
2763 "GRIT is designed to use RNA-seq, TES, and TSS data to build and quantify
2764full length transcript models. When none of these data sources are available,
2765GRIT can be run by providing a candidate set of TES or TSS sites. In
2766addition, GRIT can merge in reference junctions and gene boundaries. GRIT can
2767also be run in quantification mode, where it uses a provided GTF file and just
2768estimates transcript expression.")
2769 (license license:gpl3+)))
2770
346a829a
RW		 2771(define-public hisat
2772 (package
2773 (name "hisat")
2774 (version "0.1.4")
2775 (source (origin
2776 (method url-fetch)
2777 (uri (string-append
2778 "http://ccb.jhu.edu/software/hisat/downloads/hisat-"
2779 version "-beta-source.zip"))
2780 (sha256
2781 (base32
2782 "1k381ydranqxp09yf2y7w1d0chz5d59vb6jchi89hbb0prq19lk5"))))
2783 (build-system gnu-build-system)
2784 (arguments
e58d01fa
RW		 2785 `(#:tests? #f ;no check target
2786 #:make-flags '("allall"
2787 ;; Disable unsupported `popcnt' instructions on
2788 ;; architectures other than x86_64
2789 ,@(if (string-prefix? "x86_64"
2790 (or (%current-target-system)
2791 (%current-system)))
2792 '()
2793 '("POPCNT_CAPABILITY=0")))
346a829a 2794 #:phases
dc1d3cde
KK		 2795 (modify-phases %standard-phases
2796 (add-after 'unpack 'patch-sources
2797 (lambda _
2798 ;; XXX Cannot use snippet because zip files are not supported
2799 (substitute* "Makefile"
2800 (("^CC = .*$") "CC = gcc")
2801 (("^CPP = .*$") "CPP = g++")
2802 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
2803 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
2804 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\""))
2805 (substitute* '("hisat-build" "hisat-inspect")
2806 (("/usr/bin/env") (which "env")))
2807 #t))
2808 (replace 'install
2809 (lambda* (#:key outputs #:allow-other-keys)
2810 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
2811 (for-each (lambda (file)
2812 (install-file file bin))
2813 (find-files
2814 "."
2815 "hisat(-(build|align|inspect)(-(s|l)(-debug)*)*)*$")))
2816 #t))
2817 (delete 'configure))))
346a829a
RW		 2818 (native-inputs
2819 `(("unzip" ,unzip)))
2820 (inputs
2821 `(("perl" ,perl)
2822 ("python" ,python)
2823 ("zlib" ,zlib)))
60af3d82
RW		 2824 ;; Non-portable SSE instructions are used so building fails on platforms
2825 ;; other than x86_64.
2826 (supported-systems '("x86_64-linux"))
346a829a
RW		 2827 (home-page "http://ccb.jhu.edu/software/hisat/index.shtml")
2828 (synopsis "Hierarchical indexing for spliced alignment of transcripts")
2829 (description
2830 "HISAT is a fast and sensitive spliced alignment program for mapping
2831RNA-seq reads. In addition to one global FM index that represents a whole
2832genome, HISAT uses a large set of small FM indexes that collectively cover the
2833whole genome. These small indexes (called local indexes) combined with
2834several alignment strategies enable effective alignment of RNA-seq reads, in
2835particular, reads spanning multiple exons.")
2836 (license license:gpl3+)))
2837
e84efc50
RW		 2838(define-public hisat2
2839 (package
2840 (name "hisat2")
2841 (version "2.0.5")
2842 (source
2843 (origin
2844 (method url-fetch)
2845 ;; FIXME: a better source URL is
2846 ;; (string-append "ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2"
2847 ;; "/downloads/hisat2-" version "-source.zip")
2848 ;; with hash "0lywnr8kijwsc2aw10dwxic0n0yvip6fl3rjlvc8zzwahamy4x7g"
2849 ;; but it is currently unavailable.
2850 (uri "https://github.com/infphilo/hisat2/archive/cba6e8cb.tar.gz")
2851 (file-name (string-append name "-" version ".tar.gz"))
2852 (sha256
2853 (base32
2854 "1mf2hdsyv7cd97xm9mp9a4qws02yrj95y6w6f6cdwnq0klp81r50"))))
2855 (build-system gnu-build-system)
2856 (arguments
2857 `(#:tests? #f ; no check target
2858 #:make-flags (list "CC=gcc" "CXX=g++" "allall")
2859 #:modules ((guix build gnu-build-system)
2860 (guix build utils)
2861 (srfi srfi-26))
2862 #:phases
2863 (modify-phases %standard-phases
2864 (add-after 'unpack 'make-deterministic
2865 (lambda _
2866 (substitute* "Makefile"
2867 (("`date`") "0"))
2868 #t))
2869 (delete 'configure)
2870 (replace 'install
2871 (lambda* (#:key outputs #:allow-other-keys)
2872 (let* ((out (assoc-ref outputs "out"))
2873 (bin (string-append out "/bin/"))
2874 (doc (string-append out "/share/doc/hisat2/")))
2875 (for-each
2876 (cut install-file <> bin)
2877 (find-files "."
2878 "hisat2(-(build|align|inspect)(-(s|l)(-debug)*)*)*$"))
2879 (mkdir-p doc)
2880 (install-file "doc/manual.inc.html" doc))
2881 #t)))))
2882 (native-inputs
2883 `(("unzip" ,unzip) ; needed for archive from ftp
2884 ("perl" ,perl)
2885 ("pandoc" ,ghc-pandoc))) ; for documentation
2886 (home-page "http://ccb.jhu.edu/software/hisat2/index.shtml")
2887 (synopsis "Graph-based alignment of genomic sequencing reads")
2888 (description "HISAT2 is a fast and sensitive alignment program for mapping
2889next-generation sequencing reads (both DNA and RNA) to a population of human
2890genomes (as well as to a single reference genome). In addition to using one
2891global @dfn{graph FM} (GFM) index that represents a population of human
2892genomes, HISAT2 uses a large set of small GFM indexes that collectively cover
2893the whole genome. These small indexes, combined with several alignment
2894strategies, enable rapid and accurate alignment of sequencing reads. This new
2895indexing scheme is called a @dfn{Hierarchical Graph FM index} (HGFM).")
2896 ;; HISAT2 contains files from Bowtie2, which is released under
2897 ;; GPLv2 or later. The HISAT2 source files are released under
2898 ;; GPLv3 or later.
2899 (license license:gpl3+)))
2900
c684629f
BW		 2901(define-public hmmer
2902 (package
2903 (name "hmmer")
2904 (version "3.1b2")
79f09fa2
BW		 2905 (source
2906 (origin
2907 (method url-fetch)
2908 (uri (string-append
2909 "http://eddylab.org/software/hmmer"
2910 (version-prefix version 1) "/"
2911 version "/hmmer-" version ".tar.gz"))
2912 (sha256
2913 (base32
2914 "0djmgc0pfli0jilfx8hql1axhwhqxqb8rxg2r5rg07aw73sfs5nx"))
2915 (patches (search-patches "hmmer-remove-cpu-specificity.patch"))))
c684629f 2916 (build-system gnu-build-system)
b3546174 2917 (native-inputs `(("perl" ,perl)))
a83e6046 2918 (home-page "http://hmmer.org/")
c684629f
BW		 2919 (synopsis "Biosequence analysis using profile hidden Markov models")
2920 (description
2921 "HMMER is used for searching sequence databases for homologs of protein
2922sequences, and for making protein sequence alignments. It implements methods
2923using probabilistic models called profile hidden Markov models (profile
2924HMMs).")
2925 (license (list license:gpl3+
2926 ;; The bundled library 'easel' is distributed
2927 ;; under The Janelia Farm Software License.
2928 (license:non-copyleft
2929 "file://easel/LICENSE"
2930 "See easel/LICENSE in the distribution.")))))
2931
85652f59
RW		 2932(define-public htseq
2933 (package
2934 (name "htseq")
92971d68 2935 (version "0.9.1")
85652f59
RW		 2936 (source (origin
2937 (method url-fetch)
75e6639f 2938 (uri (pypi-uri "HTSeq" version))
85652f59
RW		 2939 (sha256
2940 (base32
92971d68 2941 "11flgb1381xdhk43bzbfm3vhnszkpqg6jk76rpa5xd1zbrvvlnxg"))))
85652f59 2942 (build-system python-build-system)
92971d68
BW		 2943 (native-inputs
2944 `(("python-cython" ,python-cython)))
0536727e
RW		 2945 ;; Numpy needs to be propagated when htseq is used as a Python library.
2946 (propagated-inputs
92971d68 2947 `(("python-numpy" ,python-numpy)))
578b05d9 2948 (inputs
92971d68
BW		 2949 `(("python-pysam" ,python-pysam)
2950 ("python-matplotlib" ,python-matplotlib)))
85652f59
RW		 2951 (home-page "http://www-huber.embl.de/users/anders/HTSeq/")
2952 (synopsis "Analysing high-throughput sequencing data with Python")
2953 (description
2954 "HTSeq is a Python package that provides infrastructure to process data
2955from high-throughput sequencing assays.")
2956 (license license:gpl3+)))
2957
92971d68
BW		 2958(define-public python2-htseq
2959 (package-with-python2 htseq))
2960
1ad15c16 2961(define-public java-htsjdk
15a3c3d4 2962 (package
1ad15c16 2963 (name "java-htsjdk")
15a3c3d4
RW		 2964 (version "1.129")
2965 (source (origin
2966 (method url-fetch)
2967 (uri (string-append
2968 "https://github.com/samtools/htsjdk/archive/"
2969 version ".tar.gz"))
2970 (file-name (string-append name "-" version ".tar.gz"))
2971 (sha256
2972 (base32
2973 "0asdk9b8jx2ij7yd6apg9qx03li8q7z3ml0qy2r2qczkra79y6fw"))
2974 (modules '((guix build utils)))
2975 ;; remove build dependency on git
2976 (snippet '(substitute* "build.xml"
2977 (("failifexecutionfails=\"true\"")
2978 "failifexecutionfails=\"false\"")))))
10b4a969 2979 (build-system ant-build-system)
15a3c3d4 2980 (arguments
10b4a969
RW		 2981 `(#:tests? #f ; test require Internet access
2982 #:make-flags
2983 (list (string-append "-Ddist=" (assoc-ref %outputs "out")
2984 "/share/java/htsjdk/"))
2985 #:build-target "all"
2986 #:phases
2987 (modify-phases %standard-phases
2988 ;; The build phase also installs the jars
2989 (delete 'install))))
15a3c3d4
RW		 2990 (home-page "http://samtools.github.io/htsjdk/")
2991 (synopsis "Java API for high-throughput sequencing data (HTS) formats")
2992 (description
2993 "HTSJDK is an implementation of a unified Java library for accessing
2994common file formats, such as SAM and VCF, used for high-throughput
2995sequencing (HTS) data. There are also an number of useful utilities for
2996manipulating HTS data.")
2997 (license license:expat)))
2998
e7c09730
RW		 2999(define-public htslib
3000 (package
3001 (name "htslib")
4fb6d128 3002 (version "1.5")
e7c09730
RW		 3003 (source (origin
3004 (method url-fetch)
3005 (uri (string-append
3006 "https://github.com/samtools/htslib/releases/download/"
3007 version "/htslib-" version ".tar.bz2"))
3008 (sha256
3009 (base32
4fb6d128 3010 "0bcjmnbwp2bib1z1bkrp95w9v2syzdwdfqww10mkb1hxlmg52ax0"))))
e7c09730
RW		 3011 (build-system gnu-build-system)
3012 (arguments
3013 `(#:phases
3014 (modify-phases %standard-phases
3015 (add-after
3016 'unpack 'patch-tests
3017 (lambda _
3018 (substitute* "test/test.pl"
3019 (("/bin/bash") (which "bash")))
3020 #t)))))
3021 (inputs
a9e4a1e6
BW		 3022 `(("openssl" ,openssl)
3023 ("curl" ,curl)
3024 ("zlib" ,zlib)))
e7c09730
RW		 3025 (native-inputs
3026 `(("perl" ,perl)))
3027 (home-page "http://www.htslib.org")
3028 (synopsis "C library for reading/writing high-throughput sequencing data")
3029 (description
3030 "HTSlib is a C library for reading/writing high-throughput sequencing
3031data. It also provides the bgzip, htsfile, and tabix utilities.")
3032 ;; Files under cram/ are released under the modified BSD license;
3033 ;; the rest is released under the Expat license
3034 (license (list license:expat license:bsd-3))))
3035
bca2c576
BW		 3036;; This package should be removed once no packages rely upon it.
3037(define htslib-1.3
3038 (package
3039 (inherit htslib)
3040 (version "1.3.1")
3041 (source (origin
3042 (method url-fetch)
3043 (uri (string-append
3044 "https://github.com/samtools/htslib/releases/download/"
3045 version "/htslib-" version ".tar.bz2"))
3046 (sha256
3047 (base32
3048 "1rja282fwdc25ql6izkhdyh8ppw8x2fs0w0js78zgkmqjlikmma9"))))))
3049
c4325f62
RW		 3050(define-public idr
3051 (package
3052 (name "idr")
3053 (version "2.0.0")
3054 (source (origin
3055 (method url-fetch)
3056 (uri (string-append
3057 "https://github.com/nboley/idr/archive/"
3058 version ".tar.gz"))
3059 (file-name (string-append name "-" version ".tar.gz"))
3060 (sha256
3061 (base32
3062 "1k3x44biak00aiv3hpm1yd6nn4hhp7n0qnbs3zh2q9sw7qr1qj5r"))))
3063 (build-system python-build-system)
3064 (arguments
14386fc7 3065 `(#:tests? #f)) ; FIXME: "ImportError: No module named 'utility'"
b7a820fc 3066 (propagated-inputs
c4325f62 3067 `(("python-scipy" ,python-scipy)
b7a820fc 3068 ("python-sympy" ,python-sympy)
c4325f62
RW		 3069 ("python-numpy" ,python-numpy)
3070 ("python-matplotlib" ,python-matplotlib)))
3071 (native-inputs
f3b98f4f 3072 `(("python-cython" ,python-cython)))
c4325f62
RW		 3073 (home-page "https://github.com/nboley/idr")
3074 (synopsis "Tool to measure the irreproducible discovery rate (IDR)")
3075 (description
3076 "The IDR (Irreproducible Discovery Rate) framework is a unified approach
3077to measure the reproducibility of findings identified from replicate
3078experiments and provide highly stable thresholds based on reproducibility.")
3079 (license license:gpl3+)))
3080
43c565d2
RW		 3081(define-public jellyfish
3082 (package
3083 (name "jellyfish")
3084 (version "2.2.4")
3085 (source (origin
3086 (method url-fetch)
3087 (uri (string-append "https://github.com/gmarcais/Jellyfish/"
3088 "releases/download/v" version
3089 "/jellyfish-" version ".tar.gz"))
3090 (sha256
3091 (base32
3092 "0a6xnynqy2ibfbfz86b9g2m2dgm7f1469pmymkpam333gi3p26nk"))))
3093 (build-system gnu-build-system)
3094 (outputs '("out" ;for library
3095 "ruby" ;for Ruby bindings
3096 "python")) ;for Python bindings
3097 (arguments
3098 `(#:configure-flags
3099 (list (string-append "--enable-ruby-binding="
3100 (assoc-ref %outputs "ruby"))
3101 (string-append "--enable-python-binding="
3102 (assoc-ref %outputs "python")))
3103 #:phases
3104 (modify-phases %standard-phases
3105 (add-before 'check 'set-SHELL-variable
3106 (lambda _
3107 ;; generator_manager.hpp either uses /bin/sh or $SHELL
3108 ;; to run tests.
3109 (setenv "SHELL" (which "bash"))
3110 #t)))))
3111 (native-inputs
3112 `(("bc" ,bc)
3113 ("time" ,time)
3114 ("ruby" ,ruby)
3115 ("python" ,python-2)))
3116 (synopsis "Tool for fast counting of k-mers in DNA")
3117 (description
3118 "Jellyfish is a tool for fast, memory-efficient counting of k-mers in
3119DNA. A k-mer is a substring of length k, and counting the occurrences of all
3120such substrings is a central step in many analyses of DNA sequence. Jellyfish
3121is a command-line program that reads FASTA and multi-FASTA files containing
3122DNA sequences. It outputs its k-mer counts in a binary format, which can be
3123translated into a human-readable text format using the @code{jellyfish dump}
3124command, or queried for specific k-mers with @code{jellyfish query}.")
3125 (home-page "http://www.genome.umd.edu/jellyfish.html")
6e8faf77
EF		 3126 ;; From their website: JELLYFISH runs on 64-bit Intel-compatible processors
3127 (supported-systems '("x86_64-linux"))
43c565d2
RW		 3128 ;; The combined work is published under the GPLv3 or later. Individual
3129 ;; files such as lib/jsoncpp.cpp are released under the Expat license.
3130 (license (list license:gpl3+ license:expat))))
3131
94ff3157
BW		 3132(define-public khmer
3133 (package
3134 (name "khmer")
3135 (version "2.0")
3136 (source
3137 (origin
3138 (method url-fetch)
3139 (uri (pypi-uri "khmer" version))
3140 (sha256
3141 (base32
3142 "0wb05shqh77v00256qlm68vbbx3kl76fyzihszbz5nhanl4ni33a"))
3143 (patches (search-patches "khmer-use-libraries.patch"))))
3144 (build-system python-build-system)
3145 (arguments
3146 `(#:phases
3147 (modify-phases %standard-phases
3148 (add-after 'unpack 'set-paths
3149 (lambda* (#:key inputs outputs #:allow-other-keys)
3150 ;; Delete bundled libraries.
3151 (delete-file-recursively "third-party/zlib")
3152 (delete-file-recursively "third-party/bzip2")
3153 ;; Replace bundled seqan.
3154 (let* ((seqan-all "third-party/seqan")
3155 (seqan-include (string-append
3156 seqan-all "/core/include")))
3157 (delete-file-recursively seqan-all)
3158 (copy-recursively (string-append (assoc-ref inputs "seqan")
3159 "/include/seqan")
3160 (string-append seqan-include "/seqan")))
3161 ;; We do not replace the bundled MurmurHash as the canonical
3162 ;; repository for this code 'SMHasher' is unsuitable for
3163 ;; providing a library. See
3164 ;; https://lists.gnu.org/archive/html/guix-devel/2016-06/msg00977.html
3165 #t))
3166 (add-after 'unpack 'set-cc
3167 (lambda _
3168 (setenv "CC" "gcc")
3169 #t))
3170 ;; It is simpler to test after installation.
3171 (delete 'check)
3172 (add-after 'install 'post-install-check
3173 (lambda* (#:key inputs outputs #:allow-other-keys)
3174 (let ((out (assoc-ref outputs "out")))
3175 (setenv "PATH"
3176 (string-append
3177 (getenv "PATH")
3178 ":"
3179 (assoc-ref outputs "out")
3180 "/bin"))
3181 (setenv "PYTHONPATH"
3182 (string-append
3183 (getenv "PYTHONPATH")
3184 ":"
3185 out
3186 "/lib/python"
3187 (string-take (string-take-right
3188 (assoc-ref inputs "python") 5) 3)
3189 "/site-packages"))
3190 (with-directory-excursion "build"
3191 (zero? (system* "nosetests" "khmer" "--attr"
3192 "!known_failing")))))))))
3193 (native-inputs
3194 `(("seqan" ,seqan)
3195 ("python-nose" ,python-nose)))
3196 (inputs
3197 `(("zlib" ,zlib)
3198 ("bzip2" ,bzip2)
3199 ("python-screed" ,python-screed)
cf1d5814
BW		 3200 ("python-bz2file" ,python-bz2file)
3201 ;; Tests fail when gcc-5 is used for compilation. Use gcc-4.9 at least
3202 ;; until the next version of khmer (likely 2.1) is released.
3203 ("gcc" ,gcc-4.9)))
94ff3157
BW		 3204 (home-page "https://khmer.readthedocs.org/")
3205 (synopsis "K-mer counting, filtering and graph traversal library")
3206 (description "The khmer software is a set of command-line tools for
3207working with DNA shotgun sequencing data from genomes, transcriptomes,
3208metagenomes and single cells. Khmer can make de novo assemblies faster, and
3209sometimes better. Khmer can also identify and fix problems with shotgun
3210data.")
8157af2e
EF		 3211 ;; When building on i686, armhf and mips64el, we get the following error:
3212 ;; error: ['khmer', 'khmer.tests', 'oxli'] require 64-bit operating system
3213 (supported-systems '("x86_64-linux"))
94ff3157
BW		 3214 (license license:bsd-3)))
3215
b9a601d9
RJ		 3216(define-public kaiju
3217 (package
3218 (name "kaiju")
3219 (version "1.5.0")
3220 (source (origin
3221 (method url-fetch)
3222 (uri (string-append
3223 "https://github.com/bioinformatics-centre/kaiju/archive/v"
3224 version ".tar.gz"))
3225 (file-name (string-append name "-" version ".tar.gz"))
3226 (sha256
3227 (base32
3228 "0afbfalfw9y39bkwnqjrh9bghs118ws1pzj5h8l0nblgn3mbjdks"))))
3229 (build-system gnu-build-system)
3230 (arguments
3231 `(#:tests? #f ; There are no tests.
3232 #:phases
3233 (modify-phases %standard-phases
3234 (delete 'configure)
3235 (add-before 'build 'move-to-src-dir
3236 (lambda _ (chdir "src") #t))
3237 (replace 'install
3238 (lambda* (#:key inputs outputs #:allow-other-keys)
3239 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
3240 (mkdir-p bin)
3241 (chdir "..")
3242 (copy-recursively "bin" bin)
3243 (copy-recursively "util" bin))
3244 #t)))))
3245 (inputs
3246 `(("perl" ,perl)))
3247 (home-page "http://kaiju.binf.ku.dk/")
3248 (synopsis "Fast and sensitive taxonomic classification for metagenomics")
3249 (description "Kaiju is a program for sensitive taxonomic classification
3250of high-throughput sequencing reads from metagenomic whole genome sequencing
3251experiments.")
3252 (license license:gpl3+)))
3253
d57e6d0f
RW		 3254(define-public macs
3255 (package
3256 (name "macs")
0d0bcaa0 3257 (version "2.1.0.20151222")
d57e6d0f
RW		 3258 (source (origin
3259 (method url-fetch)
43ec07f1 3260 (uri (pypi-uri "MACS2" version))
d57e6d0f
RW		 3261 (sha256
3262 (base32
0d0bcaa0 3263 "1r2hcz6irhcq7lwbafjks98jbn34hv05avgbdjnp6w6mlfjkf8x5"))))
d57e6d0f
RW		 3264 (build-system python-build-system)
3265 (arguments
3266 `(#:python ,python-2 ; only compatible with Python 2.7
3267 #:tests? #f)) ; no test target
3268 (inputs
3269 `(("python-numpy" ,python2-numpy)))
7bf837fd 3270 (home-page "https://github.com/taoliu/MACS/")
d57e6d0f
RW		 3271 (synopsis "Model based analysis for ChIP-Seq data")
3272 (description
3273 "MACS is an implementation of a ChIP-Seq analysis algorithm for
3274identifying transcript factor binding sites named Model-based Analysis of
3275ChIP-Seq (MACS). MACS captures the influence of genome complexity to evaluate
3276the significance of enriched ChIP regions and it improves the spatial
3277resolution of binding sites through combining the information of both
3278sequencing tag position and orientation.")
3279 (license license:bsd-3)))
3280
41ddebdd
BW		 3281(define-public mafft
3282 (package
3283 (name "mafft")
88eb119f 3284 (version "7.310")
41ddebdd
BW		 3285 (source (origin
3286 (method url-fetch)
3287 (uri (string-append
3288 "http://mafft.cbrc.jp/alignment/software/mafft-" version
3289 "-without-extensions-src.tgz"))
3290 (file-name (string-append name "-" version ".tgz"))
3291 (sha256
3292 (base32
88eb119f 3293 "0gbsaz6z2qa307kd7wfb06c3y4ikmv1hsdvlns11f6zq4w1z9pwc"))))
41ddebdd
BW		 3294 (build-system gnu-build-system)
3295 (arguments
3296 `(#:tests? #f ; no automated tests, though there are tests in the read me
3297 #:make-flags (let ((out (assoc-ref %outputs "out")))
3298 (list (string-append "PREFIX=" out)
3299 (string-append "BINDIR="
3300 (string-append out "/bin"))))
3301 #:phases
3302 (modify-phases %standard-phases
3303 (add-after 'unpack 'enter-dir
101e8f71 3304 (lambda _ (chdir "core") #t))
41ddebdd 3305 (add-after 'enter-dir 'patch-makefile
101e8f71
BW		 3306 (lambda _
3307 ;; on advice from the MAFFT authors, there is no need to
3308 ;; distribute mafft-profile, mafft-distance, or
3309 ;; mafft-homologs.rb as they are too "specialised".
3310 (substitute* "Makefile"
3311 ;; remove mafft-homologs.rb from SCRIPTS
3312 (("^SCRIPTS = mafft mafft-homologs.rb")
3313 "SCRIPTS = mafft")
3314 ;; remove mafft-homologs from MANPAGES
3315 (("^MANPAGES = mafft.1 mafft-homologs.1")
3316 "MANPAGES = mafft.1")
3317 ;; remove mafft-distance from PROGS
3318 (("^PROGS = dvtditr dndfast7 dndblast sextet5 mafft-distance")
3319 "PROGS = dvtditr dndfast7 dndblast sextet5")
3320 ;; remove mafft-profile from PROGS
3321 (("splittbfast disttbfast tbfast mafft-profile 2cl mccaskillwrap")
3322 "splittbfast disttbfast tbfast f2cl mccaskillwrap")
3323 (("^rm -f mafft-profile mafft-profile.exe") "#")
3324 (("^rm -f mafft-distance mafft-distance.exe") ")#")
3325 ;; do not install MAN pages in libexec folder
3326 (("^\t\\$\\(INSTALL\\) -m 644 \\$\\(MANPAGES\\) \
41ddebdd 3327\\$\\(DESTDIR\\)\\$\\(LIBDIR\\)") "#"))
101e8f71 3328 #t))
02f35bb5
BW		 3329 (add-after 'enter-dir 'patch-paths
3330 (lambda* (#:key inputs #:allow-other-keys)
3331 (substitute* '("pairash.c"
3332 "mafft.tmpl")
3333 (("perl") (which "perl"))
3334 (("([\"`| ])awk" _ prefix)
3335 (string-append prefix (which "awk")))
3336 (("grep") (which "grep")))
3337 #t))
101e8f71
BW		 3338 (delete 'configure)
3339 (add-after 'install 'wrap-programs
3340 (lambda* (#:key outputs #:allow-other-keys)
3341 (let* ((out (assoc-ref outputs "out"))
3342 (bin (string-append out "/bin"))
3343 (path (string-append
3344 (assoc-ref %build-inputs "coreutils") "/bin:")))
3345 (for-each (lambda (file)
3346 (wrap-program file
3347 `("PATH" ":" prefix (,path))))
3348 (find-files bin)))
3349 #t)))))
41ddebdd 3350 (inputs
02f35bb5 3351 `(("perl" ,perl)
71461f88 3352 ("ruby" ,ruby)
02f35bb5 3353 ("gawk" ,gawk)
101e8f71
BW		 3354 ("grep" ,grep)
3355 ("coreutils" ,coreutils)))
41ddebdd
BW		 3356 (home-page "http://mafft.cbrc.jp/alignment/software/")
3357 (synopsis "Multiple sequence alignment program")
3358 (description
3359 "MAFFT offers a range of multiple alignment methods for nucleotide and
3360protein sequences. For instance, it offers L-INS-i (accurate; for alignment
3361of <~200 sequences) and FFT-NS-2 (fast; for alignment of <~30,000
3362sequences).")
3363 (license (license:non-copyleft
3364 "http://mafft.cbrc.jp/alignment/software/license.txt"
3365 "BSD-3 with different formatting"))))
8fd790eb 3366
84be3b99
MB		 3367(define-public mash
3368 (package
3369 (name "mash")
3370 (version "1.1.1")
3371 (source (origin
3372 (method url-fetch)
3373 (uri (string-append
3374 "https://github.com/marbl/mash/archive/v"
3375 version ".tar.gz"))
3376 (file-name (string-append name "-" version ".tar.gz"))
3377 (sha256
3378 (base32
3379 "08znbvqq5xknfhmpp3wcj574zvi4p7i8zifi67c9qw9a6ikp42fj"))
3380 (modules '((guix build utils)))
3381 (snippet
3382 ;; Delete bundled kseq.
3383 ;; TODO: Also delete bundled murmurhash and open bloom filter.
3384 '(delete-file "src/mash/kseq.h"))))
3385 (build-system gnu-build-system)
3386 (arguments
3387 `(#:tests? #f ; No tests.
3388 #:configure-flags
3389 (list
3390 (string-append "--with-capnp=" (assoc-ref %build-inputs "capnproto"))
3391 (string-append "--with-gsl=" (assoc-ref %build-inputs "gsl")))
3392 #:make-flags (list "CC=gcc")
3393 #:phases
3394 (modify-phases %standard-phases
3395 (add-after 'unpack 'fix-includes
3396 (lambda _
3397 (substitute* '("src/mash/Sketch.cpp" "src/mash/CommandFind.cpp")
3398 (("^#include \"kseq\\.h\"")
3399 "#include \"htslib/kseq.h\""))
3400 #t))
d10092b8 3401 (add-after 'fix-includes 'autoconf
84be3b99
MB		 3402 (lambda _ (zero? (system* "autoconf")))))))
3403 (native-inputs
3404 `(("autoconf" ,autoconf)
3405 ;; Capnproto and htslib are statically embedded in the final
3406 ;; application. Therefore we also list their licenses, below.
3407 ("capnproto" ,capnproto)
3408 ("htslib" ,htslib)))
3409 (inputs
3410 `(("gsl" ,gsl)
3411 ("zlib" ,zlib)))
3412 (supported-systems '("x86_64-linux"))
3413 (home-page "https://mash.readthedocs.io")
3414 (synopsis "Fast genome and metagenome distance estimation using MinHash")
3415 (description "Mash is a fast sequence distance estimator that uses the
3416MinHash algorithm and is designed to work with genomes and metagenomes in the
3417form of assemblies or reads.")
3418 (license (list license:bsd-3 ; Mash
3419 license:expat ; HTSlib and capnproto
3420 license:public-domain ; MurmurHash 3
3421 license:cpl1.0)))) ; Open Bloom Filter
3422
8fd790eb 3423(define-public metabat
f3f68a44
BW		 3424 (package
3425 (name "metabat")
25bd1fc1 3426 (version "2.12.1")
f3f68a44
BW		 3427 (source
3428 (origin
25bd1fc1
BW		 3429 (method url-fetch)
3430 (uri (string-append "https://bitbucket.org/berkeleylab/metabat/get/v"
3431 version ".tar.gz"))
3432 (file-name (string-append name "-" version ".tar.gz"))
f3f68a44
BW		 3433 (sha256
3434 (base32
25bd1fc1
BW		 3435 "1hmvdalz3zj5sqqklg0l4npjdv37cv2hsdi1al9iby2ndxjs1b73"))
3436 (patches (search-patches "metabat-fix-compilation.patch"))))
8fd790eb
BW		 3437 (build-system gnu-build-system)
3438 (arguments
3439 `(#:phases
3440 (modify-phases %standard-phases
3441 (add-after 'unpack 'fix-includes
45469ebe
BW		 3442 (lambda _
3443 (substitute* "src/BamUtils.h"
3444 (("^#include \"bam/bam\\.h\"")
3445 "#include \"samtools/bam.h\"")
3446 (("^#include \"bam/sam\\.h\"")
3447 "#include \"samtools/sam.h\""))
3448 (substitute* "src/KseqReader.h"
3449 (("^#include \"bam/kseq\\.h\"")
3450 "#include \"htslib/kseq.h\""))
3451 #t))
8fd790eb 3452 (add-after 'unpack 'fix-scons
f3f68a44
BW		 3453 (lambda* (#:key inputs #:allow-other-keys)
3454 (substitute* "SConstruct"
3455 (("^htslib_dir += 'samtools'")
3456 (string-append "htslib_dir = '"
3457 (assoc-ref inputs "htslib")
3458 "'"))
3459 (("^samtools_dir = 'samtools'")
3460 (string-append "samtools_dir = '"
3461 (assoc-ref inputs "samtools")
3462 "'"))
3463 (("^findStaticOrShared\\('bam', hts_lib")
3464 (string-append "findStaticOrShared('bam', '"
3465 (assoc-ref inputs "samtools")
3466 "/lib'"))
3467 ;; Do not distribute README.
3468 (("^env\\.Install\\(idir_prefix, 'README\\.md'\\)") ""))
3469 #t))
8fd790eb
BW		 3470 (delete 'configure)
3471 (replace 'build
f3f68a44
BW		 3472 (lambda* (#:key inputs outputs #:allow-other-keys)
3473 (mkdir (assoc-ref outputs "out"))
3474 (zero? (system* "scons"
3475 (string-append
3476 "PREFIX="
3477 (assoc-ref outputs "out"))
3478 (string-append
3479 "BOOST_ROOT="
3480 (assoc-ref inputs "boost"))
3481 "install"))))
45469ebe 3482 ;; Check and install are carried out during build phase.
8fd790eb
BW		 3483 (delete 'check)
3484 (delete 'install))))
3485 (inputs
3486 `(("zlib" ,zlib)
3487 ("perl" ,perl)
3488 ("samtools" ,samtools)
3489 ("htslib" ,htslib)
3490 ("boost" ,boost)))
3491 (native-inputs
3492 `(("scons" ,scons)))
3493 (home-page "https://bitbucket.org/berkeleylab/metabat")
3494 (synopsis
3495 "Reconstruction of single genomes from complex microbial communities")
3496 (description
3497 "Grouping large genomic fragments assembled from shotgun metagenomic
3498sequences to deconvolute complex microbial communities, or metagenome binning,
3499enables the study of individual organisms and their interactions. MetaBAT is
3500an automated metagenome binning software, which integrates empirical
3501probabilistic distances of genome abundance and tetranucleotide frequency.")
d931a4bb
EF		 3502 ;; The source code contains inline assembly.
3503 (supported-systems '("x86_64-linux" "i686-linux"))
f3f68a44
BW		 3504 (license (license:non-copyleft "file://license.txt"
3505 "See license.txt in the distribution."))))
8fd790eb 3506
318c0aee
MB		 3507(define-public minced
3508 (package
3509 (name "minced")
3510 (version "0.2.0")
3511 (source (origin
3512 (method url-fetch)
3513 (uri (string-append
3514 "https://github.com/ctSkennerton/minced/archive/"
3515 version ".tar.gz"))
3516 (file-name (string-append name "-" version ".tar.gz"))
3517 (sha256
3518 (base32
3519 "0wxmlsapxfpxfd3ps9636h7i2xy6la8i42mwh0j2lsky63h63jp1"))))
3520 (build-system gnu-build-system)
3521 (arguments
3522 `(#:test-target "test"
3523 #:phases
3524 (modify-phases %standard-phases
3525 (delete 'configure)
3526 (add-before 'check 'fix-test
3527 (lambda _
3528 ;; Fix test for latest version.
3529 (substitute* "t/Aquifex_aeolicus_VF5.expected"
3530 (("minced:0.1.6") "minced:0.2.0"))
3531 #t))
3532 (replace 'install ; No install target.
3533 (lambda* (#:key inputs outputs #:allow-other-keys)
3534 (let* ((out (assoc-ref outputs "out"))
3535 (bin (string-append out "/bin"))
3536 (wrapper (string-append bin "/minced")))
3537 ;; Minced comes with a wrapper script that tries to figure out where
3538 ;; it is located before running the JAR. Since these paths are known
3539 ;; to us, we build our own wrapper to avoid coreutils dependency.
3540 (install-file "minced.jar" bin)
3541 (with-output-to-file wrapper
3542 (lambda _
3543 (display
3544 (string-append
3545 "#!" (assoc-ref inputs "bash") "/bin/sh\n\n"
3546 (assoc-ref inputs "jre") "/bin/java -jar "
3547 bin "/minced.jar \"$@\"\n"))))
3548 (chmod wrapper #o555)))))))
3549 (native-inputs
3550 `(("jdk" ,icedtea "jdk")))
3551 (inputs
3552 `(("bash" ,bash)
3553 ("jre" ,icedtea "out")))
3554 (home-page "https://github.com/ctSkennerton/minced")
3555 (synopsis "Mining CRISPRs in Environmental Datasets")
3556 (description
3557 "MinCED is a program to find Clustered Regularly Interspaced Short
3558Palindromic Repeats (CRISPRs) in DNA sequences. It can be used for
3559unassembled metagenomic reads, but is mainly designed for full genomes and
3560assembled metagenomic sequence.")
3561 (license license:gpl3+)))
3562
ddd82e0e
RW		 3563(define-public miso
3564 (package
3565 (name "miso")
ce4dfde0 3566 (version "0.5.4")
ddd82e0e
RW		 3567 (source (origin
3568 (method url-fetch)
34260a10 3569 (uri (pypi-uri "misopy" version))
ddd82e0e
RW		 3570 (sha256
3571 (base32
ce4dfde0 3572 "1z3x0vd8ma7pdrnywj7i3kgwl89sdkwrrn62zl7r5calqaq2hyip"))
ddd82e0e 3573 (modules '((guix build utils)))
ddd82e0e
RW		 3574 (snippet
3575 '(substitute* "setup.py"
0ec8b206
RW		 3576 ;; Use setuptools, or else the executables are not
3577 ;; installed.
3578 (("distutils.core") "setuptools")
3579 ;; use "gcc" instead of "cc" for compilation
ddd82e0e
RW		 3580 (("^defines")
3581 "cc.set_executables(
3582compiler='gcc',
3583compiler_so='gcc',
3584linker_exe='gcc',
3585linker_so='gcc -shared'); defines")))))
3586 (build-system python-build-system)
3587 (arguments
3588 `(#:python ,python-2 ; only Python 2 is supported
3589 #:tests? #f)) ; no "test" target
3590 (inputs
3591 `(("samtools" ,samtools)
3592 ("python-numpy" ,python2-numpy)
3593 ("python-pysam" ,python2-pysam)
3594 ("python-scipy" ,python2-scipy)
3595 ("python-matplotlib" ,python2-matplotlib)))
3596 (native-inputs
f3b98f4f
HG		 3597 `(("python-mock" ,python2-mock) ;for tests
3598 ("python-pytz" ,python2-pytz))) ;for tests
ddd82e0e
RW		 3599 (home-page "http://genes.mit.edu/burgelab/miso/index.html")
3600 (synopsis "Mixture of Isoforms model for RNA-Seq isoform quantitation")
3601 (description
3602 "MISO (Mixture-of-Isoforms) is a probabilistic framework that quantitates
3603the expression level of alternatively spliced genes from RNA-Seq data, and
3604identifies differentially regulated isoforms or exons across samples. By
3605modeling the generative process by which reads are produced from isoforms in
3606RNA-Seq, the MISO model uses Bayesian inference to compute the probability
3607that a read originated from a particular isoform.")
3608 (license license:gpl2)))
3609
324efb88
BW		 3610(define-public muscle
3611 (package
3612 (name "muscle")
3613 (version "3.8.1551")
3614 (source (origin
3615 (method url-fetch/tarbomb)
324efb88
BW		 3616 (uri (string-append
3617 "http://www.drive5.com/muscle/muscle_src_"
3618 version ".tar.gz"))
3619 (sha256
3620 (base32
3621 "0bj8kj7sdizy3987zx6w7axihk40fk8rn76mpbqqjcnd64i5a367"))))
3622 (build-system gnu-build-system)
3623 (arguments
3624 `(#:make-flags (list "LDLIBS = -lm")
3625 #:phases
3626 (modify-phases %standard-phases
3627 (delete 'configure)
3628 (replace 'check
3629 ;; There are no tests, so just test if it runs.
3630 (lambda _ (zero? (system* "./muscle" "-version"))))
3631 (replace 'install
3632 (lambda* (#:key outputs #:allow-other-keys)
3633 (let* ((out (assoc-ref outputs "out"))
3634 (bin (string-append out "/bin")))
3635 (install-file "muscle" bin)))))))
3636 (home-page "http://www.drive5.com/muscle")
3637 (synopsis "Multiple sequence alignment program")
3638 (description
3639 "MUSCLE aims to be a fast and accurate multiple sequence alignment
3640program for nucleotide and protein sequences.")
3641 ;; License information found in 'muscle -h' and usage.cpp.
3642 (license license:public-domain)))
3643
99268755
BW		 3644(define-public newick-utils
3645 ;; There are no recent releases so we package from git.
3646 (let ((commit "da121155a977197cab9fbb15953ca1b40b11eb87"))
3647 (package
3648 (name "newick-utils")
3649 (version (string-append "1.6-1." (string-take commit 8)))
3650 (source (origin
3651 (method git-fetch)
3652 (uri (git-reference
3653 (url "https://github.com/tjunier/newick_utils.git")
3654 (commit commit)))
3655 (file-name (string-append name "-" version "-checkout"))
3656 (sha256
3657 (base32
3658 "1hkw21rq1mwf7xp0rmbb2gqc0i6p11108m69i7mr7xcjl268pxnb"))))
3659 (build-system gnu-build-system)
3660 (arguments
3661 `(#:phases
3662 (modify-phases %standard-phases
3663 (add-after 'unpack 'autoconf
3664 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
3665 (inputs
3666 ;; XXX: TODO: Enable Lua and Guile bindings.
3667 ;; https://github.com/tjunier/newick_utils/issues/13
3668 `(("libxml2" ,libxml2)
3669 ("flex" ,flex)
3670 ("bison" ,bison)))
3671 (native-inputs
3672 `(("autoconf" ,autoconf)
3673 ("automake" ,automake)
3674 ("libtool" ,libtool)))
3675 (synopsis "Programs for working with newick format phylogenetic trees")
3676 (description
3677 "Newick-utils is a suite of utilities for processing phylogenetic trees
3678in Newick format. Functions include re-rooting, extracting subtrees,
3679trimming, pruning, condensing, drawing (ASCII graphics or SVG).")
3680 (home-page "https://github.com/tjunier/newick_utils")
3681 (license license:bsd-3))))
3682
1e44cf8b
BW		 3683(define-public orfm
3684 (package
3685 (name "orfm")
dfc83ead 3686 (version "0.7.1")
1e44cf8b
BW		 3687 (source (origin
3688 (method url-fetch)
3689 (uri (string-append
3690 "https://github.com/wwood/OrfM/releases/download/v"
3691 version "/orfm-" version ".tar.gz"))
3692 (sha256
3693 (base32
dfc83ead 3694 "16iigyr2gd8x0imzkk1dr3k5xsds9bpmwg31ayvjg0f4pir9rwqr"))))
1e44cf8b
BW		 3695 (build-system gnu-build-system)
3696 (inputs `(("zlib" ,zlib)))
6b6f7d6a
BW		 3697 (native-inputs
3698 `(("ruby-bio-commandeer" ,ruby-bio-commandeer)
3699 ("ruby-rspec" ,ruby-rspec)
3700 ("ruby" ,ruby)))
1e44cf8b
BW		 3701 (synopsis "Simple and not slow open reading frame (ORF) caller")
3702 (description
6b6f7d6a 3703 "An ORF caller finds stretches of DNA that, when translated, are not
1e44cf8b
BW		 3704interrupted by stop codons. OrfM finds and prints these ORFs.")
3705 (home-page "https://github.com/wwood/OrfM")
3706 (license license:lgpl3+)))
3707
c033f5d6
BW		 3708(define-public pplacer
3709 (let ((commit "g807f6f3"))
3710 (package
3711 (name "pplacer")
3712 ;; The commit should be updated with each version change.
3713 (version "1.1.alpha19")
3714 (source
3715 (origin
3716 (method url-fetch)
3717 (uri (string-append "https://github.com/matsen/pplacer/archive/v"
3718 version ".tar.gz"))
3719 (file-name (string-append name "-" version ".tar.gz"))
3720 (sha256
3721 (base32 "0z1lnd2s8sh6kpzg106wzbh2szw7h0hvq8syd5a6wv4rmyyz6x0f"))))
3722 (build-system ocaml-build-system)
3723 (arguments
3724 `(#:ocaml ,ocaml-4.01
3725 #:findlib ,ocaml4.01-findlib
3726 #:modules ((guix build ocaml-build-system)
3727 (guix build utils)
3728 (ice-9 ftw))
3729 #:phases
3730 (modify-phases %standard-phases
3731 (delete 'configure)
3732 (add-after 'unpack 'replace-bundled-cddlib
3733 (lambda* (#:key inputs #:allow-other-keys)
3734 (let* ((cddlib-src (assoc-ref inputs "cddlib-src"))
3735 (local-dir "cddlib_guix"))
3736 (mkdir local-dir)
3737 (with-directory-excursion local-dir
3738 (system* "tar" "xvf" cddlib-src))
3739 (let ((cddlib-src-folder
3740 (string-append local-dir "/"
3741 (list-ref (scandir local-dir) 2)
3742 "/lib-src")))
3743 (for-each
3744 (lambda (file)
3745 (copy-file file
3746 (string-append "cdd_src/" (basename file))))
3747 (find-files cddlib-src-folder ".*[ch]$")))
3748 #t)))
3749 (add-after 'unpack 'fix-makefile
3750 (lambda _
3751 ;; Remove system calls to 'git'.
3752 (substitute* "Makefile"
3753 (("^DESCRIPT:=pplacer-.*")
3754 (string-append
3755 "DESCRIPT:=pplacer-$(shell uname)-v" ,version "\n")))
3756 (substitute* "myocamlbuild.ml"
3757 (("git describe --tags --long .*\\\" with")
3758 (string-append
3759 "echo -n v" ,version "-" ,commit "\" with")))
3760 #t))
3761 (replace 'install
3762 (lambda* (#:key outputs #:allow-other-keys)
3763 (let* ((out (assoc-ref outputs "out"))
3764 (bin (string-append out "/bin")))
3765 (copy-recursively "bin" bin))
3766 #t)))))
3767 (native-inputs
3768 `(("zlib" ,zlib)
3769 ("gsl" ,gsl)
3770 ("ocaml-ounit" ,ocaml4.01-ounit)
3771 ("ocaml-batteries" ,ocaml4.01-batteries)
3772 ("ocaml-camlzip" ,ocaml4.01-camlzip)
3773 ("ocaml-csv" ,ocaml4.01-csv)
3774 ("ocaml-sqlite3" ,ocaml4.01-sqlite3)
3775 ("ocaml-xmlm" ,ocaml4.01-xmlm)
3776 ("ocaml-mcl" ,ocaml4.01-mcl)
3777 ("ocaml-gsl" ,ocaml4.01-gsl)
3778 ("cddlib-src" ,(package-source cddlib))))
3779 (propagated-inputs
3780 `(("pplacer-scripts" ,pplacer-scripts)))
3781 (synopsis "Phylogenetic placement of biological sequences")
3782 (description
3783 "Pplacer places query sequences on a fixed reference phylogenetic tree
3784to maximize phylogenetic likelihood or posterior probability according to a
3785reference alignment. Pplacer is designed to be fast, to give useful
3786information about uncertainty, and to offer advanced visualization and
3787downstream analysis.")
3788 (home-page "http://matsen.fhcrc.org/pplacer")
3789 (license license:gpl3))))
3790
3791;; This package is installed alongside 'pplacer'. It is a separate package so
3792;; that it can use the python-build-system for the scripts that are
3793;; distributed alongside the main OCaml binaries.
3794(define pplacer-scripts
3795 (package
3796 (inherit pplacer)
3797 (name "pplacer-scripts")
3798 (build-system python-build-system)
3799 (arguments
3800 `(#:python ,python-2
3801 #:phases
3802 (modify-phases %standard-phases
3803 (add-after 'unpack 'enter-scripts-dir
3804 (lambda _ (chdir "scripts")))
3805 (replace 'check
3806 (lambda _
3807 (zero? (system* "python" "-m" "unittest" "discover" "-v"))))
3808 (add-after 'install 'wrap-executables
3809 (lambda* (#:key inputs outputs #:allow-other-keys)
3810 (let* ((out (assoc-ref outputs "out"))
3811 (bin (string-append out "/bin")))
3812 (let ((path (string-append
3813 (assoc-ref inputs "hmmer") "/bin:"
3814 (assoc-ref inputs "infernal") "/bin")))
3815 (display path)
3816 (wrap-program (string-append bin "/refpkg_align.py")
3817 `("PATH" ":" prefix (,path))))
3818 (let ((path (string-append
3819 (assoc-ref inputs "hmmer") "/bin")))
3820 (wrap-program (string-append bin "/hrefpkg_query.py")
3821 `("PATH" ":" prefix (,path)))))
3822 #t)))))
3823 (inputs
3824 `(("infernal" ,infernal)
3825 ("hmmer" ,hmmer)))
3826 (propagated-inputs
3827 `(("python-biopython" ,python2-biopython)
3828 ("taxtastic" ,taxtastic)))
3829 (synopsis "Pplacer Python scripts")))
3830
19ee9201
RW		 3831(define-public python2-pbcore
3832 (package
3833 (name "python2-pbcore")
e301bfc8 3834 (version "1.2.10")
19ee9201
RW		 3835 (source (origin
3836 (method url-fetch)
ddb83129 3837 (uri (pypi-uri "pbcore" version))
19ee9201
RW		 3838 (sha256
3839 (base32
e301bfc8 3840 "1kjmv891d6qbpp4shhhvkl02ff4q5xlpnls2513sm2cjcrs52f1i"))))
19ee9201
RW		 3841 (build-system python-build-system)
3842 (arguments `(#:python ,python-2)) ; pbcore requires Python 2.7
de5bc890 3843 (propagated-inputs
19ee9201
RW		 3844 `(("python-cython" ,python2-cython)
3845 ("python-numpy" ,python2-numpy)
3846 ("python-pysam" ,python2-pysam)
3847 ("python-h5py" ,python2-h5py)))
3848 (native-inputs
de5bc890
HG		 3849 `(("python-nose" ,python2-nose)
3850 ("python-sphinx" ,python2-sphinx)
3851 ("python-pyxb" ,python2-pyxb)))
19ee9201
RW		 3852 (home-page "http://pacificbiosciences.github.io/pbcore/")
3853 (synopsis "Library for reading and writing PacBio data files")
3854 (description
3855 "The pbcore package provides Python APIs for interacting with PacBio data
3856files and writing bioinformatics applications.")
3857 (license license:bsd-3)))
3858
c61fe02c
RW		 3859(define-public python2-warpedlmm
3860 (package
3861 (name "python2-warpedlmm")
3862 (version "0.21")
3863 (source
3864 (origin
3865 (method url-fetch)
3866 (uri (string-append
3867 "https://pypi.python.org/packages/source/W/WarpedLMM/WarpedLMM-"
3868 version ".zip"))
3869 (sha256
3870 (base32
3871 "1agfz6zqa8nc6cw47yh0s3y14gkpa9wqazwcj7mwwj3ffnw39p3j"))))
3872 (build-system python-build-system)
3873 (arguments
54c85e12 3874 `(#:python ,python-2)) ; requires Python 2.7
c61fe02c
RW		 3875 (propagated-inputs
3876 `(("python-scipy" ,python2-scipy)
3877 ("python-numpy" ,python2-numpy)
3878 ("python-matplotlib" ,python2-matplotlib)
3879 ("python-fastlmm" ,python2-fastlmm)
3880 ("python-pandas" ,python2-pandas)
3881 ("python-pysnptools" ,python2-pysnptools)))
3882 (native-inputs
f3b98f4f 3883 `(("python-mock" ,python2-mock)
c61fe02c
RW		 3884 ("python-nose" ,python2-nose)
3885 ("unzip" ,unzip)))
3886 (home-page "https://github.com/PMBio/warpedLMM")
3887 (synopsis "Implementation of warped linear mixed models")
3888 (description
3889 "WarpedLMM is a Python implementation of the warped linear mixed model,
3890which automatically learns an optimal warping function (or transformation) for
3891the phenotype as it models the data.")
3892 (license license:asl2.0)))
3893
2c16316e 3894(define-public pbtranscript-tofu
698bd297 3895 (let ((commit "8f5467fe6a4472bcfb4226c8720993c8507adfe4"))
2c16316e
RW		 3896 (package
3897 (name "pbtranscript-tofu")
698bd297 3898 (version (string-append "2.2.3." (string-take commit 7)))
2c16316e
RW		 3899 (source (origin
3900 (method git-fetch)
3901 (uri (git-reference
3902 (url "https://github.com/PacificBiosciences/cDNA_primer.git")
3903 (commit commit)))
9a067efd 3904 (file-name (string-append name "-" version "-checkout"))
2c16316e
RW		 3905 (sha256
3906 (base32
9a067efd
RW		 3907 "1lgnpi35ihay42qx0b6yl3kkgra723i413j33kvs0kvs61h82w0f"))
3908 (modules '((guix build utils)))
3909 (snippet
3910 '(begin
3911 ;; remove bundled Cython sources
3912 (delete-file "pbtranscript-tofu/pbtranscript/Cython-0.20.1.tar.gz")
3913 #t))))
2c16316e
RW		 3914 (build-system python-build-system)
3915 (arguments
3916 `(#:python ,python-2
cdc2bb50
MB		 3917 ;; FIXME: Tests fail with "No such file or directory:
3918 ;; pbtools/pbtranscript/modified_bx_intervals/intersection_unique.so"
3919 #:tests? #f
2c16316e 3920 #:phases
9a067efd
RW		 3921 (modify-phases %standard-phases
3922 (add-after 'unpack 'enter-directory
3923 (lambda _
3924 (chdir "pbtranscript-tofu/pbtranscript/")
3925 #t))
3926 ;; With setuptools version 18.0 and later this setup.py hack causes
3927 ;; a build error, so we disable it.
3928 (add-after 'enter-directory 'patch-setuppy
3929 (lambda _
3930 (substitute* "setup.py"
3931 (("if 'setuptools.extension' in sys.modules:")
3932 "if False:"))
3933 #t)))))
2c16316e 3934 (inputs
9a067efd 3935 `(("python-numpy" ,python2-numpy)
2c16316e 3936 ("python-bx-python" ,python2-bx-python)
c5372108
RW		 3937 ("python-networkx" ,python2-networkx)
3938 ("python-scipy" ,python2-scipy)
9a067efd
RW		 3939 ("python-pbcore" ,python2-pbcore)
3940 ("python-h5py" ,python2-h5py)))
2c16316e 3941 (native-inputs
9a067efd 3942 `(("python-cython" ,python2-cython)
f3b98f4f 3943 ("python-nose" ,python2-nose)))
2c16316e
RW		 3944 (home-page "https://github.com/PacificBiosciences/cDNA_primer")
3945 (synopsis "Analyze transcriptome data generated with the Iso-Seq protocol")
3946 (description
3947 "pbtranscript-tofu contains scripts to analyze transcriptome data
3948generated using the PacBio Iso-Seq protocol.")
3949 (license license:bsd-3))))
3950
024130d2
BW		 3951(define-public prank
3952 (package
3953 (name "prank")
3954 (version "150803")
3955 (source (origin
3956 (method url-fetch)
3957 (uri (string-append
3958 "http://wasabiapp.org/download/prank/prank.source."
3959 version ".tgz"))
3960 (sha256
3961 (base32
3962 "0am4z94fs3w2n5xpfls9zda61vq7qqz4q2i7b9hlsxz5q4j3kfm4"))))
3963 (build-system gnu-build-system)
3964 (arguments
3965 `(#:phases
3966 (modify-phases %standard-phases
3967 (add-after 'unpack 'enter-src-dir
3968 (lambda _
3969 (chdir "src")
3970 #t))
62d00095
EF		 3971 (add-after 'unpack 'remove-m64-flag
3972 ;; Prank will build with the correct 'bit-ness' without this flag
3973 ;; and this allows building on 32-bit machines.
3974 (lambda _ (substitute* "src/Makefile"
3975 (("-m64") ""))
3976 #t))
024130d2
BW		 3977 (delete 'configure)
3978 (replace 'install
3979 (lambda* (#:key outputs #:allow-other-keys)
3980 (let* ((out (assoc-ref outputs "out"))
3981 (bin (string-append out "/bin"))
3982 (man (string-append out "/share/man/man1"))
3983 (path (string-append
3984 (assoc-ref %build-inputs "mafft") "/bin:"
3985 (assoc-ref %build-inputs "exonerate") "/bin:"
3986 (assoc-ref %build-inputs "bppsuite") "/bin")))
3987 (install-file "prank" bin)
3988 (wrap-program (string-append bin "/prank")
3989 `("PATH" ":" prefix (,path)))
3990 (install-file "prank.1" man))
3991 #t)))))
3992 (inputs
3993 `(("mafft" ,mafft)
3994 ("exonerate" ,exonerate)
3995 ("bppsuite" ,bppsuite)))
3996 (home-page "http://wasabiapp.org/software/prank/")
3997 (synopsis "Probabilistic multiple sequence alignment program")
3998 (description
3999 "PRANK is a probabilistic multiple sequence alignment program for DNA,
4000codon and amino-acid sequences. It is based on a novel algorithm that treats
4001insertions correctly and avoids over-estimation of the number of deletion
4002events. In addition, PRANK borrows ideas from maximum likelihood methods used
4003in phylogenetics and correctly takes into account the evolutionary distances
4004between sequences. Lastly, PRANK allows for defining a potential structure
4005for sequences to be aligned and then, simultaneously with the alignment,
4006predicts the locations of structural units in the sequences.")
4007 (license license:gpl2+)))
4008
31a9d653
BW		 4009(define-public proteinortho
4010 (package
4011 (name "proteinortho")
8c864901 4012 (version "5.16b")
31a9d653
BW		 4013 (source
4014 (origin
4015 (method url-fetch)
4016 (uri
4017 (string-append
4018 "http://www.bioinf.uni-leipzig.de/Software/proteinortho/proteinortho_v"
4019 version "_src.tar.gz"))
4020 (sha256
4021 (base32
8c864901 4022 "1wl0dawpssqwfjvr651r4wlww8hhjin8nba6xh71ks7sbypx886j"))))
31a9d653
BW		 4023 (build-system gnu-build-system)
4024 (arguments
4025 `(#:test-target "test"
4026 #:phases
4027 (modify-phases %standard-phases
4028 (replace 'configure
4029 ;; There is no configure script, so we modify the Makefile directly.
4030 (lambda* (#:key outputs #:allow-other-keys)
4031 (substitute* "Makefile"
4032 (("INSTALLDIR=.*")
4033 (string-append
4034 "INSTALLDIR=" (assoc-ref outputs "out") "/bin\n")))
4035 #t))
4036 (add-before 'install 'make-install-directory
4037 ;; The install directory is not created during 'make install'.
4038 (lambda* (#:key outputs #:allow-other-keys)
4039 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
4040 #t))
4041 (add-after 'install 'wrap-programs
4042 (lambda* (#:key inputs outputs #:allow-other-keys)
4043 (let* ((path (getenv "PATH"))
4044 (out (assoc-ref outputs "out"))
4045 (binary (string-append out "/bin/proteinortho5.pl")))
4046 (wrap-program binary `("PATH" ":" prefix (,path))))
4047 #t)))))
4048 (inputs
4049 `(("perl" ,perl)
4050 ("python" ,python-2)
4051 ("blast+" ,blast+)))
4052 (home-page "http://www.bioinf.uni-leipzig.de/Software/proteinortho")
4053 (synopsis "Detect orthologous genes across species")
4054 (description
4055 "Proteinortho is a tool to detect orthologous genes across different
4056species. For doing so, it compares similarities of given gene sequences and
4057clusters them to find significant groups. The algorithm was designed to handle
4058large-scale data and can be applied to hundreds of species at once.")
4059 (license license:gpl2+)))
4060
846e3409
RW		 4061(define-public pyicoteo
4062 (package
4063 (name "pyicoteo")
4064 (version "2.0.7")
4065 (source
4066 (origin
4067 (method url-fetch)
4068 (uri (string-append "https://bitbucket.org/regulatorygenomicsupf/"
4069 "pyicoteo/get/v" version ".tar.bz2"))
4070 (file-name (string-append name "-" version ".tar.bz2"))
4071 (sha256
4072 (base32
4073 "0d6087f29xp8wxwlj111c3sylli98n0l8ry58c51ixzq0zfm50wa"))))
4074 (build-system python-build-system)
4075 (arguments
4076 `(#:python ,python-2 ; does not work with Python 3
4077 #:tests? #f)) ; there are no tests
4078 (inputs
4079 `(("python2-matplotlib" ,python2-matplotlib)))
4080 (home-page "https://bitbucket.org/regulatorygenomicsupf/pyicoteo")
4081 (synopsis "Analyze high-throughput genetic sequencing data")
4082 (description
4083 "Pyicoteo is a suite of tools for the analysis of high-throughput genetic
4084sequencing data. It works with genomic coordinates. There are currently six
4085different command-line tools:
4086
4087@enumerate
4088@item pyicoregion: for generating exploratory regions automatically;
4089@item pyicoenrich: for differential enrichment between two conditions;
4090@item pyicoclip: for calling CLIP-Seq peaks without a control;
4091@item pyicos: for genomic coordinates manipulation;
4092@item pyicoller: for peak calling on punctuated ChIP-Seq;
4093@item pyicount: to count how many reads from N experiment files overlap in a
4094 region file;
4095@item pyicotrocol: to combine operations from pyicoteo.
4096@end enumerate\n")
4097 (license license:gpl3+)))
4098
af860475
BW		 4099(define-public prodigal
4100 (package
4101 (name "prodigal")
e70f7a23 4102 (version "2.6.3")
af860475
BW		 4103 (source (origin
4104 (method url-fetch)
4105 (uri (string-append
4106 "https://github.com/hyattpd/Prodigal/archive/v"
4107 version ".tar.gz"))
4108 (file-name (string-append name "-" version ".tar.gz"))
4109 (sha256
4110 (base32
e70f7a23 4111 "17srxkqd3jc77xk15pfbgg1a9xahqg7337w95mrsia7mpza4l2c9"))))
af860475
BW		 4112 (build-system gnu-build-system)
4113 (arguments
4114 `(#:tests? #f ;no check target
4115 #:make-flags (list (string-append "INSTALLDIR="
4116 (assoc-ref %outputs "out")
4117 "/bin"))
4118 #:phases
4119 (modify-phases %standard-phases
4120 (delete 'configure))))
4121 (home-page "http://prodigal.ornl.gov")
4122 (synopsis "Protein-coding gene prediction for Archaea and Bacteria")
4123 (description
4124 "Prodigal runs smoothly on finished genomes, draft genomes, and
4125metagenomes, providing gene predictions in GFF3, Genbank, or Sequin table
4126format. It runs quickly, in an unsupervised fashion, handles gaps, handles
4127partial genes, and identifies translation initiation sites.")
4128 (license license:gpl3+)))
608dd932 4129
ceb62d54
BW		 4130(define-public roary
4131 (package
4132 (name "roary")
da4a707c 4133 (version "3.11.0")
ceb62d54
BW		 4134 (source
4135 (origin
4136 (method url-fetch)
4137 (uri (string-append
4138 "mirror://cpan/authors/id/A/AJ/AJPAGE/Bio-Roary-"
4139 version ".tar.gz"))
4140 (sha256
4141 (base32
da4a707c 4142 "10lw78x1xzvn7xzvnmh4bm3cak3ah5cssapl0yidvhaj1f44h29i"))))
ceb62d54
BW		 4143 (build-system perl-build-system)
4144 (arguments
4145 `(#:phases
4146 (modify-phases %standard-phases
4147 (delete 'configure)
4148 (delete 'build)
4149 (replace 'check
4150 (lambda _
4151 ;; The tests are not run by default, so we run each test file
4152 ;; directly.
4153 (setenv "PATH" (string-append (getcwd) "/bin" ":"
4154 (getenv "PATH")))
4155 (setenv "PERL5LIB" (string-append (getcwd) "/lib" ":"
4156 (getenv "PERL5LIB")))
4157 (zero? (length (filter (lambda (file)
4158 (display file)(display "\n")
4159 (not (zero? (system* "perl" file))))
4160 (find-files "t" ".*\\.t$"))))))
4161 (replace 'install
4162 ;; There is no 'install' target in the Makefile.
4163 (lambda* (#:key outputs #:allow-other-keys)
4164 (let* ((out (assoc-ref outputs "out"))
4165 (bin (string-append out "/bin"))
4166 (perl (string-append out "/lib/perl5/site_perl"))
4167 (roary-plots "contrib/roary_plots"))
4168 (mkdir-p bin)
4169 (mkdir-p perl)
4170 (copy-recursively "bin" bin)
4171 (copy-recursively "lib" perl)
4172 #t)))
4173 (add-after 'install 'wrap-programs
4174 (lambda* (#:key inputs outputs #:allow-other-keys)
4175 (let* ((out (assoc-ref outputs "out"))
4176 (perl5lib (getenv "PERL5LIB"))
4177 (path (getenv "PATH")))
4178 (for-each (lambda (prog)
4179 (let ((binary (string-append out "/" prog)))
4180 (wrap-program binary
4181 `("PERL5LIB" ":" prefix
4182 (,(string-append perl5lib ":" out
4183 "/lib/perl5/site_perl"))))
4184 (wrap-program binary
4185 `("PATH" ":" prefix
4186 (,(string-append path ":" out "/bin"))))))
4187 (find-files "bin" ".*[^R]$"))
4188 (let ((file
4189 (string-append out "/bin/roary-create_pan_genome_plots.R"))
4190 (r-site-lib (getenv "R_LIBS_SITE"))
4191 (coreutils-path
4192 (string-append (assoc-ref inputs "coreutils") "/bin")))
4193 (wrap-program file
4194 `("R_LIBS_SITE" ":" prefix
4195 (,(string-append r-site-lib ":" out "/site-library/"))))
4196 (wrap-program file
4197 `("PATH" ":" prefix
4198 (,(string-append coreutils-path ":" out "/bin"))))))
4199 #t)))))
4200 (native-inputs
4201 `(("perl-env-path" ,perl-env-path)
4202 ("perl-test-files" ,perl-test-files)
4203 ("perl-test-most" ,perl-test-most)
4204 ("perl-test-output" ,perl-test-output)))
4205 (inputs
4206 `(("perl-array-utils" ,perl-array-utils)
4207 ("bioperl" ,bioperl-minimal)
da4a707c 4208 ("perl-digest-md5-file" ,perl-digest-md5-file)
ceb62d54
BW		 4209 ("perl-exception-class" ,perl-exception-class)
4210 ("perl-file-find-rule" ,perl-file-find-rule)
4211 ("perl-file-grep" ,perl-file-grep)
4212 ("perl-file-slurper" ,perl-file-slurper)
4213 ("perl-file-which" ,perl-file-which)
4214 ("perl-graph" ,perl-graph)
4215 ("perl-graph-readwrite" ,perl-graph-readwrite)
4216 ("perl-log-log4perl" ,perl-log-log4perl)
4217 ("perl-moose" ,perl-moose)
4218 ("perl-perlio-utf8_strict" ,perl-perlio-utf8_strict)
4219 ("perl-text-csv" ,perl-text-csv)
4220 ("bedtools" ,bedtools)
4221 ("cd-hit" ,cd-hit)
4222 ("blast+" ,blast+)
4223 ("mcl" ,mcl)
4224 ("parallel" ,parallel)
4225 ("prank" ,prank)
4226 ("mafft" ,mafft)
4227 ("fasttree" ,fasttree)
4228 ("grep" ,grep)
4229 ("sed" ,sed)
4230 ("gawk" ,gawk)
2d7c4ae3 4231 ("r-minimal" ,r-minimal)
ceb62d54
BW		 4232 ("r-ggplot2" ,r-ggplot2)
4233 ("coreutils" ,coreutils)))
4234 (home-page "http://sanger-pathogens.github.io/Roary")
4235 (synopsis "High speed stand-alone pan genome pipeline")
4236 (description
4237 "Roary is a high speed stand alone pan genome pipeline, which takes
4238annotated assemblies in GFF3 format (produced by the Prokka program) and
4239calculates the pan genome. Using a standard desktop PC, it can analyse
4240datasets with thousands of samples, without compromising the quality of the
4241results. 128 samples can be analysed in under 1 hour using 1 GB of RAM and a
4242single processor. Roary is not intended for metagenomics or for comparing
4243extremely diverse sets of genomes.")
4244 (license license:gpl3)))
4245
608dd932
BW		 4246(define-public raxml
4247 (package
4248 (name "raxml")
e9e0fab0 4249 (version "8.2.10")
608dd932
BW		 4250 (source
4251 (origin
4252 (method url-fetch)
4253 (uri
4254 (string-append
4255 "https://github.com/stamatak/standard-RAxML/archive/v"
4256 version ".tar.gz"))
4257 (file-name (string-append name "-" version ".tar.gz"))
4258 (sha256
4259 (base32
e9e0fab0 4260 "13s7aspfdcfr6asynwdg1x6vznys6pzap5f8wsffbnnwpkkg9ya8"))))
608dd932
BW		 4261 (build-system gnu-build-system)
4262 (arguments
4263 `(#:tests? #f ; There are no tests.
4264 ;; Use 'standard' Makefile rather than SSE or AVX ones.
4265 #:make-flags (list "-f" "Makefile.HYBRID.gcc")
4266 #:phases
4267 (modify-phases %standard-phases
4268 (delete 'configure)
4269 (replace 'install
4270 (lambda* (#:key outputs #:allow-other-keys)
4271 (let* ((out (assoc-ref outputs "out"))
4272 (bin (string-append out "/bin"))
4273 (executable "raxmlHPC-HYBRID"))
4274 (install-file executable bin)
4275 (symlink (string-append bin "/" executable) "raxml"))
4276 #t)))))
4277 (inputs
4278 `(("openmpi" ,openmpi)))
4279 (home-page "http://sco.h-its.org/exelixis/web/software/raxml/index.html")
4280 (synopsis "Randomized Axelerated Maximum Likelihood phylogenetic trees")
4281 (description
4282 "RAxML is a tool for phylogenetic analysis and post-analysis of large
4283phylogenies.")
2805f9b2
EF		 4284 ;; The source includes x86 specific code
4285 (supported-systems '("x86_64-linux" "i686-linux"))
608dd932 4286 (license license:gpl2+)))
af860475 4287
66e3eff1
RW		 4288(define-public rsem
4289 (package
4290 (name "rsem")
4291 (version "1.2.20")
4292 (source
4293 (origin
4294 (method url-fetch)
4295 (uri
4296 (string-append "http://deweylab.biostat.wisc.edu/rsem/src/rsem-"
4297 version ".tar.gz"))
4298 (sha256
4299 (base32 "0nzdc0j0hjllhsd5f2xli95dafm3nawskigs140xzvjk67xh0r9q"))
fc1adab1 4300 (patches (search-patches "rsem-makefile.patch"))
66e3eff1
RW		 4301 (modules '((guix build utils)))
4302 (snippet
4303 '(begin
4304 ;; remove bundled copy of boost
4305 (delete-file-recursively "boost")
4306 #t))))
4307 (build-system gnu-build-system)
4308 (arguments
4309 `(#:tests? #f ;no "check" target
4310 #:phases
4311 (modify-phases %standard-phases
4312 ;; No "configure" script.
4313 ;; Do not build bundled samtools library.
4314 (replace 'configure
4315 (lambda _
4316 (substitute* "Makefile"
4317 (("^all : sam/libbam.a") "all : "))
4318 #t))
4319 (replace 'install
4320 (lambda* (#:key outputs #:allow-other-keys)
4321 (let* ((out (string-append (assoc-ref outputs "out")))
4322 (bin (string-append out "/bin/"))
4323 (perl (string-append out "/lib/perl5/site_perl")))
4324 (mkdir-p bin)
4325 (mkdir-p perl)
4326 (for-each (lambda (file)
f3860753 4327 (install-file file bin))
66e3eff1 4328 (find-files "." "rsem-.*"))
f3860753 4329 (install-file "rsem_perl_utils.pm" perl))
66e3eff1
RW		 4330 #t))
4331 (add-after
4332 'install 'wrap-program
4333 (lambda* (#:key outputs #:allow-other-keys)
4334 (let ((out (assoc-ref outputs "out")))
4335 (for-each (lambda (prog)
4336 (wrap-program (string-append out "/bin/" prog)
4337 `("PERL5LIB" ":" prefix
4338 (,(string-append out "/lib/perl5/site_perl")))))
4339 '("rsem-plot-transcript-wiggles"
4340 "rsem-calculate-expression"
4341 "rsem-generate-ngvector"
4342 "rsem-run-ebseq"
4343 "rsem-prepare-reference")))
4344 #t)))))
4345 (inputs
4346 `(("boost" ,boost)
4347 ("ncurses" ,ncurses)
2d7c4ae3 4348 ("r-minimal" ,r-minimal)
66e3eff1
RW		 4349 ("perl" ,perl)
4350 ("samtools" ,samtools-0.1)
4351 ("zlib" ,zlib)))
4352 (home-page "http://deweylab.biostat.wisc.edu/rsem/")
4353 (synopsis "Estimate gene expression levels from RNA-Seq data")
4354 (description
4355 "RSEM is a software package for estimating gene and isoform expression
4356levels from RNA-Seq data. The RSEM package provides a user-friendly
4357interface, supports threads for parallel computation of the EM algorithm,
4358single-end and paired-end read data, quality scores, variable-length reads and
4359RSPD estimation. In addition, it provides posterior mean and 95% credibility
4360interval estimates for expression levels. For visualization, it can generate
4361BAM and Wiggle files in both transcript-coordinate and genomic-coordinate.")
4362 (license license:gpl3+)))
4363
8622a072
RW		 4364(define-public rseqc
4365 (package
4366 (name "rseqc")
4367 (version "2.6.1")
4368 (source
4369 (origin
4370 (method url-fetch)
4371 (uri
4372 (string-append "mirror://sourceforge/rseqc/"
de67e922 4373 "RSeQC-" version ".tar.gz"))
8622a072 4374 (sha256
8214b7fb 4375 (base32 "15ly0254yi032qzkdplg00q144qfdsd986gh62829rl5bkxhj330"))
8622a072
RW		 4376 (modules '((guix build utils)))
4377 (snippet
4378 '(begin
4379 ;; remove bundled copy of pysam
4380 (delete-file-recursively "lib/pysam")
4381 (substitute* "setup.py"
4382 ;; remove dependency on outdated "distribute" module
4383 (("^from distribute_setup import use_setuptools") "")
4384 (("^use_setuptools\\(\\)") "")
4385 ;; do not use bundled copy of pysam
4386 (("^have_pysam = False") "have_pysam = True"))))))
4387 (build-system python-build-system)
4388 (arguments `(#:python ,python-2))
4389 (inputs
4390 `(("python-cython" ,python2-cython)
4391 ("python-pysam" ,python2-pysam)
4392 ("python-numpy" ,python2-numpy)
8622a072
RW		 4393 ("zlib" ,zlib)))
4394 (native-inputs
4395 `(("python-nose" ,python2-nose)))
4396 (home-page "http://rseqc.sourceforge.net/")
4397 (synopsis "RNA-seq quality control package")
4398 (description
4399 "RSeQC provides a number of modules that can comprehensively evaluate
4400high throughput sequence data, especially RNA-seq data. Some basic modules
4401inspect sequence quality, nucleotide composition bias, PCR bias and GC bias,
4402while RNA-seq specific modules evaluate sequencing saturation, mapped reads
4403distribution, coverage uniformity, strand specificity, etc.")
4404 (license license:gpl3+)))
4405
ec946638
RW		 4406(define-public seek
4407 ;; There are no release tarballs. According to the installation
4408 ;; instructions at http://seek.princeton.edu/installation.jsp, the latest
4409 ;; stable release is identified by this changeset ID.
4410 (let ((changeset "2329130")
4411 (revision "1"))
4412 (package
4413 (name "seek")
4414 (version (string-append "0-" revision "." changeset))
4415 (source (origin
4416 (method hg-fetch)
4417 (uri (hg-reference
4418 (url "https://bitbucket.org/libsleipnir/sleipnir")
4419 (changeset changeset)))
4420 (sha256
4421 (base32
4422 "0qrvilwh18dpbhkf92qvxbmay0j75ra3jg2wrhz67gf538zzphsx"))))
4423 (build-system gnu-build-system)
4424 (arguments
4425 `(#:modules ((srfi srfi-1)
4426 (guix build gnu-build-system)
4427 (guix build utils))
4428 #:phases
4429 (let ((dirs '("SeekMiner"
4430 "SeekEvaluator"
4431 "SeekPrep"
4432 "Distancer"
4433 "Data2DB"
4434 "PCL2Bin")))
4435 (modify-phases %standard-phases
4561498e 4436 (add-before 'configure 'bootstrap
ec946638
RW		 4437 (lambda _
4438 (zero? (system* "bash" "gen_auto"))))
4439 (add-after 'build 'build-additional-tools
4440 (lambda* (#:key make-flags #:allow-other-keys)
4441 (every (lambda (dir)
4442 (with-directory-excursion (string-append "tools/" dir)
4443 (zero? (apply system* "make" make-flags))))
4444 dirs)))
4445 (add-after 'install 'install-additional-tools
4446 (lambda* (#:key make-flags #:allow-other-keys)
4447 (fold (lambda (dir result)
4448 (with-directory-excursion (string-append "tools/" dir)
4449 (and result
4450 (zero? (apply system*
4451 `("make" ,@make-flags "install"))))))
4452 #t dirs)))))))
4453 (inputs
4454 `(("gsl" ,gsl)
4455 ("boost" ,boost)
4456 ("libsvm" ,libsvm)
4457 ("readline" ,readline)
4458 ("gengetopt" ,gengetopt)
4459 ("log4cpp" ,log4cpp)))
4460 (native-inputs
4461 `(("autoconf" ,autoconf)
4462 ("automake" ,automake)
4463 ("perl" ,perl)))
4464 (home-page "http://seek.princeton.edu")
4465 (synopsis "Gene co-expression search engine")
4466 (description
4467 "SEEK is a computational gene co-expression search engine. SEEK provides
4468biologists with a way to navigate the massive human expression compendium that
4469now contains thousands of expression datasets. SEEK returns a robust ranking
4470of co-expressed genes in the biological area of interest defined by the user's
4471query genes. It also prioritizes thousands of expression datasets according
4472to the user's query of interest.")
4473 (license license:cc-by3.0))))
4474
4e10a221
RW		 4475(define-public samtools
4476 (package
4477 (name "samtools")
63fa0ffc 4478 (version "1.5")
4e10a221
RW		 4479 (source
4480 (origin
4481 (method url-fetch)
4482 (uri
de67e922 4483 (string-append "mirror://sourceforge/samtools/samtools/"
4e10a221
RW		 4484 version "/samtools-" version ".tar.bz2"))
4485 (sha256
4486 (base32
63fa0ffc 4487 "1xidmv0jmfy7l0kb32hdnlshcxgzi1hmygvig0cqrq1fhckdlhl5"))))
4e10a221
RW		 4488 (build-system gnu-build-system)
4489 (arguments
c4473411 4490 `(#:modules ((ice-9 ftw)
5bdda30b
RW		 4491 (ice-9 regex)
4492 (guix build gnu-build-system)
4493 (guix build utils))
c4473411 4494 #:make-flags (list (string-append "prefix=" (assoc-ref %outputs "out")))
4ab16440 4495 #:configure-flags (list "--with-ncurses" "--with-htslib=system")
4e10a221 4496 #:phases
c842d425
BW		 4497 (modify-phases %standard-phases
4498 (add-after 'unpack 'patch-tests
4499 (lambda _
4500 (substitute* "test/test.pl"
4501 ;; The test script calls out to /bin/bash
4502 (("/bin/bash") (which "bash")))
b2d748ed 4503 #t))
c842d425
BW		 4504 (add-after 'install 'install-library
4505 (lambda* (#:key outputs #:allow-other-keys)
4506 (let ((lib (string-append (assoc-ref outputs "out") "/lib")))
4507 (install-file "libbam.a" lib)
4508 #t)))
4509 (add-after 'install 'install-headers
4510 (lambda* (#:key outputs #:allow-other-keys)
4511 (let ((include (string-append (assoc-ref outputs "out")
4512 "/include/samtools/")))
4513 (for-each (lambda (file)
4514 (install-file file include))
4515 (scandir "." (lambda (name) (string-match "\\.h$" name))))
4516 #t))))))
4e10a221 4517 (native-inputs `(("pkg-config" ,pkg-config)))
4ab16440
BW		 4518 (inputs
4519 `(("htslib" ,htslib)
4520 ("ncurses" ,ncurses)
4521 ("perl" ,perl)
4522 ("python" ,python)
4523 ("zlib" ,zlib)))
4e10a221
RW		 4524 (home-page "http://samtools.sourceforge.net")
4525 (synopsis "Utilities to efficiently manipulate nucleotide sequence alignments")
4526 (description
4527 "Samtools implements various utilities for post-processing nucleotide
4528sequence alignments in the SAM, BAM, and CRAM formats, including indexing,
4529variant calling (in conjunction with bcftools), and a simple alignment
4530viewer.")
4531 (license license:expat)))
d3517eda 4532
0b84a0aa
RW		 4533(define-public samtools-0.1
4534 ;; This is the most recent version of the 0.1 line of samtools. The input
4535 ;; and output formats differ greatly from that used and produced by samtools
4536 ;; 1.x and is still used in many bioinformatics pipelines.
4537 (package (inherit samtools)
4538 (version "0.1.19")
4539 (source
4540 (origin
4541 (method url-fetch)
4542 (uri
de67e922 4543 (string-append "mirror://sourceforge/samtools/samtools/"
0b84a0aa
RW		 4544 version "/samtools-" version ".tar.bz2"))
4545 (sha256
4546 (base32 "1m33xsfwz0s8qi45lylagfllqg7fphf4dr0780rsvw75av9wk06h"))))
4547 (arguments
2309ed68
RW		 4548 `(#:tests? #f ;no "check" target
4549 ,@(substitute-keyword-arguments (package-arguments samtools)
4550 ((#:make-flags flags)
4551 `(cons "LIBCURSES=-lncurses" ,flags))
4552 ((#:phases phases)
4553 `(modify-phases ,phases
4554 (replace 'install
4555 (lambda* (#:key outputs #:allow-other-keys)
4556 (let ((bin (string-append
4557 (assoc-ref outputs "out") "/bin")))
4558 (mkdir-p bin)
f3860753 4559 (install-file "samtools" bin)
b2d748ed 4560 #t)))
2309ed68
RW		 4561 (delete 'patch-tests)
4562 (delete 'configure))))))))
0b84a0aa 4563
fe4c37c2 4564(define-public mosaik
698bd297 4565 (let ((commit "5c25216d3522d6a33e53875cd76a6d65001e4e67"))
fe4c37c2
RW		 4566 (package
4567 (name "mosaik")
4568 (version "2.2.30")
4569 (source (origin
4570 ;; There are no release tarballs nor tags.
4571 (method git-fetch)
4572 (uri (git-reference
4573 (url "https://github.com/wanpinglee/MOSAIK.git")
4574 (commit commit)))
4575 (file-name (string-append name "-" version))
4576 (sha256
4577 (base32
4578 "17gj3s07cm77r41z92awh0bim7w7q7fbn0sf5nkqmcm1vw052qgw"))))
4579 (build-system gnu-build-system)
4580 (arguments
4581 `(#:tests? #f ; no tests
4582 #:make-flags (list "CC=gcc")
4583 #:phases
4584 (modify-phases %standard-phases
4585 (replace 'configure
4586 (lambda _ (chdir "src") #t))
4587 (replace 'install
4588 (lambda* (#:key outputs #:allow-other-keys)
4589 (let ((bin (string-append (assoc-ref outputs "out")
4590 "/bin")))
4591 (mkdir-p bin)
4592 (copy-recursively "../bin" bin)
4593 #t))))))
4594 (inputs
4595 `(("perl" ,perl)
4596 ("zlib" ,zlib)))
029d9f77 4597 (supported-systems '("x86_64-linux"))
0c6c9c00 4598 (home-page "https://github.com/wanpinglee/MOSAIK")
fe4c37c2
RW		 4599 (synopsis "Map nucleotide sequence reads to reference genomes")
4600 (description
4601 "MOSAIK is a program for mapping second and third-generation sequencing
4602reads to a reference genome. MOSAIK can align reads generated by all the
4603major sequencing technologies, including Illumina, Applied Biosystems SOLiD,
4604Roche 454, Ion Torrent and Pacific BioSciences SMRT.")
4605 ;; MOSAIK is released under the GPLv2+ with the exception of third-party
4606 ;; code released into the public domain:
4607 ;; 1. fastlz by Ariya Hidayat - http://www.fastlz.org/
4608 ;; 2. MD5 implementation - RSA Data Security, RFC 1321
4609 (license (list license:gpl2+ license:public-domain)))))
4610
282c5087
RW		 4611(define-public ngs-sdk
4612 (package
4613 (name "ngs-sdk")
6c4ccf32 4614 (version "1.3.0")
282c5087
RW		 4615 (source
4616 (origin
4617 (method url-fetch)
4618 (uri
4619 (string-append "https://github.com/ncbi/ngs/archive/"
4620 version ".tar.gz"))
4621 (file-name (string-append name "-" version ".tar.gz"))
4622 (sha256
4623 (base32
6c4ccf32 4624 "1wiyf4c6nm2j87pv015cbi0qny5byf3pbvcw3likifz5dl56ag40"))))
282c5087
RW		 4625 (build-system gnu-build-system)
4626 (arguments
4627 `(#:parallel-build? #f ; not supported
4628 #:tests? #f ; no "check" target
4629 #:phases
dc1d3cde
KK		 4630 (modify-phases %standard-phases
4631 (replace 'configure
4632 (lambda* (#:key outputs #:allow-other-keys)
4633 (let ((out (assoc-ref outputs "out")))
4634 ;; Allow 'konfigure.perl' to find 'package.prl'.
4635 (setenv "PERL5LIB"
4636 (string-append ".:" (getenv "PERL5LIB")))
4637
4638 ;; The 'configure' script doesn't recognize things like
4639 ;; '--enable-fast-install'.
4640 (zero? (system* "./configure"
4641 (string-append "--build-prefix=" (getcwd) "/build")
4642 (string-append "--prefix=" out))))))
4643 (add-after 'unpack 'enter-dir
4644 (lambda _ (chdir "ngs-sdk") #t)))))
282c5087 4645 (native-inputs `(("perl" ,perl)))
a0dadf0c
AE		 4646 ;; According to the test
4647 ;; unless ($MARCH =~ /x86_64/i || $MARCH =~ /i?86/i)
4648 ;; in ngs-sdk/setup/konfigure.perl
ab29be81 4649 (supported-systems '("i686-linux" "x86_64-linux"))
282c5087
RW		 4650 (home-page "https://github.com/ncbi/ngs")
4651 (synopsis "API for accessing Next Generation Sequencing data")
4652 (description
4653 "NGS is a domain-specific API for accessing reads, alignments and pileups
4654produced from Next Generation Sequencing. The API itself is independent from
4655any particular back-end implementation, and supports use of multiple back-ends
4656simultaneously.")
4657 (license license:public-domain)))
4658
1ad15c16 4659(define-public java-ngs
2651a5e6 4660 (package (inherit ngs-sdk)
1ad15c16 4661 (name "java-ngs")
2651a5e6
RW		 4662 (arguments
4663 `(,@(substitute-keyword-arguments
4664 `(#:modules ((guix build gnu-build-system)
4665 (guix build utils)
4666 (srfi srfi-1)
4667 (srfi srfi-26))
4668 ,@(package-arguments ngs-sdk))
4669 ((#:phases phases)
614a8977
RW		 4670 `(modify-phases ,phases
4671 (replace 'enter-dir (lambda _ (chdir "ngs-java") #t)))))))
2651a5e6 4672 (inputs
d2540f80 4673 `(("jdk" ,icedtea "jdk")
2651a5e6
RW		 4674 ("ngs-sdk" ,ngs-sdk)))
4675 (synopsis "Java bindings for NGS SDK")))
4676
75dd2424
RW		 4677(define-public ncbi-vdb
4678 (package
4679 (name "ncbi-vdb")
5021f547 4680 (version "2.8.2")
75dd2424
RW		 4681 (source
4682 (origin
4683 (method url-fetch)
4684 (uri
4685 (string-append "https://github.com/ncbi/ncbi-vdb/archive/"
4686 version ".tar.gz"))
4687 (file-name (string-append name "-" version ".tar.gz"))
4688 (sha256
4689 (base32
5021f547 4690 "1acn4bv81mfl137qnbn9995mjjhwd36pm0b7qli1iw5skrxa9j8m"))))
75dd2424
RW		 4691 (build-system gnu-build-system)
4692 (arguments
4693 `(#:parallel-build? #f ; not supported
4694 #:tests? #f ; no "check" target
4695 #:phases
70111198 4696 (modify-phases %standard-phases
0691851a
BW		 4697 (add-before 'configure 'set-perl-search-path
4698 (lambda _
4699 ;; Work around "dotless @INC" build failure.
4700 (setenv "PERL5LIB"
4701 (string-append (getcwd) "/setup:"
4702 (getenv "PERL5LIB")))
4703 #t))
70111198
RW		 4704 (replace 'configure
4705 (lambda* (#:key inputs outputs #:allow-other-keys)
4706 (let ((out (assoc-ref outputs "out")))
4707 ;; Override include path for libmagic
4708 (substitute* "setup/package.prl"
4709 (("name => 'magic', Include => '/usr/include'")
4710 (string-append "name=> 'magic', Include => '"
4711 (assoc-ref inputs "libmagic")
4712 "/include" "'")))
75dd2424 4713
70111198
RW		 4714 ;; Install kdf5 library (needed by sra-tools)
4715 (substitute* "build/Makefile.install"
4716 (("LIBRARIES_TO_INSTALL =")
4717 "LIBRARIES_TO_INSTALL = kdf5.$(VERSION_LIBX) kdf5.$(VERSION_SHLX)"))
75dd2424 4718
70111198
RW		 4719 (substitute* "build/Makefile.env"
4720 (("CFLAGS =" prefix)
4721 (string-append prefix "-msse2 ")))
675d7ae2 4722
558e2307
RW		 4723 ;; Override search path for ngs-java
4724 (substitute* "setup/package.prl"
4725 (("/usr/local/ngs/ngs-java")
4726 (assoc-ref inputs "java-ngs")))
4727
70111198
RW		 4728 ;; The 'configure' script doesn't recognize things like
4729 ;; '--enable-fast-install'.
4730 (zero? (system*
4731 "./configure"
4732 (string-append "--build-prefix=" (getcwd) "/build")
4733 (string-append "--prefix=" (assoc-ref outputs "out"))
4734 (string-append "--debug")
4735 (string-append "--with-xml2-prefix="
4736 (assoc-ref inputs "libxml2"))
4737 (string-append "--with-ngs-sdk-prefix="
4738 (assoc-ref inputs "ngs-sdk"))
70111198
RW		 4739 (string-append "--with-hdf5-prefix="
4740 (assoc-ref inputs "hdf5")))))))
4741 (add-after 'install 'install-interfaces
4742 (lambda* (#:key outputs #:allow-other-keys)
4743 ;; Install interface libraries. On i686 the interface libraries
4744 ;; are installed to "linux/gcc/i386", so we need to use the Linux
4745 ;; architecture name ("i386") instead of the target system prefix
4746 ;; ("i686").
4747 (mkdir (string-append (assoc-ref outputs "out") "/ilib"))
4748 (copy-recursively (string-append "build/ncbi-vdb/linux/gcc/"
4749 ,(system->linux-architecture
4750 (or (%current-target-system)
4751 (%current-system)))
4752 "/rel/ilib")
4753 (string-append (assoc-ref outputs "out")
4754 "/ilib"))
4755 ;; Install interface headers
4756 (copy-recursively "interfaces"
4757 (string-append (assoc-ref outputs "out")
4758 "/include"))
d5e17162
RW		 4759 #t))
4760 ;; These files are needed by sra-tools.
4761 (add-after 'install 'install-configuration-files
4762 (lambda* (#:key outputs #:allow-other-keys)
4763 (let ((target (string-append (assoc-ref outputs "out") "/kfg")))
4764 (mkdir target)
4765 (install-file "libs/kfg/default.kfg" target)
4766 (install-file "libs/kfg/certs.kfg" target))
70111198 4767 #t)))))
75dd2424
RW		 4768 (inputs
4769 `(("libxml2" ,libxml2)
4770 ("ngs-sdk" ,ngs-sdk)
1ad15c16 4771 ("java-ngs" ,java-ngs)
75dd2424
RW		 4772 ("libmagic" ,file)
4773 ("hdf5" ,hdf5)))
4774 (native-inputs `(("perl" ,perl)))
675d7ae2
RW		 4775 ;; NCBI-VDB requires SSE capability.
4776 (supported-systems '("i686-linux" "x86_64-linux"))
75dd2424
RW		 4777 (home-page "https://github.com/ncbi/ncbi-vdb")
4778 (synopsis "Database engine for genetic information")
4779 (description
4780 "The NCBI-VDB library implements a highly compressed columnar data
4781warehousing engine that is most often used to store genetic information.
4782Databases are stored in a portable image within the file system, and can be
4783accessed/downloaded on demand across HTTP.")
4784 (license license:public-domain)))
4785
cc6ed477
RW		 4786(define-public plink
4787 (package
4788 (name "plink")
4789 (version "1.07")
4790 (source
4791 (origin
4792 (method url-fetch)
4793 (uri (string-append
4794 "http://pngu.mgh.harvard.edu/~purcell/plink/dist/plink-"
4795 version "-src.zip"))
4796 (sha256
4797 (base32 "0as8gxm4pjyc8dxmm1sl873rrd7wn5qs0l29nqfnl31x8i467xaa"))
0dbb7ac2
EF		 4798 (patches (search-patches "plink-1.07-unclobber-i.patch"
4799 "plink-endian-detection.patch"))))
cc6ed477
RW		 4800 (build-system gnu-build-system)
4801 (arguments
4802 '(#:tests? #f ;no "check" target
4803 #:make-flags (list (string-append "LIB_LAPACK="
4804 (assoc-ref %build-inputs "lapack")
4805 "/lib/liblapack.so")
4806 "WITH_LAPACK=1"
4807 "FORCE_DYNAMIC=1"
4808 ;; disable phoning home
4809 "WITH_WEBCHECK=")
4810 #:phases
4811 (modify-phases %standard-phases
4812 ;; no "configure" script
4813 (delete 'configure)
4814 (replace 'install
4815 (lambda* (#:key outputs #:allow-other-keys)
4816 (let ((bin (string-append (assoc-ref outputs "out")
4817 "/bin/")))
96c46210 4818 (install-file "plink" bin)
cc6ed477
RW		 4819 #t))))))
4820 (inputs
4821 `(("zlib" ,zlib)
4822 ("lapack" ,lapack)))
4823 (native-inputs
4824 `(("unzip" ,unzip)))
4825 (home-page "http://pngu.mgh.harvard.edu/~purcell/plink/")
4826 (synopsis "Whole genome association analysis toolset")
4827 (description
4828 "PLINK is a whole genome association analysis toolset, designed to
4829perform a range of basic, large-scale analyses in a computationally efficient
4830manner. The focus of PLINK is purely on analysis of genotype/phenotype data,
4831so there is no support for steps prior to this (e.g. study design and
4832planning, generating genotype or CNV calls from raw data). Through
4833integration with gPLINK and Haploview, there is some support for the
4834subsequent visualization, annotation and storage of results.")
4835 ;; Code is released under GPLv2, except for fisher.h, which is under
4836 ;; LGPLv2.1+
4837 (license (list license:gpl2 license:lgpl2.1+))))
4838
dad66da4
RW		 4839(define-public plink-ng
4840 (package (inherit plink)
4841 (name "plink-ng")
4842 (version "1.90b4")
4843 (source
4844 (origin
4845 (method url-fetch)
4846 (uri (string-append "https://github.com/chrchang/plink-ng/archive/v"
4847 version ".tar.gz"))
4848 (file-name (string-append name "-" version ".tar.gz"))
4849 (sha256
4850 (base32 "09ixrds009aczjswxr2alcb774mksq5g0v78dgjjn1h4dky0kf9a"))))
4851 (build-system gnu-build-system)
4852 (arguments
4853 '(#:tests? #f ;no "check" target
4854 #:make-flags (list "BLASFLAGS=-llapack -lopenblas"
4855 "CFLAGS=-Wall -O2 -DDYNAMIC_ZLIB=1"
4856 "ZLIB=-lz"
4857 "-f" "Makefile.std")
4858 #:phases
4859 (modify-phases %standard-phases
4860 (add-after 'unpack 'chdir
4861 (lambda _ (chdir "1.9") #t))
4862 (delete 'configure) ; no "configure" script
4863 (replace 'install
4864 (lambda* (#:key outputs #:allow-other-keys)
4865 (let ((bin (string-append (assoc-ref outputs "out")
4866 "/bin/")))
4867 (install-file "plink" bin)
4868 #t))))))
4869 (inputs
4870 `(("zlib" ,zlib)
4871 ("lapack" ,lapack)
4872 ("openblas" ,openblas)))
4873 (home-page "https://www.cog-genomics.org/plink/")
4874 (license license:gpl3+)))
4875
c6a24d6e
RW		 4876(define-public smithlab-cpp
4877 (let ((revision "1")
698bd297 4878 (commit "728a097bec88c6f4b8528b685932049e660eff2e"))
c6a24d6e
RW		 4879 (package
4880 (name "smithlab-cpp")
698bd297 4881 (version (string-append "0." revision "." (string-take commit 7)))
c6a24d6e
RW		 4882 (source (origin
4883 (method git-fetch)
4884 (uri (git-reference
4885 (url "https://github.com/smithlabcode/smithlab_cpp.git")
4886 (commit commit)))
4887 (file-name (string-append name "-" version "-checkout"))
4888 (sha256
4889 (base32
4890 "0d476lmj312xk77kr9fzrv7z1bv96yfyx0w7y62ycmnfbx32ll74"))))
4891 (build-system gnu-build-system)
4892 (arguments
4893 `(#:modules ((guix build gnu-build-system)
4894 (guix build utils)
4895 (srfi srfi-26))
4896 #:tests? #f ;no "check" target
4897 #:phases
4898 (modify-phases %standard-phases
4899 (add-after 'unpack 'use-samtools-headers
4900 (lambda _
4901 (substitute* '("SAM.cpp"
4902 "SAM.hpp")
4903 (("sam.h") "samtools/sam.h"))
4904 #t))
4905 (replace 'install
4906 (lambda* (#:key outputs #:allow-other-keys)
4907 (let* ((out (assoc-ref outputs "out"))
4908 (lib (string-append out "/lib"))
4909 (include (string-append out "/include/smithlab-cpp")))
4910 (mkdir-p lib)
4911 (mkdir-p include)
4912 (for-each (cut install-file <> lib)
4913 (find-files "." "\\.o$"))
4914 (for-each (cut install-file <> include)
4915 (find-files "." "\\.hpp$")))
4916 #t))
4917 (delete 'configure))))
4918 (inputs
4919 `(("samtools" ,samtools-0.1)
4920 ("zlib" ,zlib)))
4921 (home-page "https://github.com/smithlabcode/smithlab_cpp")
4922 (synopsis "C++ helper library for functions used in Smith lab projects")
4923 (description
4924 "Smithlab CPP is a C++ library that includes functions used in many of
4925the Smith lab bioinformatics projects, such as a wrapper around Samtools data
4926structures, classes for genomic regions, mapped sequencing reads, etc.")
4927 (license license:gpl3+))))
4928
56e373ef
RW		 4929(define-public preseq
4930 (package
4931 (name "preseq")
b49c5a58 4932 (version "2.0")
56e373ef
RW		 4933 (source (origin
4934 (method url-fetch)
b49c5a58
RW		 4935 (uri (string-append "https://github.com/smithlabcode/"
4936 "preseq/archive/v" version ".tar.gz"))
4937 (file-name (string-append name "-" version ".tar.gz"))
56e373ef 4938 (sha256
b49c5a58 4939 (base32 "08r684l50pnxjpvmhzjgqq56yv9rfw90k8vx0nsrnrzk8mf9hsdq"))
56e373ef
RW		 4940 (modules '((guix build utils)))
4941 (snippet
4942 ;; Remove bundled samtools.
b49c5a58 4943 '(delete-file-recursively "samtools"))))
56e373ef
RW		 4944 (build-system gnu-build-system)
4945 (arguments
4946 `(#:tests? #f ;no "check" target
4947 #:phases
4948 (modify-phases %standard-phases
56e373ef 4949 (delete 'configure))
b49c5a58
RW		 4950 #:make-flags
4951 (list (string-append "PREFIX="
4952 (assoc-ref %outputs "out"))
4953 (string-append "LIBBAM="
4954 (assoc-ref %build-inputs "samtools")
4955 "/lib/libbam.a")
4956 (string-append "SMITHLAB_CPP="
4957 (assoc-ref %build-inputs "smithlab-cpp")
4958 "/lib")
4959 "PROGS=preseq"
4960 "INCLUDEDIRS=$(SMITHLAB_CPP)/../include/smithlab-cpp $(SAMTOOLS_DIR)")))
56e373ef
RW		 4961 (inputs
4962 `(("gsl" ,gsl)
4963 ("samtools" ,samtools-0.1)
b49c5a58 4964 ("smithlab-cpp" ,smithlab-cpp)
56e373ef
RW		 4965 ("zlib" ,zlib)))
4966 (home-page "http://smithlabresearch.org/software/preseq/")
4967 (synopsis "Program for analyzing library complexity")
4968 (description
4969 "The preseq package is aimed at predicting and estimating the complexity
4970of a genomic sequencing library, equivalent to predicting and estimating the
4971number of redundant reads from a given sequencing depth and how many will be
4972expected from additional sequencing using an initial sequencing experiment.
4973The estimates can then be used to examine the utility of further sequencing,
4974optimize the sequencing depth, or to screen multiple libraries to avoid low
4975complexity samples.")
4976 (license license:gpl3+)))
4977
9ded1457
BW		 4978(define-public python-screed
4979 (package
4980 (name "python-screed")
4981 (version "0.9")
4982 (source
4983 (origin
4984 (method url-fetch)
4985 (uri (pypi-uri "screed" version))
4986 (sha256
4987 (base32
4988 "18czszp9fkx3j6jr7y5kp6dfialscgddk05mw1zkhh2zhn0jd8i0"))))
4989 (build-system python-build-system)
4990 (arguments
4991 `(#:phases
4992 (modify-phases %standard-phases
4993 (replace 'check
4994 (lambda _
4995 (setenv "PYTHONPATH"
4996 (string-append (getenv "PYTHONPATH") ":."))
4997 (zero? (system* "nosetests" "--attr" "!known_failing")))))))
4998 (native-inputs
4999 `(("python-nose" ,python-nose)))
5000 (inputs
5001 `(("python-bz2file" ,python-bz2file)))
7bf837fd 5002 (home-page "https://github.com/dib-lab/screed/")
9ded1457
BW		 5003 (synopsis "Short read sequence database utilities")
5004 (description "Screed parses FASTA and FASTQ files and generates databases.
5005Values such as sequence name, sequence description, sequence quality and the
5006sequence itself can be retrieved from these databases.")
5007 (license license:bsd-3)))
5008
5009(define-public python2-screed
5c31f4aa 5010 (package-with-python2 python-screed))
9ded1457 5011
51c64999
RW		 5012(define-public sra-tools
5013 (package
5014 (name "sra-tools")
79849358 5015 (version "2.8.2-1")
51c64999
RW		 5016 (source
5017 (origin
5018 (method url-fetch)
5019 (uri
5020 (string-append "https://github.com/ncbi/sra-tools/archive/"
5021 version ".tar.gz"))
5022 (file-name (string-append name "-" version ".tar.gz"))
5023 (sha256
5024 (base32
79849358 5025 "1camsijmvv2s45mb4iyf44ghl4gkd4rl0viphpcgl3ccchy32a0g"))))
51c64999
RW		 5026 (build-system gnu-build-system)
5027 (arguments
5028 `(#:parallel-build? #f ; not supported
5029 #:tests? #f ; no "check" target
2320e76b 5030 #:make-flags
79849358
RW		 5031 (list (string-append "DEFAULT_CRT="
5032 (assoc-ref %build-inputs "ncbi-vdb")
5033 "/kfg/certs.kfg")
5034 (string-append "DEFAULT_KFG="
5035 (assoc-ref %build-inputs "ncbi-vdb")
5036 "/kfg/default.kfg")
5037 (string-append "VDB_LIBDIR="
2320e76b
RW		 5038 (assoc-ref %build-inputs "ncbi-vdb")
5039 ,(if (string-prefix? "x86_64"
5040 (or (%current-target-system)
5041 (%current-system)))
5042 "/lib64"
5043 "/lib32")))
51c64999 5044 #:phases
beebe431 5045 (modify-phases %standard-phases
0691851a
BW		 5046 (add-before 'configure 'set-perl-search-path
5047 (lambda _
5048 ;; Work around "dotless @INC" build failure.
5049 (setenv "PERL5LIB"
5050 (string-append (getcwd) "/setup:"
5051 (getenv "PERL5LIB")))
5052 #t))
beebe431
RW		 5053 (replace 'configure
5054 (lambda* (#:key inputs outputs #:allow-other-keys)
5055 ;; The build system expects a directory containing the sources and
5056 ;; raw build output of ncbi-vdb, including files that are not
5057 ;; installed. Since we are building against an installed version of
5058 ;; ncbi-vdb, the following modifications are needed.
5059 (substitute* "setup/konfigure.perl"
5060 ;; Make the configure script look for the "ilib" directory of
5061 ;; "ncbi-vdb" without first checking for the existence of a
5062 ;; matching library in its "lib" directory.
5063 (("^ my \\$f = File::Spec->catdir\\(\\$libdir, \\$lib\\);")
5064 "my $f = File::Spec->catdir($ilibdir, $ilib);")
5065 ;; Look for interface libraries in ncbi-vdb's "ilib" directory.
5066 (("my \\$ilibdir = File::Spec->catdir\\(\\$builddir, 'ilib'\\);")
5067 "my $ilibdir = File::Spec->catdir($dir, 'ilib');"))
51c64999 5068
beebe431
RW		 5069 ;; Dynamic linking
5070 (substitute* "tools/copycat/Makefile"
5071 (("smagic-static") "lmagic"))
2320e76b 5072
beebe431
RW		 5073 ;; The 'configure' script doesn't recognize things like
5074 ;; '--enable-fast-install'.
5075 (zero? (system*
5076 "./configure"
5077 (string-append "--build-prefix=" (getcwd) "/build")
5078 (string-append "--prefix=" (assoc-ref outputs "out"))
5079 (string-append "--debug")
5080 (string-append "--with-fuse-prefix="
5081 (assoc-ref inputs "fuse"))
5082 (string-append "--with-magic-prefix="
5083 (assoc-ref inputs "libmagic"))
5084 ;; TODO: building with libxml2 fails with linker errors
5085 ;; (string-append "--with-xml2-prefix="
5086 ;; (assoc-ref inputs "libxml2"))
5087 (string-append "--with-ncbi-vdb-sources="
5088 (assoc-ref inputs "ncbi-vdb"))
5089 (string-append "--with-ncbi-vdb-build="
5090 (assoc-ref inputs "ncbi-vdb"))
5091 (string-append "--with-ngs-sdk-prefix="
5092 (assoc-ref inputs "ngs-sdk"))
5093 (string-append "--with-hdf5-prefix="
1108a920
RW		 5094 (assoc-ref inputs "hdf5"))))))
5095 ;; This version of sra-tools fails to build with glibc because of a
5096 ;; naming conflict. glibc-2.25/include/bits/mathcalls.h already
5097 ;; contains a definition of "canonicalize", so we rename it.
5098 ;;
5099 ;; See upstream bug report:
5100 ;; https://github.com/ncbi/sra-tools/issues/67
5101 (add-after 'unpack 'patch-away-glibc-conflict
5102 (lambda _
5103 (substitute* "tools/bam-loader/bam.c"
5104 (("canonicalize\\(" line)
5105 (string-append "sra_tools_" line)))
5106 #t)))))
51c64999
RW		 5107 (native-inputs `(("perl" ,perl)))
5108 (inputs
5109 `(("ngs-sdk" ,ngs-sdk)
5110 ("ncbi-vdb" ,ncbi-vdb)
5111 ("libmagic" ,file)
5112 ("fuse" ,fuse)
5113 ("hdf5" ,hdf5)
5114 ("zlib" ,zlib)))
5115 (home-page "http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software")
5116 (synopsis "Tools and libraries for reading and writing sequencing data")
5117 (description
5118 "The SRA Toolkit from NCBI is a collection of tools and libraries for
5119reading of sequencing files from the Sequence Read Archive (SRA) database and
5120writing files into the .sra format.")
5121 (license license:public-domain)))
5122
d3517eda
RW		 5123(define-public seqan
5124 (package
5125 (name "seqan")
5126 (version "1.4.2")
5127 (source (origin
5128 (method url-fetch)
5129 (uri (string-append "http://packages.seqan.de/seqan-library/"
5130 "seqan-library-" version ".tar.bz2"))
5131 (sha256
5132 (base32
5133 "05s3wrrwn50f81aklfm65i4a749zag1vr8z03k21xm0pdxy47yvp"))))
5134 ;; The documentation is 7.8MB and the includes are 3.6MB heavy, so it
5135 ;; makes sense to split the outputs.
5136 (outputs '("out" "doc"))
5137 (build-system trivial-build-system)
5138 (arguments
5139 `(#:modules ((guix build utils))
5140 #:builder
5141 (begin
5142 (use-modules (guix build utils))
5143 (let ((tar (assoc-ref %build-inputs "tar"))
5144 (bzip (assoc-ref %build-inputs "bzip2"))
5145 (out (assoc-ref %outputs "out"))
5146 (doc (assoc-ref %outputs "doc")))
5147 (setenv "PATH" (string-append tar "/bin:" bzip "/bin"))
5148 (system* "tar" "xvf" (assoc-ref %build-inputs "source"))
5149 (chdir (string-append "seqan-library-" ,version))
5150 (copy-recursively "include" (string-append out "/include"))
5151 (copy-recursively "share" (string-append doc "/share"))))))
5152 (native-inputs
5153 `(("source" ,source)
5154 ("tar" ,tar)
5155 ("bzip2" ,bzip2)))
5156 (home-page "http://www.seqan.de")
5157 (synopsis "Library for nucleotide sequence analysis")
5158 (description
5159 "SeqAn is a C++ library of efficient algorithms and data structures for
5160the analysis of sequences with the focus on biological data. It contains
5161algorithms and data structures for string representation and their
5162manipulation, online and indexed string search, efficient I/O of
5163bioinformatics file formats, sequence alignment, and more.")
5164 (license license:bsd-3)))
ce7155d5 5165
d708b7a9
BW		 5166(define-public seqmagick
5167 (package
5168 (name "seqmagick")
5169 (version "0.6.1")
5170 (source
5171 (origin
5172 (method url-fetch)
5173 (uri (string-append
5174 "https://pypi.python.org/packages/source/s/seqmagick/seqmagick-"
5175 version ".tar.gz"))
5176 (sha256
5177 (base32
5178 "0cgn477n74gsl4qdaakrrhi953kcsd4q3ivk2lr18x74s3g4ma1d"))))
5179 (build-system python-build-system)
5180 (arguments
5181 ;; python2 only, see https://github.com/fhcrc/seqmagick/issues/56
5182 `(#:python ,python-2
5183 #:phases
5184 (modify-phases %standard-phases
5185 ;; Current test in setup.py does not work as of 0.6.1,
5186 ;; so use nose to run tests instead for now. See
5187 ;; https://github.com/fhcrc/seqmagick/issues/55
5188 (replace 'check (lambda _ (zero? (system* "nosetests")))))))
5189 (inputs
7bba95b7
BW		 5190 ;; biopython-1.66 is required due to
5191 ;; https://github.com/fhcrc/seqmagick/issues/59
5192 ;; When that issue is resolved the 'python2-biopython-1.66' package
5193 ;; should be removed.
5194 `(("python-biopython" ,python2-biopython-1.66)))
d708b7a9 5195 (native-inputs
f3b98f4f 5196 `(("python-nose" ,python2-nose)))
7bf837fd 5197 (home-page "https://github.com/fhcrc/seqmagick")
d708b7a9
BW		 5198 (synopsis "Tools for converting and modifying sequence files")
5199 (description
5200 "Bioinformaticians often have to convert sequence files between formats
5201and do little manipulations on them, and it's not worth writing scripts for
5202that. Seqmagick is a utility to expose the file format conversion in
5203BioPython in a convenient way. Instead of having a big mess of scripts, there
5204is one that takes arguments.")
5205 (license license:gpl3)))
5206
66daf78c
BW		 5207(define-public seqtk
5208 (package
5209 (name "seqtk")
5210 (version "1.2")
5211 (source (origin
5212 (method url-fetch)
5213 (uri (string-append
5214 "https://github.com/lh3/seqtk/archive/v"
5215 version ".tar.gz"))
5216 (file-name (string-append name "-" version ".tar.gz"))
5217 (sha256
5218 (base32
5219 "0ywdyzpmfiz2wp6ampbzqg4y8bj450nfgqarpamg045b8mk32lxx"))
5220 (modules '((guix build utils)))
5221 (snippet
5222 '(begin
5223 ;; Remove extraneous header files, as is done in the seqtk
5224 ;; master branch.
5225 (for-each (lambda (file) (delete-file file))
5226 (list "ksort.h" "kstring.h" "kvec.h"))
5227 #t))))
5228 (build-system gnu-build-system)
5229 (arguments
5230 `(#:phases
5231 (modify-phases %standard-phases
5232 (delete 'configure)
5233 (replace 'check
5234 ;; There are no tests, so we just run a sanity check.
5235 (lambda _ (zero? (system* "./seqtk" "seq"))))
5236 (replace 'install
5237 (lambda* (#:key outputs #:allow-other-keys)
5238 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5239 (install-file "seqtk" bin)))))))
5240 (inputs
5241 `(("zlib" ,zlib)))
5242 (home-page "https://github.com/lh3/seqtk")
5243 (synopsis "Toolkit for processing biological sequences in FASTA/Q format")
5244 (description
5245 "Seqtk is a fast and lightweight tool for processing sequences in the
5246FASTA or FASTQ format. It parses both FASTA and FASTQ files which can be
5247optionally compressed by gzip.")
5248 (license license:expat)))
5249
5f7e17be
BW		 5250(define-public snap-aligner
5251 (package
5252 (name "snap-aligner")
5253 (version "1.0beta.18")
5254 (source (origin
5255 (method url-fetch)
5256 (uri (string-append
5257 "https://github.com/amplab/snap/archive/v"
5258 version ".tar.gz"))
5259 (file-name (string-append name "-" version ".tar.gz"))
5260 (sha256
5261 (base32
5262 "1vnsjwv007k1fl1q7d681kbwn6bc66cgw6h16hym6gvyy71qv2ly"))))
5263 (build-system gnu-build-system)
5264 (arguments
5265 '(#:phases
5266 (modify-phases %standard-phases
5267 (delete 'configure)
5268 (replace 'check (lambda _ (zero? (system* "./unit_tests"))))
5269 (replace 'install
5270 (lambda* (#:key outputs #:allow-other-keys)
5271 (let* ((out (assoc-ref outputs "out"))
5272 (bin (string-append out "/bin")))
5f7e17be
BW		 5273 (install-file "snap-aligner" bin)
5274 (install-file "SNAPCommand" bin)
5275 #t))))))
5276 (native-inputs
5277 `(("zlib" ,zlib)))
5278 (home-page "http://snap.cs.berkeley.edu/")
5279 (synopsis "Short read DNA sequence aligner")
5280 (description
5281 "SNAP is a fast and accurate aligner for short DNA reads. It is
5282optimized for modern read lengths of 100 bases or higher, and takes advantage
5283of these reads to align data quickly through a hash-based indexing scheme.")
3e6fdd5f
EF		 5284 ;; 32-bit systems are not supported by the unpatched code.
5285 ;; Following the bug reports https://github.com/amplab/snap/issues/68 and
5286 ;; https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=812378 we see that
5287 ;; systems without a lot of memory cannot make good use of this program.
5288 (supported-systems '("x86_64-linux"))
5f7e17be
BW		 5289 (license license:asl2.0)))
5290
bcadaf00
BW		 5291(define-public sortmerna
5292 (package
5293 (name "sortmerna")
849485f5 5294 (version "2.1b")
bcadaf00
BW		 5295 (source
5296 (origin
5297 (method url-fetch)
5298 (uri (string-append
5299 "https://github.com/biocore/sortmerna/archive/"
5300 version ".tar.gz"))
5301 (file-name (string-append name "-" version ".tar.gz"))
5302 (sha256
5303 (base32
849485f5 5304 "1ghaghvd82af9j5adavxh77g7hm247d1r69m3fbi6f1jdivj5ldk"))))
bcadaf00
BW		 5305 (build-system gnu-build-system)
5306 (outputs '("out" ;for binaries
5307 "db")) ;for sequence databases
5308 (arguments
5309 `(#:phases
5310 (modify-phases %standard-phases
5311 (replace 'install
5312 (lambda* (#:key outputs #:allow-other-keys)
5313 (let* ((out (assoc-ref outputs "out"))
5314 (bin (string-append out "/bin"))
5315 (db (assoc-ref outputs "db"))
5316 (share
5317 (string-append db "/share/sortmerna/rRNA_databases")))
5318 (install-file "sortmerna" bin)
5319 (install-file "indexdb_rna" bin)
5320 (for-each (lambda (file)
5321 (install-file file share))
5322 (find-files "rRNA_databases" ".*fasta"))
5323 #t))))))
849485f5
BW		 5324 (inputs
5325 `(("zlib" ,zlib)))
bcadaf00
BW		 5326 (home-page "http://bioinfo.lifl.fr/RNA/sortmerna")
5327 (synopsis "Biological sequence analysis tool for NGS reads")
5328 (description
5329 "SortMeRNA is a biological sequence analysis tool for filtering, mapping
5330and operational taxonomic unit (OTU) picking of next generation
5331sequencing (NGS) reads. The core algorithm is based on approximate seeds and
5332allows for fast and sensitive analyses of nucleotide sequences. The main
5333application of SortMeRNA is filtering rRNA from metatranscriptomic data.")
2324541b
EF		 5334 ;; The source includes x86 specific code
5335 (supported-systems '("x86_64-linux" "i686-linux"))
bcadaf00
BW		 5336 (license license:lgpl3)))
5337
ce7155d5
RW		 5338(define-public star
5339 (package
5340 (name "star")
939b6b11 5341 (version "2.5.3a")
ce7155d5
RW		 5342 (source (origin
5343 (method url-fetch)
3062d750
RW		 5344 (uri (string-append "https://github.com/alexdobin/STAR/archive/"
5345 version ".tar.gz"))
5346 (file-name (string-append name "-" version ".tar.gz"))
ce7155d5
RW		 5347 (sha256
5348 (base32
939b6b11 5349 "013wirlz8lllgjyagl48l75n1isxyabqb3sj7qlsl0x1rmvqw99a"))
ce7155d5
RW		 5350 (modules '((guix build utils)))
5351 (snippet
3062d750
RW		 5352 '(begin
5353 (substitute* "source/Makefile"
5354 (("/bin/rm") "rm"))
5355 ;; Remove pre-built binaries and bundled htslib sources.
5356 (delete-file-recursively "bin/MacOSX_x86_64")
5357 (delete-file-recursively "bin/Linux_x86_64")
c9e9154e 5358 (delete-file-recursively "bin/Linux_x86_64_static")
3062d750
RW		 5359 (delete-file-recursively "source/htslib")
5360 #t))))
ce7155d5
RW		 5361 (build-system gnu-build-system)
5362 (arguments
5363 '(#:tests? #f ;no check target
5364 #:make-flags '("STAR")
5365 #:phases
c0266e8d
RW		 5366 (modify-phases %standard-phases
5367 (add-after 'unpack 'enter-source-dir
5368 (lambda _ (chdir "source") #t))
3062d750
RW		 5369 (add-after 'enter-source-dir 'do-not-use-bundled-htslib
5370 (lambda _
5371 (substitute* "Makefile"
5372 (("(Depend.list: \\$\\(SOURCES\\) parametersDefault\\.xxd) htslib"
5373 _ prefix) prefix))
5374 (substitute* '("BAMfunctions.cpp"
5375 "signalFromBAM.h"
5376 "bam_cat.h"
5377 "bam_cat.c"
5378 "STAR.cpp"
5379 "bamRemoveDuplicates.cpp")
5380 (("#include \"htslib/([^\"]+\\.h)\"" _ header)
5381 (string-append "#include <" header ">")))
5382 (substitute* "IncludeDefine.h"
5383 (("\"htslib/(htslib/[^\"]+.h)\"" _ header)
5384 (string-append "<" header ">")))
5385 #t))
c0266e8d
RW		 5386 (replace 'install
5387 (lambda* (#:key outputs #:allow-other-keys)
5388 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5389 (install-file "STAR" bin))
5390 #t))
5391 (delete 'configure))))
ce7155d5 5392 (native-inputs
9fc513ad 5393 `(("xxd" ,xxd)))
ce7155d5 5394 (inputs
3062d750
RW		 5395 `(("htslib" ,htslib)
5396 ("zlib" ,zlib)))
ce7155d5
RW		 5397 (home-page "https://github.com/alexdobin/STAR")
5398 (synopsis "Universal RNA-seq aligner")
5399 (description
5400 "The Spliced Transcripts Alignment to a Reference (STAR) software is
5401based on a previously undescribed RNA-seq alignment algorithm that uses
5402sequential maximum mappable seed search in uncompressed suffix arrays followed
5403by seed clustering and stitching procedure. In addition to unbiased de novo
5404detection of canonical junctions, STAR can discover non-canonical splices and
5405chimeric (fusion) transcripts, and is also capable of mapping full-length RNA
5406sequences.")
c11f79a4
BW		 5407 ;; Only 64-bit systems are supported according to the README.
5408 (supported-systems '("x86_64-linux" "mips64el-linux"))
ce7155d5
RW		 5409 ;; STAR is licensed under GPLv3 or later; htslib is MIT-licensed.
5410 (license license:gpl3+)))
de07c0db 5411
dbf4ed7c
RW		 5412(define-public subread
5413 (package
5414 (name "subread")
c180533b 5415 (version "1.5.1")
dbf4ed7c
RW		 5416 (source (origin
5417 (method url-fetch)
de67e922
LF		 5418 (uri (string-append "mirror://sourceforge/subread/subread-"
5419 version "/subread-" version "-source.tar.gz"))
dbf4ed7c
RW		 5420 (sha256
5421 (base32
c180533b 5422 "0gn5zhbvllks0mmdg3qlmsbg91p2mpdc2wixwfqpi85yzfrh8hcy"))))
dbf4ed7c
RW		 5423 (build-system gnu-build-system)
5424 (arguments
5425 `(#:tests? #f ;no "check" target
104c1986
RW		 5426 ;; The CC and CCFLAGS variables are set to contain a lot of x86_64
5427 ;; optimizations by default, so we override these flags such that x86_64
5428 ;; flags are only added when the build target is an x86_64 system.
5429 #:make-flags
5430 (list (let ((system ,(or (%current-target-system)
5431 (%current-system)))
5432 (flags '("-ggdb" "-fomit-frame-pointer"
5433 "-ffast-math" "-funroll-loops"
5434 "-fmessage-length=0"
5435 "-O9" "-Wall" "-DMAKE_FOR_EXON"
5436 "-DMAKE_STANDALONE"
5437 "-DSUBREAD_VERSION=\\\"${SUBREAD_VERSION}\\\""))
5438 (flags64 '("-mmmx" "-msse" "-msse2" "-msse3")))
5439 (if (string-prefix? "x86_64" system)
5440 (string-append "CCFLAGS=" (string-join (append flags flags64)))
5441 (string-append "CCFLAGS=" (string-join flags))))
5442 "-f" "Makefile.Linux"
5443 "CC=gcc ${CCFLAGS}")
dbf4ed7c 5444 #:phases
dc1d3cde
KK		 5445 (modify-phases %standard-phases
5446 (add-after 'unpack 'enter-dir
5447 (lambda _ (chdir "src") #t))
5448 (replace 'install
5449 (lambda* (#:key outputs #:allow-other-keys)
5450 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5451 (mkdir-p bin)
5452 (copy-recursively "../bin" bin))))
dbf4ed7c 5453 ;; no "configure" script
dc1d3cde 5454 (delete 'configure))))
dbf4ed7c
RW		 5455 (inputs `(("zlib" ,zlib)))
5456 (home-page "http://bioinf.wehi.edu.au/subread-package/")
5457 (synopsis "Tool kit for processing next-gen sequencing data")
5458 (description
5459 "The subread package contains the following tools: subread aligner, a
5460general-purpose read aligner; subjunc aligner: detecting exon-exon junctions
5461and mapping RNA-seq reads; featureCounts: counting mapped reads for genomic
5462features; exactSNP: a SNP caller that discovers SNPs by testing signals
5463against local background noises.")
5464 (license license:gpl3+)))
5465
d15d981e
RW		 5466(define-public stringtie
5467 (package
5468 (name "stringtie")
5469 (version "1.2.1")
5470 (source (origin
5471 (method url-fetch)
5472 (uri (string-append "http://ccb.jhu.edu/software/stringtie/dl/"
5473 "stringtie-" version ".tar.gz"))
5474 (sha256
5475 (base32
5476 "1cqllsc1maq4kh92isi8yadgzbmnf042hlnalpk3y59aph1z3bfz"))
5477 (modules '((guix build utils)))
5478 (snippet
5479 '(begin
5480 (delete-file-recursively "samtools-0.1.18")
5481 #t))))
5482 (build-system gnu-build-system)
5483 (arguments
5484 `(#:tests? #f ;no test suite
5485 #:phases
5486 (modify-phases %standard-phases
5487 ;; no configure script
5488 (delete 'configure)
5489 (add-before 'build 'use-system-samtools
5490 (lambda _
5491 (substitute* "Makefile"
5492 (("stringtie: \\$\\{BAM\\}/libbam\\.a")
5493 "stringtie: "))
5494 (substitute* '("gclib/GBam.h"
5495 "gclib/GBam.cpp")
5496 (("#include \"(bam|sam|kstring).h\"" _ header)
5497 (string-append "#include <samtools/" header ".h>")))
5498 #t))
0d2c0562
RW		 5499 (add-after 'unpack 'remove-duplicate-typedef
5500 (lambda _
5501 ;; This typedef conflicts with the typedef in
5502 ;; glibc-2.25/include/bits/types.h
5503 (substitute* "gclib/GThreads.h"
5504 (("typedef long long __intmax_t;") ""))
5505 #t))
d15d981e
RW		 5506 (replace 'install
5507 (lambda* (#:key outputs #:allow-other-keys)
5508 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5509 (install-file "stringtie" bin)
5510 #t))))))
5511 (inputs
5512 `(("samtools" ,samtools-0.1)
5513 ("zlib" ,zlib)))
5514 (home-page "http://ccb.jhu.edu/software/stringtie/")
5515 (synopsis "Transcript assembly and quantification for RNA-Seq data")
5516 (description
5517 "StringTie is a fast and efficient assembler of RNA-Seq sequence
5518alignments into potential transcripts. It uses a novel network flow algorithm
5519as well as an optional de novo assembly step to assemble and quantitate
5520full-length transcripts representing multiple splice variants for each gene
5521locus. Its input can include not only the alignments of raw reads used by
5522other transcript assemblers, but also alignments of longer sequences that have
5523been assembled from those reads. To identify differentially expressed genes
5524between experiments, StringTie's output can be processed either by the
5525Cuffdiff or Ballgown programs.")
5526 (license license:artistic2.0)))
5527
ad0ae297
BW		 5528(define-public taxtastic
5529 (package
5530 (name "taxtastic")
3cbfc149 5531 (version "0.6.4")
ad0ae297
BW		 5532 (source (origin
5533 (method url-fetch)
3cbfc149 5534 (uri (pypi-uri "taxtastic" version))
ad0ae297
BW		 5535 (sha256
5536 (base32
3cbfc149 5537 "0s79z8kfl853x7l4h8ms05k31q87aw62nrchlk20w9n227j35929"))))
ad0ae297
BW		 5538 (build-system python-build-system)
5539 (arguments
5540 `(#:python ,python-2
5541 #:phases
5542 (modify-phases %standard-phases
5543 (replace 'check
5544 (lambda _
5545 (zero? (system* "python" "-m" "unittest" "discover" "-v")))))))
5546 (propagated-inputs
5547 `(("python-sqlalchemy" ,python2-sqlalchemy)
5548 ("python-decorator" ,python2-decorator)
5549 ("python-biopython" ,python2-biopython)
5550 ("python-pandas" ,python2-pandas)))
5551 (home-page "https://github.com/fhcrc/taxtastic")
5552 (synopsis "Tools for taxonomic naming and annotation")
5553 (description
5554 "Taxtastic is software written in python used to build and maintain
5555reference packages i.e. collections of reference trees, reference alignments,
5556profiles, and associated taxonomic information.")
5557 (license license:gpl3+)))
5558
de07c0db
RW		 5559(define-public vcftools
5560 (package
5561 (name "vcftools")
f4322542 5562 (version "0.1.15")
de07c0db
RW		 5563 (source (origin
5564 (method url-fetch)
5565 (uri (string-append
9b36e256
RJ		 5566 "https://github.com/vcftools/vcftools/releases/download/v"
5567 version "/vcftools-" version ".tar.gz"))
de07c0db
RW		 5568 (sha256
5569 (base32
f4322542 5570 "1qw30c45wihgy632rbz4rh3njnwj4msj46l1rsgdhyg6bgypmr1i"))))
de07c0db
RW		 5571 (build-system gnu-build-system)
5572 (arguments
5573 `(#:tests? #f ; no "check" target
5574 #:make-flags (list
7c3958e1 5575 "CFLAGS=-O2" ; override "-m64" flag
de07c0db
RW		 5576 (string-append "PREFIX=" (assoc-ref %outputs "out"))
5577 (string-append "MANDIR=" (assoc-ref %outputs "out")
9b36e256
RJ		 5578 "/share/man/man1"))))
5579 (native-inputs
5580 `(("pkg-config" ,pkg-config)))
de07c0db
RW		 5581 (inputs
5582 `(("perl" ,perl)
5583 ("zlib" ,zlib)))
9b36e256 5584 (home-page "https://vcftools.github.io/")
de07c0db
RW		 5585 (synopsis "Tools for working with VCF files")
5586 (description
5587 "VCFtools is a program package designed for working with VCF files, such
5588as those generated by the 1000 Genomes Project. The aim of VCFtools is to
5589provide easily accessible methods for working with complex genetic variation
5590data in the form of VCF files.")
5591 ;; The license is declared as LGPLv3 in the README and
9b36e256 5592 ;; at https://vcftools.github.io/license.html
de07c0db 5593 (license license:lgpl3)))
9c38b540 5594
35aa90a1
RW		 5595(define-public infernal
5596 (package
5597 (name "infernal")
5598 (version "1.1.2")
5599 (source (origin
5600 (method url-fetch)
5601 (uri (string-append "http://eddylab.org/software/infernal/"
5602 "infernal-" version ".tar.gz"))
5603 (sha256
5604 (base32
5605 "0sr2hiz3qxfwqpz3whxr6n82p3x27336v3f34iqznp10hks2935c"))))
5606 (build-system gnu-build-system)
5607 (native-inputs
5608 `(("perl" ,perl))) ; for tests
5609 (home-page "http://eddylab.org/infernal/")
5610 (synopsis "Inference of RNA alignments")
5611 (description "Infernal (\"INFERence of RNA ALignment\") is a tool for
5612searching DNA sequence databases for RNA structure and sequence similarities.
5613It is an implementation of a special case of profile stochastic context-free
5614grammars called @dfn{covariance models} (CMs). A CM is like a sequence
5615profile, but it scores a combination of sequence consensus and RNA secondary
5616structure consensus, so in many cases, it is more capable of identifying RNA
5617homologs that conserve their secondary structure more than their primary
5618sequence.")
48409ef2
EF		 5619 ;; Infernal 1.1.2 requires VMX or SSE capability for parallel instructions.
5620 (supported-systems '("i686-linux" "x86_64-linux"))
35aa90a1
RW		 5621 (license license:bsd-3)))
5622
b91cfa22
RW		 5623(define-public r-centipede
5624 (package
5625 (name "r-centipede")
5626 (version "1.2")
5627 (source (origin
5628 (method url-fetch)
5629 (uri (string-append "http://download.r-forge.r-project.org/"
5630 "src/contrib/CENTIPEDE_" version ".tar.gz"))
5631 (sha256
5632 (base32
5633 "1hsx6qgwr0i67fhy9257zj7s0ppncph2hjgbia5nn6nfmj0ax6l9"))))
5634 (build-system r-build-system)
5635 (home-page "http://centipede.uchicago.edu/")
5636 (synopsis "Predict transcription factor binding sites")
5637 (description
5638 "CENTIPEDE applies a hierarchical Bayesian mixture model to infer regions
5639of the genome that are bound by particular transcription factors. It starts
5640by identifying a set of candidate binding sites, and then aims to classify the
5641sites according to whether each site is bound or not bound by a transcription
5642factor. CENTIPEDE is an unsupervised learning algorithm that discriminates
5643between two different types of motif instances using as much relevant
5644information as possible.")
5645 (license (list license:gpl2+ license:gpl3+))))
5646
7b3df1e5
BW		 5647(define-public r-vegan
5648 (package
5649 (name "r-vegan")
db2e4386 5650 (version "2.4-4")
7b3df1e5
BW		 5651 (source
5652 (origin
5653 (method url-fetch)
5654 (uri (cran-uri "vegan" version))
5655 (sha256
5656 (base32
db2e4386 5657 "1n57dzv2aid6iqd9fkqik401sidqanhzsawyak94qbiyh6dbd1x9"))))
7b3df1e5 5658 (build-system r-build-system)
7b3df1e5 5659 (native-inputs
db2e4386 5660 `(("gfortran" ,gfortran)))
7b3df1e5
BW		 5661 (propagated-inputs
5662 `(("r-cluster" ,r-cluster)
5663 ("r-lattice" ,r-lattice)
aeb64f3c 5664 ("r-mass" ,r-mass)
7b3df1e5
BW		 5665 ("r-mgcv" ,r-mgcv)
5666 ("r-permute" ,r-permute)))
5667 (home-page "https://cran.r-project.org/web/packages/vegan")
5668 (synopsis "Functions for community ecology")
5669 (description
5670 "The vegan package provides tools for descriptive community ecology. It
5671has most basic functions of diversity analysis, community ordination and
5672dissimilarity analysis. Most of its multivariate tools can be used for other
5673data types as well.")
5674 (license license:gpl2+)))
5675
8c6de588
RW		 5676(define-public r-annotate
5677 (package
5678 (name "r-annotate")
2d4e887d 5679 (version "1.56.0")
8c6de588
RW		 5680 (source
5681 (origin
5682 (method url-fetch)
5683 (uri (bioconductor-uri "annotate" version))
5684 (sha256
5685 (base32
2d4e887d 5686 "0wlrp3v2jxw9is98ap39dfi7z97kmw1wv1xi4h7yfh12zpj2r8l0"))))
8c6de588
RW		 5687 (build-system r-build-system)
5688 (propagated-inputs
5689 `(("r-annotationdbi" ,r-annotationdbi)
5690 ("r-biobase" ,r-biobase)
5691 ("r-biocgenerics" ,r-biocgenerics)
5692 ("r-dbi" ,r-dbi)
d0f0579e 5693 ("r-rcurl" ,r-rcurl)
8c6de588
RW		 5694 ("r-xml" ,r-xml)
5695 ("r-xtable" ,r-xtable)))
5696 (home-page
5713bbf1 5697 "https://bioconductor.org/packages/annotate")
8c6de588 5698 (synopsis "Annotation for microarrays")
d1e4ad1b 5699 (description "This package provides R environments for the annotation of
8c6de588
RW		 5700microarrays.")
5701 (license license:artistic2.0)))
5702
07a664cd
RW		 5703(define-public r-geneplotter
5704 (package
5705 (name "r-geneplotter")
cb4d3ff2 5706 (version "1.56.0")
07a664cd
RW		 5707 (source
5708 (origin
5709 (method url-fetch)
5710 (uri (bioconductor-uri "geneplotter" version))
5711 (sha256
5712 (base32
cb4d3ff2 5713 "1z3g7frc1iviwrsv2dlm4nqvkc0685h4va0388yfxn102ln8wwma"))))
07a664cd
RW		 5714 (build-system r-build-system)
5715 (propagated-inputs
5716 `(("r-annotate" ,r-annotate)
5717 ("r-annotationdbi" ,r-annotationdbi)
5718 ("r-biobase" ,r-biobase)
5719 ("r-biocgenerics" ,r-biocgenerics)
5720 ("r-lattice" ,r-lattice)
5721 ("r-rcolorbrewer" ,r-rcolorbrewer)))
5713bbf1 5722 (home-page "https://bioconductor.org/packages/geneplotter")
07a664cd
RW		 5723 (synopsis "Graphics functions for genomic data")
5724 (description
5725 "This package provides functions for plotting genomic data.")
5726 (license license:artistic2.0)))
5727
2301fd3e
RW		 5728(define-public r-genefilter
5729 (package
5730 (name "r-genefilter")
554fac74 5731 (version "1.60.0")
2301fd3e
RW		 5732 (source
5733 (origin
5734 (method url-fetch)
5735 (uri (bioconductor-uri "genefilter" version))
5736 (sha256
5737 (base32
554fac74 5738 "173swlg6gj4kdllbqvyiw5dggbcxiwlwpqmllsv4dxzn7h25i3g7"))))
2301fd3e
RW		 5739 (build-system r-build-system)
5740 (native-inputs
5741 `(("gfortran" ,gfortran)))
5742 (propagated-inputs
5743 `(("r-annotate" ,r-annotate)
5744 ("r-annotationdbi" ,r-annotationdbi)
5745 ("r-biobase" ,r-biobase)
aeb64f3c
RW		 5746 ("r-s4vectors" ,r-s4vectors)
5747 ("r-survival" ,r-survival)))
5713bbf1 5748 (home-page "https://bioconductor.org/packages/genefilter")
2301fd3e
RW		 5749 (synopsis "Filter genes from high-throughput experiments")
5750 (description
5751 "This package provides basic functions for filtering genes from
5752high-throughput sequencing experiments.")
5753 (license license:artistic2.0)))
5754
ad34f0ac
RW		 5755(define-public r-deseq2
5756 (package
5757 (name "r-deseq2")
0166dc07 5758 (version "1.18.0")
ad34f0ac
RW		 5759 (source
5760 (origin
5761 (method url-fetch)
5762 (uri (bioconductor-uri "DESeq2" version))
5763 (sha256
5764 (base32
0166dc07 5765 "1hcxnkkjfvz4hj8iqidshwsjq7jnl1z7wj63dvcwlx1zx5aichyh"))))
ad34f0ac
RW		 5766 (properties `((upstream-name . "DESeq2")))
5767 (build-system r-build-system)
ad34f0ac
RW		 5768 (propagated-inputs
5769 `(("r-biobase" ,r-biobase)
5770 ("r-biocgenerics" ,r-biocgenerics)
5771 ("r-biocparallel" ,r-biocparallel)
5772 ("r-genefilter" ,r-genefilter)
5773 ("r-geneplotter" ,r-geneplotter)
5774 ("r-genomicranges" ,r-genomicranges)
5775 ("r-ggplot2" ,r-ggplot2)
5776 ("r-hmisc" ,r-hmisc)
5777 ("r-iranges" ,r-iranges)
5778 ("r-locfit" ,r-locfit)
5779 ("r-rcpp" ,r-rcpp)
5780 ("r-rcpparmadillo" ,r-rcpparmadillo)
5781 ("r-s4vectors" ,r-s4vectors)
5782 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5783 (home-page "https://bioconductor.org/packages/DESeq2")
ad34f0ac
RW		 5784 (synopsis "Differential gene expression analysis")
5785 (description
5786 "This package provides functions to estimate variance-mean dependence in
5787count data from high-throughput nucleotide sequencing assays and test for
5788differential expression based on a model using the negative binomial
5789distribution.")
5790 (license license:lgpl3+)))
5791
86763fdd
RW		 5792(define-public r-dexseq
5793 (package
5794 (name "r-dexseq")
3fd010ff 5795 (version "1.24.0")
86763fdd
RW		 5796 (source
5797 (origin
5798 (method url-fetch)
5799 (uri (bioconductor-uri "DEXSeq" version))
5800 (sha256
5801 (base32
3fd010ff 5802 "0qxwnz2ffhav9slcn095k206cfza9i3i5l7w1154plf08gpy1d1d"))))
86763fdd
RW		 5803 (properties `((upstream-name . "DEXSeq")))
5804 (build-system r-build-system)
5805 (propagated-inputs
5806 `(("r-annotationdbi" ,r-annotationdbi)
5807 ("r-biobase" ,r-biobase)
5808 ("r-biocgenerics" ,r-biocgenerics)
5809 ("r-biocparallel" ,r-biocparallel)
5810 ("r-biomart" ,r-biomart)
5811 ("r-deseq2" ,r-deseq2)
5812 ("r-genefilter" ,r-genefilter)
5813 ("r-geneplotter" ,r-geneplotter)
5814 ("r-genomicranges" ,r-genomicranges)
5815 ("r-hwriter" ,r-hwriter)
5816 ("r-iranges" ,r-iranges)
5817 ("r-rcolorbrewer" ,r-rcolorbrewer)
5818 ("r-rsamtools" ,r-rsamtools)
5819 ("r-s4vectors" ,r-s4vectors)
5820 ("r-statmod" ,r-statmod)
5821 ("r-stringr" ,r-stringr)
5822 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5823 (home-page "https://bioconductor.org/packages/DEXSeq")
86763fdd
RW		 5824 (synopsis "Inference of differential exon usage in RNA-Seq")
5825 (description
5826 "This package is focused on finding differential exon usage using RNA-seq
5827exon counts between samples with different experimental designs. It provides
5828functions that allows the user to make the necessary statistical tests based
5829on a model that uses the negative binomial distribution to estimate the
5830variance between biological replicates and generalized linear models for
5831testing. The package also provides functions for the visualization and
5832exploration of the results.")
5833 (license license:gpl3+)))
5834
e8163773
RW		 5835(define-public r-annotationforge
5836 (package
5837 (name "r-annotationforge")
727706ad 5838 (version "1.20.0")
e8163773
RW		 5839 (source
5840 (origin
5841 (method url-fetch)
5842 (uri (bioconductor-uri "AnnotationForge" version))
5843 (sha256
5844 (base32
727706ad 5845 "01vbrf76vqfvxh6vpfxkjwccxggnha3byqzj333glqz2b6kwx5q1"))))
e8163773
RW		 5846 (properties
5847 `((upstream-name . "AnnotationForge")))
5848 (build-system r-build-system)
5849 (propagated-inputs
5850 `(("r-annotationdbi" ,r-annotationdbi)
5851 ("r-biobase" ,r-biobase)
5852 ("r-biocgenerics" ,r-biocgenerics)
5853 ("r-dbi" ,r-dbi)
55cd914c 5854 ("r-rcurl" ,r-rcurl)
e8163773
RW		 5855 ("r-rsqlite" ,r-rsqlite)
5856 ("r-s4vectors" ,r-s4vectors)
5857 ("r-xml" ,r-xml)))
5713bbf1 5858 (home-page "https://bioconductor.org/packages/AnnotationForge")
e8163773
RW		 5859 (synopsis "Code for building annotation database packages")
5860 (description
5861 "This package provides code for generating Annotation packages and their
5862databases. Packages produced are intended to be used with AnnotationDbi.")
5863 (license license:artistic2.0)))
5864
cd9e7dc7
RW		 5865(define-public r-rbgl
5866 (package
5867 (name "r-rbgl")
c25154e5 5868 (version "1.54.0")
cd9e7dc7
RW		 5869 (source
5870 (origin
5871 (method url-fetch)
5872 (uri (bioconductor-uri "RBGL" version))
5873 (sha256
5874 (base32
c25154e5 5875 "18jad23i3899ypv4bg3l47cvvs3qnj1pqis2p9x0135yv5y6wnv7"))))
cd9e7dc7
RW		 5876 (properties `((upstream-name . "RBGL")))
5877 (build-system r-build-system)
5878 (propagated-inputs `(("r-graph" ,r-graph)))
5713bbf1 5879 (home-page "https://www.bioconductor.org/packages/RBGL")
cd9e7dc7
RW		 5880 (synopsis "Interface to the Boost graph library")
5881 (description
5882 "This package provides a fairly extensive and comprehensive interface to
5883the graph algorithms contained in the Boost library.")
5884 (license license:artistic2.0)))
5885
ad740ff8
RW		 5886(define-public r-gseabase
5887 (package
5888 (name "r-gseabase")
9f52e53d 5889 (version "1.40.0")
ad740ff8
RW		 5890 (source
5891 (origin
5892 (method url-fetch)
5893 (uri (bioconductor-uri "GSEABase" version))
5894 (sha256
5895 (base32
9f52e53d 5896 "0kpkl6c5lrar6ip7wlhvd5axqlb9lb5l3lgbdb3dlih32c3nz0yq"))))
ad740ff8
RW		 5897 (properties `((upstream-name . "GSEABase")))
5898 (build-system r-build-system)
5899 (propagated-inputs
5900 `(("r-annotate" ,r-annotate)
5901 ("r-annotationdbi" ,r-annotationdbi)
5902 ("r-biobase" ,r-biobase)
5903 ("r-biocgenerics" ,r-biocgenerics)
5904 ("r-graph" ,r-graph)
5905 ("r-xml" ,r-xml)))
5713bbf1 5906 (home-page "https://bioconductor.org/packages/GSEABase")
ad740ff8
RW		 5907 (synopsis "Gene set enrichment data structures and methods")
5908 (description
5909 "This package provides classes and methods to support @dfn{Gene Set
5910Enrichment Analysis} (GSEA).")
5911 (license license:artistic2.0)))
5912
1a1931f7
RW		 5913(define-public r-category
5914 (package
5915 (name "r-category")
2404cc42 5916 (version "2.44.0")
1a1931f7
RW		 5917 (source
5918 (origin
5919 (method url-fetch)
5920 (uri (bioconductor-uri "Category" version))
5921 (sha256
5922 (base32
2404cc42 5923 "0mkav04vbla0xfa0dssxdd0rjs589sxi83xklf5iq5hj3dm8y0i8"))))
1a1931f7
RW		 5924 (properties `((upstream-name . "Category")))
5925 (build-system r-build-system)
5926 (propagated-inputs
5927 `(("r-annotate" ,r-annotate)
5928 ("r-annotationdbi" ,r-annotationdbi)
5929 ("r-biobase" ,r-biobase)
5930 ("r-biocgenerics" ,r-biocgenerics)
5931 ("r-genefilter" ,r-genefilter)
5932 ("r-graph" ,r-graph)
5933 ("r-gseabase" ,r-gseabase)
5934 ("r-matrix" ,r-matrix)
5935 ("r-rbgl" ,r-rbgl)
2404cc42 5936 ("r-dbi" ,r-dbi)))
5713bbf1 5937 (home-page "https://bioconductor.org/packages/Category")
1a1931f7
RW		 5938 (synopsis "Category analysis")
5939 (description
5940 "This package provides a collection of tools for performing category
5941analysis.")
5942 (license license:artistic2.0)))
5943
89f40c5e
RW		 5944(define-public r-gostats
5945 (package
5946 (name "r-gostats")
eb3f5cc7 5947 (version "2.44.0")
89f40c5e
RW		 5948 (source
5949 (origin
5950 (method url-fetch)
5951 (uri (bioconductor-uri "GOstats" version))
5952 (sha256
5953 (base32
eb3f5cc7 5954 "04gqfdlx9fxf97qf0l28x4aaqvl10n6v58qiz5fiaw05sbj1pf1i"))))
89f40c5e
RW		 5955 (properties `((upstream-name . "GOstats")))
5956 (build-system r-build-system)
5957 (propagated-inputs
5958 `(("r-annotate" ,r-annotate)
5959 ("r-annotationdbi" ,r-annotationdbi)
5960 ("r-annotationforge" ,r-annotationforge)
5961 ("r-biobase" ,r-biobase)
5962 ("r-category" ,r-category)
5963 ("r-go-db" ,r-go-db)
5964 ("r-graph" ,r-graph)
eb3f5cc7 5965 ("r-rgraphviz" ,r-rgraphviz)
89f40c5e 5966 ("r-rbgl" ,r-rbgl)))
5713bbf1 5967 (home-page "https://bioconductor.org/packages/GOstats")
89f40c5e
RW		 5968 (synopsis "Tools for manipulating GO and microarrays")
5969 (description
5970 "This package provides a set of tools for interacting with GO and
5971microarray data. A variety of basic manipulation tools for graphs, hypothesis
5972testing and other simple calculations.")
5973 (license license:artistic2.0)))
5974
cb99d457
RW		 5975(define-public r-shortread
5976 (package
5977 (name "r-shortread")
843826e2 5978 (version "1.36.0")
cb99d457
RW		 5979 (source
5980 (origin
5981 (method url-fetch)
5982 (uri (bioconductor-uri "ShortRead" version))
5983 (sha256
5984 (base32
843826e2 5985 "06mknlsmd4hnaxzdjapgvp2kgdnf9w103y500dsac5jgsz4vwzcz"))))
cb99d457
RW		 5986 (properties `((upstream-name . "ShortRead")))
5987 (build-system r-build-system)
5988 (inputs
5989 `(("zlib" ,zlib)))
5990 (propagated-inputs
5991 `(("r-biobase" ,r-biobase)
5992 ("r-biocgenerics" ,r-biocgenerics)
5993 ("r-biocparallel" ,r-biocparallel)
5994 ("r-biostrings" ,r-biostrings)
5995 ("r-genomeinfodb" ,r-genomeinfodb)
5996 ("r-genomicalignments" ,r-genomicalignments)
5997 ("r-genomicranges" ,r-genomicranges)
5998 ("r-hwriter" ,r-hwriter)
5999 ("r-iranges" ,r-iranges)
6000 ("r-lattice" ,r-lattice)
6001 ("r-latticeextra" ,r-latticeextra)
6002 ("r-rsamtools" ,r-rsamtools)
6003 ("r-s4vectors" ,r-s4vectors)
6004 ("r-xvector" ,r-xvector)
6005 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 6006 (home-page "https://bioconductor.org/packages/ShortRead")
cb99d457
RW		 6007 (synopsis "FASTQ input and manipulation tools")
6008 (description
6009 "This package implements sampling, iteration, and input of FASTQ files.
6010It includes functions for filtering and trimming reads, and for generating a
6011quality assessment report. Data are represented as
6012@code{DNAStringSet}-derived objects, and easily manipulated for a diversity of
6013purposes. The package also contains legacy support for early single-end,
6014ungapped alignment formats.")
6015 (license license:artistic2.0)))
6016
7f903d73
RW		 6017(define-public r-systempiper
6018 (package
6019 (name "r-systempiper")
b5375525 6020 (version "1.12.0")
7f903d73
RW		 6021 (source
6022 (origin
6023 (method url-fetch)
6024 (uri (bioconductor-uri "systemPipeR" version))
6025 (sha256
6026 (base32
b5375525 6027 "11mj8pjq5vj25768vmagpzv74fvi3p3kdk5zdlznqyiaggri04cv"))))
7f903d73
RW		 6028 (properties `((upstream-name . "systemPipeR")))
6029 (build-system r-build-system)
6030 (propagated-inputs
6031 `(("r-annotate" ,r-annotate)
6032 ("r-batchjobs" ,r-batchjobs)
6033 ("r-biocgenerics" ,r-biocgenerics)
6034 ("r-biostrings" ,r-biostrings)
6035 ("r-deseq2" ,r-deseq2)
6036 ("r-edger" ,r-edger)
6037 ("r-genomicfeatures" ,r-genomicfeatures)
6038 ("r-genomicranges" ,r-genomicranges)
6039 ("r-ggplot2" ,r-ggplot2)
6040 ("r-go-db" ,r-go-db)
6041 ("r-gostats" ,r-gostats)
6042 ("r-limma" ,r-limma)
6043 ("r-pheatmap" ,r-pheatmap)
6044 ("r-rjson" ,r-rjson)
6045 ("r-rsamtools" ,r-rsamtools)
6046 ("r-shortread" ,r-shortread)
6047 ("r-summarizedexperiment" ,r-summarizedexperiment)
6048 ("r-variantannotation" ,r-variantannotation)))
6049 (home-page "https://github.com/tgirke/systemPipeR")
6050 (synopsis "Next generation sequencing workflow and reporting environment")
6051 (description
6052 "This R package provides tools for building and running automated
6053end-to-end analysis workflows for a wide range of @dfn{next generation
6054sequence} (NGS) applications such as RNA-Seq, ChIP-Seq, VAR-Seq and Ribo-Seq.
6055Important features include a uniform workflow interface across different NGS
6056applications, automated report generation, and support for running both R and
6057command-line software, such as NGS aligners or peak/variant callers, on local
6058computers or compute clusters. Efficient handling of complex sample sets and
6059experimental designs is facilitated by a consistently implemented sample
6060annotation infrastructure.")
6061 (license license:artistic2.0)))
6062
684f29bd
RW		 6063(define-public r-grohmm
6064 (package
6065 (name "r-grohmm")
704fe4d1 6066 (version "1.12.0")
684f29bd
RW		 6067 (source
6068 (origin
6069 (method url-fetch)
6070 (uri (bioconductor-uri "groHMM" version))
6071 (sha256
6072 (base32
704fe4d1 6073 "0cjkj0ypyc4dfi9s8dh88kh6q4xlpnc0wal7njg4b4gqj0l2hva7"))))
684f29bd
RW		 6074 (properties `((upstream-name . "groHMM")))
6075 (build-system r-build-system)
6076 (propagated-inputs
6077 `(("r-genomeinfodb" ,r-genomeinfodb)
6078 ("r-genomicalignments" ,r-genomicalignments)
6079 ("r-genomicranges" ,r-genomicranges)
6080 ("r-iranges" ,r-iranges)
aeb64f3c 6081 ("r-mass" ,r-mass)
684f29bd
RW		 6082 ("r-rtracklayer" ,r-rtracklayer)
6083 ("r-s4vectors" ,r-s4vectors)))
6084 (home-page "https://github.com/Kraus-Lab/groHMM")
6085 (synopsis "GRO-seq analysis pipeline")
6086 (description
6087 "This package provides a pipeline for the analysis of GRO-seq data.")
6088 (license license:gpl3+)))
6089
f3cfe451
RW		 6090(define-public r-txdb-hsapiens-ucsc-hg19-knowngene
6091 (package
6092 (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
6093 (version "3.2.2")
6094 (source (origin
6095 (method url-fetch)
6096 ;; We cannot use bioconductor-uri here because this tarball is
6097 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 6098 (uri (string-append "https://bioconductor.org/packages/"
f3cfe451
RW		 6099 "release/data/annotation/src/contrib"
6100 "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
6101 version ".tar.gz"))
6102 (sha256
6103 (base32
6104 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
6105 (properties
6106 `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
6107 (build-system r-build-system)
6108 ;; As this package provides little more than a very large data file it
6109 ;; doesn't make sense to build substitutes.
6110 (arguments `(#:substitutable? #f))
6111 (propagated-inputs
6112 `(("r-genomicfeatures" ,r-genomicfeatures)))
6113 (home-page
5713bbf1 6114 "https://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
f3cfe451
RW		 6115 (synopsis "Annotation package for human genome in TxDb format")
6116 (description
6117 "This package provides an annotation database of Homo sapiens genome
6118data. It is derived from the UCSC hg19 genome and based on the \"knownGene\"
6119track. The database is exposed as a @code{TxDb} object.")
6120 (license license:artistic2.0)))
6121
325c039c
RJ		 6122(define-public r-sparql
6123 (package
6124 (name "r-sparql")
6125 (version "1.16")
6126 (source (origin
6127 (method url-fetch)
6128 (uri (cran-uri "SPARQL" version))
6129 (sha256
6130 (base32
6131 "0gak1q06yyhdmcxb2n3v0h9gr1vqd0viqji52wpw211qp6r6dcrc"))))
6132 (properties `((upstream-name . "SPARQL")))
6133 (build-system r-build-system)
6134 (propagated-inputs
6135 `(("r-rcurl" ,r-rcurl)
6136 ("r-xml" ,r-xml)))
6137 (home-page "http://cran.r-project.org/web/packages/SPARQL")
6138 (synopsis "SPARQL client for R")
6139 (description "This package provides an interface to use SPARQL to pose
6140SELECT or UPDATE queries to an end-point.")
6141 ;; The only license indication is found in the DESCRIPTION file,
6142 ;; which states GPL-3. So we cannot assume GPLv3+.
6143 (license license:gpl3)))
6144
a2950fa4
BW		 6145(define-public vsearch
6146 (package
6147 (name "vsearch")
7c087150 6148 (version "2.6.0")
a2950fa4
BW		 6149 (source
6150 (origin
6151 (method url-fetch)
6152 (uri (string-append
6153 "https://github.com/torognes/vsearch/archive/v"
6154 version ".tar.gz"))
6155 (file-name (string-append name "-" version ".tar.gz"))
6156 (sha256
6157 (base32
7c087150 6158 "0rplgpvsdkxw0k371ckxrp6i77jn93ckhslaazwbyd85m83nkynr"))
206af46f 6159 (patches (search-patches "vsearch-unbundle-cityhash.patch"))
a2950fa4
BW		 6160 (snippet
6161 '(begin
206af46f
BW		 6162 ;; Remove bundled cityhash sources. The vsearch source is adjusted
6163 ;; for this in the patch.
cf6edaba
BW		 6164 (delete-file "src/city.h")
6165 (delete-file "src/citycrc.h")
6166 (delete-file "src/city.cc")
a2950fa4
BW		 6167 #t))))
6168 (build-system gnu-build-system)
6169 (arguments
6170 `(#:phases
6171 (modify-phases %standard-phases
d10092b8
KK		 6172 (add-after 'unpack 'autogen
6173 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
a2950fa4
BW		 6174 (inputs
6175 `(("zlib" ,zlib)
6176 ("bzip2" ,bzip2)
6177 ("cityhash" ,cityhash)))
6178 (native-inputs
6179 `(("autoconf" ,autoconf)
6180 ("automake" ,automake)))
6181 (synopsis "Sequence search tools for metagenomics")
6182 (description
6183 "VSEARCH supports DNA sequence searching, clustering, chimera detection,
6184dereplication, pairwise alignment, shuffling, subsampling, sorting and
6185masking. The tool takes advantage of parallelism in the form of SIMD
6186vectorization as well as multiple threads to perform accurate alignments at
6187high speed. VSEARCH uses an optimal global aligner (full dynamic programming
6188Needleman-Wunsch).")
6189 (home-page "https://github.com/torognes/vsearch")
6f04e515
BW		 6190 ;; vsearch uses non-portable SSE intrinsics so building fails on other
6191 ;; platforms.
6192 (supported-systems '("x86_64-linux"))
a2950fa4
BW		 6193 ;; Dual licensed; also includes public domain source.
6194 (license (list license:gpl3 license:bsd-2))))
6195
07837874
RW		 6196(define-public pardre
6197 (package
6198 (name "pardre")
7922ab8f
BW		 6199 ;; The source of 1.1.5 changed in place, so we append "-1" to the version.
6200 (version "1.1.5-1")
07837874
RW		 6201 (source
6202 (origin
6203 (method url-fetch)
6204 (uri (string-append "mirror://sourceforge/pardre/ParDRe-rel"
7922ab8f 6205 "1.1.5" ".tar.gz"))
07837874
RW		 6206 (sha256
6207 (base32
7922ab8f 6208 "17j73nc0viq4f6qj50nrndsrif5d6b71q8fl87m54psiv0ilns2b"))))
07837874
RW		 6209 (build-system gnu-build-system)
6210 (arguments
6211 `(#:tests? #f ; no tests included
6212 #:phases
6213 (modify-phases %standard-phases
6214 (delete 'configure)
6215 (replace 'install
6216 (lambda* (#:key outputs #:allow-other-keys)
6217 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
07837874
RW		 6218 (install-file "ParDRe" bin)
6219 #t))))))
6220 (inputs
6221 `(("openmpi" ,openmpi)
6222 ("zlib" ,zlib)))
6223 (synopsis "Parallel tool to remove duplicate DNA reads")
6224 (description
6225 "ParDRe is a parallel tool to remove duplicate genetic sequence reads.
6226Duplicate reads can be seen as identical or nearly identical sequences with
6227some mismatches. This tool lets users avoid the analysis of unnecessary
6228reads, reducing the time of subsequent procedures with the
6229dataset (e.g. assemblies, mappings, etc.). The tool is implemented with MPI
6230in order to exploit the parallel capabilities of multicore clusters. It is
6231faster than multithreaded counterparts (end of 2015) for the same number of
6232cores and, thanks to the message-passing technology, it can be executed on
6233clusters.")
6234 (home-page "https://sourceforge.net/projects/pardre/")
6235 (license license:gpl3+)))
6236
e4a44a6a
BW		 6237(define-public ruby-bio-kseq
6238 (package
6239 (name "ruby-bio-kseq")
6240 (version "0.0.2")
6241 (source
6242 (origin
6243 (method url-fetch)
6244 (uri (rubygems-uri "bio-kseq" version))
6245 (sha256
6246 (base32
6247 "1xyaha46khb5jc6wzkbf7040jagac49jbimn0vcrzid0j8jdikrz"))))
6248 (build-system ruby-build-system)
6249 (arguments
6250 `(#:test-target "spec"))
6251 (native-inputs
6252 `(("bundler" ,bundler)
6253 ("ruby-rspec" ,ruby-rspec)
6254 ("ruby-rake-compiler" ,ruby-rake-compiler)))
6255 (inputs
6256 `(("zlib" ,zlib)))
6257 (synopsis "Ruby bindings for the kseq.h FASTA/Q parser")
6258 (description
6259 "@code{Bio::Kseq} provides ruby bindings to the @code{kseq.h} FASTA and
6260FASTQ parsing code. It provides a fast iterator over sequences and their
6261quality scores.")
6262 (home-page "https://github.com/gusevfe/bio-kseq")
6263 (license license:expat)))
6264
9c38b540
PP		 6265(define-public bio-locus
6266 (package
6267 (name "bio-locus")
6268 (version "0.0.7")
6269 (source
6270 (origin
6271 (method url-fetch)
6272 (uri (rubygems-uri "bio-locus" version))
6273 (sha256
6274 (base32
6275 "02vmrxyimkj9sahsp4zhfhnmbvz6dbbqz1y01vglf8cbwvkajfl0"))))
6276 (build-system ruby-build-system)
6277 (native-inputs
6278 `(("ruby-rspec" ,ruby-rspec)))
6279 (synopsis "Tool for fast querying of genome locations")
6280 (description
6281 "Bio-locus is a tabix-like tool for fast querying of genome
6282locations. Many file formats in bioinformatics contain records that
6283start with a chromosome name and a position for a SNP, or a start-end
6284position for indels. Bio-locus allows users to store this chr+pos or
6285chr+pos+alt information in a database.")
6286 (home-page "https://github.com/pjotrp/bio-locus")
6287 (license license:expat)))
edb15985 6288
b2bddb07
PP		 6289(define-public bio-blastxmlparser
6290 (package
6291 (name "bio-blastxmlparser")
6292 (version "2.0.4")
6293 (source (origin
6294 (method url-fetch)
6295 (uri (rubygems-uri "bio-blastxmlparser" version))
6296 (sha256
6297 (base32
6298 "1wf4qygcmdjgcqm6flmvsagfr1gs9lf63mj32qv3z1f481zc5692"))))
6299 (build-system ruby-build-system)
6300 (propagated-inputs
6301 `(("ruby-bio-logger" ,ruby-bio-logger)
6302 ("ruby-nokogiri" ,ruby-nokogiri)))
6303 (inputs
6304 `(("ruby-rspec" ,ruby-rspec)))
6305 (synopsis "Fast big data BLAST XML parser and library")
6306 (description
6307 "Very fast parallel big-data BLAST XML file parser which can be used as
6308command line utility. Use blastxmlparser to: Parse BLAST XML; filter output;
6309generate FASTA, JSON, YAML, RDF, JSON-LD, HTML, CSV, tabular output etc.")
7bf837fd 6310 (home-page "https://github.com/pjotrp/blastxmlparser")
b2bddb07
PP		 6311 (license license:expat)))
6312
edb15985
PP		 6313(define-public bioruby
6314 (package
6315 (name "bioruby")
dbf9d371 6316 (version "1.5.1")
edb15985
PP		 6317 (source
6318 (origin
6319 (method url-fetch)
6320 (uri (rubygems-uri "bio" version))
6321 (sha256
6322 (base32
dbf9d371 6323 "0hdl0789c9n4mprnx5pgd46bfwl8d000rqpamj5h6kkjgspijv49"))))
edb15985
PP		 6324 (build-system ruby-build-system)
6325 (propagated-inputs
6326 `(("ruby-libxml" ,ruby-libxml)))
6327 (native-inputs
6328 `(("which" ,which))) ; required for test phase
6329 (arguments
6330 `(#:phases
6331 (modify-phases %standard-phases
6332 (add-before 'build 'patch-test-command
6333 (lambda _
6334 (substitute* '("test/functional/bio/test_command.rb")
6335 (("/bin/sh") (which "sh")))
6336 (substitute* '("test/functional/bio/test_command.rb")
6337 (("/bin/ls") (which "ls")))
6338 (substitute* '("test/functional/bio/test_command.rb")
6339 (("which") (which "which")))
6340 (substitute* '("test/functional/bio/test_command.rb",
6341 "test/data/command/echoarg2.sh")
6342 (("/bin/echo") (which "echo")))
6343 #t)))))
6344 (synopsis "Ruby library, shell and utilities for bioinformatics")
6345 (description "BioRuby comes with a comprehensive set of Ruby development
6346tools and libraries for bioinformatics and molecular biology. BioRuby has
6347components for sequence analysis, pathway analysis, protein modelling and
6348phylogenetic analysis; it supports many widely used data formats and provides
6349easy access to databases, external programs and public web services, including
6350BLAST, KEGG, GenBank, MEDLINE and GO.")
6351 (home-page "http://bioruby.org/")
6352 ;; Code is released under Ruby license, except for setup
6353 ;; (LGPLv2.1+) and scripts in samples (which have GPL2 and GPL2+)
6354 (license (list license:ruby license:lgpl2.1+ license:gpl2+ ))))
a5002ae7 6355
9fba89e8
RW		 6356(define-public r-acsnminer
6357 (package
6358 (name "r-acsnminer")
0b54b4c9 6359 (version "0.16.8.25")
9fba89e8
RW		 6360 (source (origin
6361 (method url-fetch)
6362 (uri (cran-uri "ACSNMineR" version))
6363 (sha256
6364 (base32
0b54b4c9 6365 "0gh604s8qall6zfjlwcg2ilxjvz08dplf9k5g47idhv43scm748l"))))
9fba89e8
RW		 6366 (properties `((upstream-name . "ACSNMineR")))
6367 (build-system r-build-system)
6368 (propagated-inputs
6369 `(("r-ggplot2" ,r-ggplot2)
6370 ("r-gridextra" ,r-gridextra)))
6371 (home-page "http://cran.r-project.org/web/packages/ACSNMineR")
6372 (synopsis "Gene enrichment analysis")
6373 (description
6374 "This package provides tools to compute and represent gene set enrichment
6375or depletion from your data based on pre-saved maps from the @dfn{Atlas of
6376Cancer Signalling Networks} (ACSN) or user imported maps. The gene set
6377enrichment can be run with hypergeometric test or Fisher exact test, and can
6378use multiple corrections. Visualization of data can be done either by
6379barplots or heatmaps.")
6380 (license license:gpl2+)))
6381
d29b25c4
RW		 6382(define-public r-biocgenerics
6383 (package
6384 (name "r-biocgenerics")
d61fc4e1 6385 (version "0.24.0")
d29b25c4
RW		 6386 (source (origin
6387 (method url-fetch)
6388 (uri (bioconductor-uri "BiocGenerics" version))
6389 (sha256
6390 (base32
d61fc4e1 6391 "03wxvhxyrhipbgcg83lqlfn7p9gbzzrnl48y0dq7303xgp232zai"))))
d29b25c4 6392 (properties
1d216b6e 6393 `((upstream-name . "BiocGenerics")))
d29b25c4 6394 (build-system r-build-system)
5713bbf1 6395 (home-page "https://bioconductor.org/packages/BiocGenerics")
d29b25c4
RW		 6396 (synopsis "S4 generic functions for Bioconductor")
6397 (description
6398 "This package provides S4 generic functions needed by many Bioconductor
6399packages.")
6400 (license license:artistic2.0)))
6401
eb24341f
RJ		 6402(define-public r-biocinstaller
6403 (package
6404 (name "r-biocinstaller")
32414616 6405 (version "1.28.0")
eb24341f
RJ		 6406 (source (origin
6407 (method url-fetch)
6408 (uri (bioconductor-uri "BiocInstaller" version))
6409 (sha256
6410 (base32
32414616 6411 "19fga27bv6q9v5mpil74y76lahmnwvpg2h33rdx1r79nvljkd19d"))))
eb24341f
RJ		 6412 (properties
6413 `((upstream-name . "BiocInstaller")))
6414 (build-system r-build-system)
5713bbf1 6415 (home-page "https://bioconductor.org/packages/BiocInstaller")
eb24341f
RJ		 6416 (synopsis "Install Bioconductor packages")
6417 (description "This package is used to install and update R packages from
6418Bioconductor, CRAN, and Github.")
6419 (license license:artistic2.0)))
6420
207ce8fb
RJ		 6421(define-public r-biocviews
6422 (package
6423 (name "r-biocviews")
c7782c3f 6424 (version "1.46.0")
207ce8fb
RJ		 6425 (source (origin
6426 (method url-fetch)
6427 (uri (bioconductor-uri "biocViews" version))
6428 (sha256
6429 (base32
c7782c3f 6430 "09zyqj1kqc089lmh9sliy0acanx9zimcasvp71dsrg2bqm08r1md"))))
207ce8fb
RJ		 6431 (properties
6432 `((upstream-name . "biocViews")))
6433 (build-system r-build-system)
6434 (propagated-inputs
6435 `(("r-biobase" ,r-biobase)
6436 ("r-graph" ,r-graph)
6437 ("r-rbgl" ,r-rbgl)
6438 ("r-rcurl" ,r-rcurl)
6439 ("r-xml" ,r-xml)
207ce8fb 6440 ("r-runit" ,r-runit)))
5713bbf1 6441 (home-page "https://bioconductor.org/packages/biocViews")
207ce8fb
RJ		 6442 (synopsis "Bioconductor package categorization helper")
6443 (description "The purpose of biocViews is to create HTML pages that
6444categorize packages in a Bioconductor package repository according to keywords,
6445also known as views, in a controlled vocabulary.")
6446 (license license:artistic2.0)))
6447
2abfc5b8
RJ		 6448(define-public r-bookdown
6449 (package
6450 (name "r-bookdown")
6bd27981 6451 (version "0.5")
2abfc5b8
RJ		 6452 (source (origin
6453 (method url-fetch)
6454 (uri (cran-uri "bookdown" version))
6455 (sha256
6456 (base32
6bd27981 6457 "0zm63kr4f4kja4qpwkzl119zzyciqj7ihajfqgfjpgb4dzaiycxp"))))
2abfc5b8
RJ		 6458 (build-system r-build-system)
6459 (propagated-inputs
6460 `(("r-htmltools" ,r-htmltools)
6461 ("r-knitr" ,r-knitr)
6462 ("r-rmarkdown" ,r-rmarkdown)
6463 ("r-yaml" ,r-yaml)))
6464 (home-page "https://github.com/rstudio/bookdown")
6465 (synopsis "Authoring books and technical documents with R markdown")
6466 (description "This package provides output formats and utilities for
6467authoring books and technical documents with R Markdown.")
6468 (license license:gpl3)))
6469
99df12cd
RJ		 6470(define-public r-biocstyle
6471 (package
6472 (name "r-biocstyle")
2167eafc 6473 (version "2.6.0")
99df12cd
RJ		 6474 (source (origin
6475 (method url-fetch)
6476 (uri (bioconductor-uri "BiocStyle" version))
6477 (sha256
6478 (base32
2167eafc 6479 "05f2j9fx8s5gh4f8qkl6wcz32ghz04wxhqb3xxcn1bj24qd7x1x8"))))
99df12cd
RJ		 6480 (properties
6481 `((upstream-name . "BiocStyle")))
6482 (build-system r-build-system)
3bef24c9
RJ		 6483 (propagated-inputs
6484 `(("r-bookdown" ,r-bookdown)
6485 ("r-knitr" ,r-knitr)
6486 ("r-rmarkdown" ,r-rmarkdown)
6487 ("r-yaml" ,r-yaml)))
5713bbf1 6488 (home-page "https://bioconductor.org/packages/BiocStyle")
99df12cd
RJ		 6489 (synopsis "Bioconductor formatting styles")
6490 (description "This package provides standard formatting styles for
6491Bioconductor PDF and HTML documents. Package vignettes illustrate use and
6492functionality.")
6493 (license license:artistic2.0)))
6494
4644644a
RJ		 6495(define-public r-bioccheck
6496 (package
6497 (name "r-bioccheck")
7373b416 6498 (version "1.14.0")
4644644a
RJ		 6499 (source (origin
6500 (method url-fetch)
6501 (uri (bioconductor-uri "BiocCheck" version))
6502 (sha256
6503 (base32
7373b416 6504 "1nzp8kgw13z9pgf885rplj6k37jcldfhbz0adqclxr2gq0yalmyx"))))
4644644a
RJ		 6505 (properties
6506 `((upstream-name . "BiocCheck")))
6507 (build-system r-build-system)
6508 (arguments
6509 '(#:phases
6510 (modify-phases %standard-phases
6511 ;; This package can be used by calling BiocCheck(<package>) from
6512 ;; within R, or by running R CMD BiocCheck <package>. This phase
6513 ;; makes sure the latter works. For this to work, the BiocCheck
6514 ;; script must be somewhere on the PATH (not the R bin directory).
6515 (add-after 'install 'install-bioccheck-subcommand
6516 (lambda* (#:key outputs #:allow-other-keys)
6517 (let* ((out (assoc-ref outputs "out"))
6518 (dest-dir (string-append out "/bin"))
6519 (script-dir
6520 (string-append out "/site-library/BiocCheck/script/")))
6521 (mkdir-p dest-dir)
6522 (symlink (string-append script-dir "/checkBadDeps.R")
6523 (string-append dest-dir "/checkBadDeps.R"))
6524 (symlink (string-append script-dir "/BiocCheck")
6525 (string-append dest-dir "/BiocCheck")))
6526 #t)))))
4644644a 6527 (propagated-inputs
aeb64f3c
RW		 6528 `(("r-codetools" ,r-codetools)
6529 ("r-graph" ,r-graph)
4644644a
RJ		 6530 ("r-httr" ,r-httr)
6531 ("r-optparse" ,r-optparse)
4644644a 6532 ("r-biocinstaller" ,r-biocinstaller)
7373b416
RW		 6533 ("r-biocviews" ,r-biocviews)
6534 ("r-stringdist" ,r-stringdist)))
5713bbf1 6535 (home-page "https://bioconductor.org/packages/BiocCheck")
4644644a
RJ		 6536 (synopsis "Executes Bioconductor-specific package checks")
6537 (description "This package contains tools to perform additional quality
6538checks on R packages that are to be submitted to the Bioconductor repository.")
6539 (license license:artistic2.0)))
6540
2acaaee5
RJ		 6541(define-public r-getopt
6542 (package
6543 (name "r-getopt")
6544 (version "1.20.0")
6545 (source
6546 (origin
6547 (method url-fetch)
6548 (uri (cran-uri "getopt" version))
6549 (sha256
6550 (base32
6551 "00f57vgnzmg7cz80rjmjz1556xqcmx8nhrlbbhaq4w7gl2ibl87r"))))
6552 (build-system r-build-system)
6553 (home-page "https://github.com/trevorld/getopt")
6554 (synopsis "Command-line option processor for R")
6555 (description
6556 "This package is designed to be used with Rscript to write shebang
6557scripts that accept short and long options. Many users will prefer to
6558use the packages @code{optparse} or @code{argparse} which add extra
6559features like automatically generated help options and usage texts,
6560support for default values, positional argument support, etc.")
6561 (license license:gpl2+)))
6562
c79ad57a
RJ		 6563(define-public r-optparse
6564 (package
6565 (name "r-optparse")
7150f1c3 6566 (version "1.4.4")
c79ad57a
RJ		 6567 (source
6568 (origin
6569 (method url-fetch)
6570 (uri (cran-uri "optparse" version))
6571 (sha256
6572 (base32
7150f1c3 6573 "1ff4wmsszrb3spwfp7ynfs8w11qpy1sdzfxm1wk8dqqvdwris7qb"))))
c79ad57a
RJ		 6574 (build-system r-build-system)
6575 (propagated-inputs
6576 `(("r-getopt" ,r-getopt)))
6577 (home-page
6578 "https://github.com/trevorld/optparse")
6579 (synopsis "Command line option parser")
6580 (description
6581 "This package provides a command line parser inspired by Python's
6582@code{optparse} library to be used with Rscript to write shebang scripts
6583that accept short and long options.")
6584 (license license:gpl2+)))
6585
247d498a
RJ		 6586(define-public r-dnacopy
6587 (package
6588 (name "r-dnacopy")
94fd86e5 6589 (version "1.52.0")
247d498a
RJ		 6590 (source (origin
6591 (method url-fetch)
6592 (uri (bioconductor-uri "DNAcopy" version))
6593 (sha256
6594 (base32
94fd86e5 6595 "127il5rlg1hzjlhwhs64x3nm18p00q1pd9ckb2b9ifl0rax95wai"))))
247d498a
RJ		 6596 (properties
6597 `((upstream-name . "DNAcopy")))
6598 (build-system r-build-system)
6599 (inputs
6600 `(("gfortran" ,gfortran)))
5697fdc3 6601 (home-page "https://bioconductor.org/packages/DNAcopy")
247d498a
RJ		 6602 (synopsis "Implementation of a circular binary segmentation algorithm")
6603 (description "This package implements the circular binary segmentation (CBS)
6604algorithm to segment DNA copy number data and identify genomic regions with
6605abnormal copy number.")
6606 (license license:gpl2+)))
6607
7485129e
RW		 6608(define-public r-s4vectors
6609 (package
6610 (name "r-s4vectors")
41f0f949 6611 (version "0.16.0")
7485129e
RW		 6612 (source (origin
6613 (method url-fetch)
6614 (uri (bioconductor-uri "S4Vectors" version))
6615 (sha256
6616 (base32
41f0f949 6617 "03s8vz33nl6mivjb7dbvj702dkypi340lji1sjban03fyyls0hw0"))))
7485129e 6618 (properties
1d216b6e 6619 `((upstream-name . "S4Vectors")))
7485129e
RW		 6620 (build-system r-build-system)
6621 (propagated-inputs
6622 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6623 (home-page "https://bioconductor.org/packages/S4Vectors")
7485129e
RW		 6624 (synopsis "S4 implementation of vectors and lists")
6625 (description
6626 "The S4Vectors package defines the @code{Vector} and @code{List} virtual
6627classes and a set of generic functions that extend the semantic of ordinary
6628vectors and lists in R. Package developers can easily implement vector-like
6629or list-like objects as concrete subclasses of @code{Vector} or @code{List}.
6630In addition, a few low-level concrete subclasses of general interest (e.g.
6631@code{DataFrame}, @code{Rle}, and @code{Hits}) are implemented in the
6632S4Vectors package itself.")
6633 (license license:artistic2.0)))
6634
274da826
RW		 6635(define-public r-seqinr
6636 (package
6637 (name "r-seqinr")
023aa8ff 6638 (version "3.4-5")
274da826
RW		 6639 (source
6640 (origin
6641 (method url-fetch)
6642 (uri (cran-uri "seqinr" version))
6643 (sha256
6644 (base32
023aa8ff 6645 "17zv0n5cji17izwmwg0jcbxbjl3w5rls91w15svcnlpxjms38ahn"))))
274da826
RW		 6646 (build-system r-build-system)
6647 (propagated-inputs
3b851cd4
RW		 6648 `(("r-ade4" ,r-ade4)
6649 ("r-segmented" ,r-segmented)))
274da826
RW		 6650 (inputs
6651 `(("zlib" ,zlib)))
6652 (home-page "http://seqinr.r-forge.r-project.org/")
6653 (synopsis "Biological sequences retrieval and analysis")
6654 (description
6655 "This package provides tools for exploratory data analysis and data
6656visualization of biological sequence (DNA and protein) data. It also includes
6657utilities for sequence data management under the ACNUC system.")
6658 (license license:gpl2+)))
6659
78addcb0
RW		 6660(define-public r-iranges
6661 (package
6662 (name "r-iranges")
9e482c20 6663 (version "2.12.0")
78addcb0
RW		 6664 (source (origin
6665 (method url-fetch)
6666 (uri (bioconductor-uri "IRanges" version))
6667 (sha256
6668 (base32
9e482c20 6669 "1vqczb9wlxsmpwpqig6j1dmiblcfpq6mgnq8qwzcrvddm4cp47m5"))))
78addcb0 6670 (properties
1d216b6e 6671 `((upstream-name . "IRanges")))
78addcb0
RW		 6672 (build-system r-build-system)
6673 (propagated-inputs
6674 `(("r-biocgenerics" ,r-biocgenerics)
6675 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6676 (home-page "https://bioconductor.org/packages/IRanges")
78addcb0
RW		 6677 (synopsis "Infrastructure for manipulating intervals on sequences")
6678 (description
6679 "This package provides efficient low-level and highly reusable S4 classes
6680for storing ranges of integers, RLE vectors (Run-Length Encoding), and, more
6681generally, data that can be organized sequentially (formally defined as
6682@code{Vector} objects), as well as views on these @code{Vector} objects.
6683Efficient list-like classes are also provided for storing big collections of
6684instances of the basic classes. All classes in the package use consistent
6685naming and share the same rich and consistent \"Vector API\" as much as
6686possible.")
6687 (license license:artistic2.0)))
6688
ffef27f3
RJ		 6689(define-public r-genomeinfodbdata
6690 (package
6691 (name "r-genomeinfodbdata")
261b38a9 6692 (version "0.99.1")
ffef27f3
RJ		 6693 (source (origin
6694 (method url-fetch)
90f83099
EF		 6695 ;; We cannot use bioconductor-uri here because this tarball is
6696 ;; located under "data/annotation/" instead of "bioc/".
6697 (uri (string-append "https://bioconductor.org/packages/release/"
6698 "data/annotation/src/contrib/GenomeInfoDbData_"
6699 version ".tar.gz"))
ffef27f3
RJ		 6700 (sha256
6701 (base32
261b38a9 6702 "0hipipvyvrh75n68hsjg35sxbcfzrghzxv547vnkk2f8ya99g01r"))))
ffef27f3
RJ		 6703 (properties
6704 `((upstream-name . "GenomeInfoDbData")))
6705 (build-system r-build-system)
5713bbf1 6706 (home-page "https://bioconductor.org/packages/GenomeInfoDbData")
ffef27f3
RJ		 6707 (synopsis "Species and taxonomy ID look up tables for GenomeInfoDb")
6708 (description "This package contains data for mapping between NCBI taxonomy
6709ID and species. It is used by functions in the GenomeInfoDb package.")
6710 (license license:artistic2.0)))
6711
bf7764b7
RW		 6712(define-public r-genomeinfodb
6713 (package
6714 (name "r-genomeinfodb")
8aab0235 6715 (version "1.14.0")
bf7764b7
RW		 6716 (source (origin
6717 (method url-fetch)
6718 (uri (bioconductor-uri "GenomeInfoDb" version))
6719 (sha256
6720 (base32
8aab0235 6721 "1jhm0imkac4gvchbjxj408aakk39xdv2fyh818d3lk295bz6bnyp"))))
bf7764b7 6722 (properties
1d216b6e 6723 `((upstream-name . "GenomeInfoDb")))
bf7764b7
RW		 6724 (build-system r-build-system)
6725 (propagated-inputs
6726 `(("r-biocgenerics" ,r-biocgenerics)
38b99ccc 6727 ("r-genomeinfodbdata" ,r-genomeinfodbdata)
bf7764b7 6728 ("r-iranges" ,r-iranges)
4cd07e48 6729 ("r-rcurl" ,r-rcurl)
bf7764b7 6730 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6731 (home-page "https://bioconductor.org/packages/GenomeInfoDb")
bf7764b7
RW		 6732 (synopsis "Utilities for manipulating chromosome identifiers")
6733 (description
6734 "This package contains data and functions that define and allow
6735translation between different chromosome sequence naming conventions (e.g.,
6736\"chr1\" versus \"1\"), including a function that attempts to place sequence
6737names in their natural, rather than lexicographic, order.")
6738 (license license:artistic2.0)))
6739
744004a3
RJ		 6740(define-public r-edger
6741 (package
6742 (name "r-edger")
010ab2ff 6743 (version "3.20.1")
744004a3
RJ		 6744 (source (origin
6745 (method url-fetch)
6746 (uri (bioconductor-uri "edgeR" version))
6747 (sha256
6748 (base32
010ab2ff 6749 "01qnxwr9rmz8r5ga3hvjk632365ga2aygx71mxkk7jiad2pjznsp"))))
744004a3
RJ		 6750 (properties `((upstream-name . "edgeR")))
6751 (build-system r-build-system)
6752 (propagated-inputs
5e48005f 6753 `(("r-limma" ,r-limma)
47055b27 6754 ("r-locfit" ,r-locfit)
010ab2ff 6755 ("r-rcpp" ,r-rcpp)
47055b27 6756 ("r-statmod" ,r-statmod))) ;for estimateDisp
744004a3
RJ		 6757 (home-page "http://bioinf.wehi.edu.au/edgeR")
6758 (synopsis "EdgeR does empirical analysis of digital gene expression data")
6759 (description "This package can do differential expression analysis of
6760RNA-seq expression profiles with biological replication. It implements a range
6761of statistical methodology based on the negative binomial distributions,
6762including empirical Bayes estimation, exact tests, generalized linear models
6763and quasi-likelihood tests. It be applied to differential signal analysis of
6764other types of genomic data that produce counts, including ChIP-seq, SAGE and
6765CAGE.")
6766 (license license:gpl2+)))
6767
b669d9c4
RJ		 6768(define-public r-variantannotation
6769 (package
6770 (name "r-variantannotation")
7f1bac35 6771 (version "1.24.0")
b669d9c4
RJ		 6772 (source (origin
6773 (method url-fetch)
6774 (uri (bioconductor-uri "VariantAnnotation" version))
6775 (sha256
6776 (base32
7f1bac35 6777 "1lllp2vgyfbrar1yg28ji7am470hfzrzxm1bgdk68xpnrwcgcl25"))))
b669d9c4
RJ		 6778 (properties
6779 `((upstream-name . "VariantAnnotation")))
6780 (inputs
6781 `(("zlib" ,zlib)))
6782 (propagated-inputs
6783 `(("r-annotationdbi" ,r-annotationdbi)
37d96f1d 6784 ("r-biobase" ,r-biobase)
b669d9c4 6785 ("r-biocgenerics" ,r-biocgenerics)
37d96f1d 6786 ("r-biostrings" ,r-biostrings)
b669d9c4
RJ		 6787 ("r-bsgenome" ,r-bsgenome)
6788 ("r-dbi" ,r-dbi)
6789 ("r-genomeinfodb" ,r-genomeinfodb)
6790 ("r-genomicfeatures" ,r-genomicfeatures)
6791 ("r-genomicranges" ,r-genomicranges)
37d96f1d 6792 ("r-iranges" ,r-iranges)
b669d9c4
RJ		 6793 ("r-summarizedexperiment" ,r-summarizedexperiment)
6794 ("r-rsamtools" ,r-rsamtools)
37d96f1d
RW		 6795 ("r-rtracklayer" ,r-rtracklayer)
6796 ("r-s4vectors" ,r-s4vectors)
6797 ("r-xvector" ,r-xvector)
b669d9c4
RJ		 6798 ("r-zlibbioc" ,r-zlibbioc)))
6799 (build-system r-build-system)
6800 (home-page "https://bioconductor.org/packages/VariantAnnotation")
6801 (synopsis "Package for annotation of genetic variants")
6802 (description "This R package can annotate variants, compute amino acid
6803coding changes and predict coding outcomes.")
6804 (license license:artistic2.0)))
6805
7d4224d7
RJ		 6806(define-public r-limma
6807 (package
6808 (name "r-limma")
f642c0e4 6809 (version "3.34.0")
7d4224d7
RJ		 6810 (source (origin
6811 (method url-fetch)
6812 (uri (bioconductor-uri "limma" version))
6813 (sha256
6814 (base32
f642c0e4 6815 "0a15gsaky0hfrkx8wrrmp0labzxpq6m2hrd33zl206wyas8bqzcs"))))
7d4224d7
RJ		 6816 (build-system r-build-system)
6817 (home-page "http://bioinf.wehi.edu.au/limma")
6818 (synopsis "Package for linear models for microarray and RNA-seq data")
6819 (description "This package can be used for the analysis of gene expression
6820studies, especially the use of linear models for analysing designed experiments
6821and the assessment of differential expression. The analysis methods apply to
6822different technologies, including microarrays, RNA-seq, and quantitative PCR.")
6823 (license license:gpl2+)))
6824
0e7d5560
RW		 6825(define-public r-xvector
6826 (package
6827 (name "r-xvector")
2b6ae8bf 6828 (version "0.18.0")
0e7d5560
RW		 6829 (source (origin
6830 (method url-fetch)
6831 (uri (bioconductor-uri "XVector" version))
6832 (sha256
6833 (base32
2b6ae8bf 6834 "1i4i3kdxr78lr1kcxq657p11ybi7kq10c8kyaqyh6gfc8i9rhvmk"))))
0e7d5560 6835 (properties
1d216b6e 6836 `((upstream-name . "XVector")))
0e7d5560
RW		 6837 (build-system r-build-system)
6838 (arguments
6839 `(#:phases
6840 (modify-phases %standard-phases
6841 (add-after 'unpack 'use-system-zlib
6842 (lambda _
6843 (substitute* "DESCRIPTION"
6844 (("zlibbioc, ") ""))
6845 (substitute* "NAMESPACE"
6846 (("import\\(zlibbioc\\)") ""))
6847 #t)))))
6848 (inputs
6849 `(("zlib" ,zlib)))
6850 (propagated-inputs
6851 `(("r-biocgenerics" ,r-biocgenerics)
6852 ("r-iranges" ,r-iranges)
6853 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6854 (home-page "https://bioconductor.org/packages/XVector")
0e7d5560
RW		 6855 (synopsis "Representation and manpulation of external sequences")
6856 (description
6857 "This package provides memory efficient S4 classes for storing sequences
6858\"externally\" (behind an R external pointer, or on disk).")
6859 (license license:artistic2.0)))
6860
e2cd1d0f
RW		 6861(define-public r-genomicranges
6862 (package
6863 (name "r-genomicranges")
aa5c1f61 6864 (version "1.30.0")
e2cd1d0f
RW		 6865 (source (origin
6866 (method url-fetch)
6867 (uri (bioconductor-uri "GenomicRanges" version))
6868 (sha256
6869 (base32
aa5c1f61 6870 "10ra2sjn17h6gilm9iz0cygp9ijpgbirljlc4drwrnivnw9cmi2a"))))
e2cd1d0f 6871 (properties
1d216b6e 6872 `((upstream-name . "GenomicRanges")))
e2cd1d0f
RW		 6873 (build-system r-build-system)
6874 (propagated-inputs
6875 `(("r-biocgenerics" ,r-biocgenerics)
6876 ("r-genomeinfodb" ,r-genomeinfodb)
92a740af
RW		 6877 ("r-iranges" ,r-iranges)
6878 ("r-s4vectors" ,r-s4vectors)
e2cd1d0f 6879 ("r-xvector" ,r-xvector)))
5713bbf1 6880 (home-page "https://bioconductor.org/packages/GenomicRanges")
e2cd1d0f
RW		 6881 (synopsis "Representation and manipulation of genomic intervals")
6882 (description
6883 "This package provides tools to efficiently represent and manipulate
6884genomic annotations and alignments is playing a central role when it comes to
6885analyzing high-throughput sequencing data (a.k.a. NGS data). The
6886GenomicRanges package defines general purpose containers for storing and
6887manipulating genomic intervals and variables defined along a genome.")
6888 (license license:artistic2.0)))
6889
555e3399
RW		 6890(define-public r-biobase
6891 (package
6892 (name "r-biobase")
ca521236 6893 (version "2.38.0")
555e3399
RW		 6894 (source (origin
6895 (method url-fetch)
6896 (uri (bioconductor-uri "Biobase" version))
6897 (sha256
6898 (base32
ca521236 6899 "1cgm1ja1kp56zdlzyy9ggbkfn8r2vbsd4hncmz8g4hjd47fg18kg"))))
555e3399
RW		 6900 (properties
6901 `((upstream-name . "Biobase")))
6902 (build-system r-build-system)
6903 (propagated-inputs
6904 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6905 (home-page "https://bioconductor.org/packages/Biobase")
555e3399
RW		 6906 (synopsis "Base functions for Bioconductor")
6907 (description
6908 "This package provides functions that are needed by many other packages
6909on Bioconductor or which replace R functions.")
6910 (license license:artistic2.0)))
6911
8b7bce74
RW		 6912(define-public r-annotationdbi
6913 (package
6914 (name "r-annotationdbi")
0f8d98f2 6915 (version "1.40.0")
8b7bce74
RW		 6916 (source (origin
6917 (method url-fetch)
6918 (uri (bioconductor-uri "AnnotationDbi" version))
6919 (sha256
6920 (base32
0f8d98f2 6921 "1dh4qs1a757n640gs34lf6z2glc96nan86x0sqaw5csadl2rhnlc"))))
8b7bce74
RW		 6922 (properties
6923 `((upstream-name . "AnnotationDbi")))
6924 (build-system r-build-system)
6925 (propagated-inputs
6926 `(("r-biobase" ,r-biobase)
6927 ("r-biocgenerics" ,r-biocgenerics)
6928 ("r-dbi" ,r-dbi)
6929 ("r-iranges" ,r-iranges)
6930 ("r-rsqlite" ,r-rsqlite)
6931 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6932 (home-page "https://bioconductor.org/packages/AnnotationDbi")
8b7bce74
RW		 6933 (synopsis "Annotation database interface")
6934 (description
6935 "This package provides user interface and database connection code for
6936annotation data packages using SQLite data storage.")
6937 (license license:artistic2.0)))
6938
c465fa72
RW		 6939(define-public r-biomart
6940 (package
6941 (name "r-biomart")
42e11d33 6942 (version "2.34.0")
c465fa72
RW		 6943 (source (origin
6944 (method url-fetch)
6945 (uri (bioconductor-uri "biomaRt" version))
6946 (sha256
6947 (base32
42e11d33 6948 "1dn3ysf0vb3mmg2b3380g0j1ajf88x4rh7fddfp990h2xlnsy2cx"))))
c465fa72
RW		 6949 (properties
6950 `((upstream-name . "biomaRt")))
6951 (build-system r-build-system)
6952 (propagated-inputs
6953 `(("r-annotationdbi" ,r-annotationdbi)
42e11d33 6954 ("r-progress" ,r-progress)
c465fa72 6955 ("r-rcurl" ,r-rcurl)
42e11d33 6956 ("r-stringr" ,r-stringr)
c465fa72 6957 ("r-xml" ,r-xml)))
5713bbf1 6958 (home-page "https://bioconductor.org/packages/biomaRt")
c465fa72
RW		 6959 (synopsis "Interface to BioMart databases")
6960 (description
6961 "biomaRt provides an interface to a growing collection of databases
6962implementing the @url{BioMart software suite, http://www.biomart.org}. The
6963package enables retrieval of large amounts of data in a uniform way without
6964the need to know the underlying database schemas or write complex SQL queries.
6965Examples of BioMart databases are Ensembl, COSMIC, Uniprot, HGNC, Gramene,
6966Wormbase and dbSNP mapped to Ensembl. These major databases give biomaRt
6967users direct access to a diverse set of data and enable a wide range of
6968powerful online queries from gene annotation to database mining.")
6969 (license license:artistic2.0)))
6970
e91d362e
RW		 6971(define-public r-biocparallel
6972 (package
6973 (name "r-biocparallel")
a044c7f4 6974 (version "1.12.0")
e91d362e
RW		 6975 (source (origin
6976 (method url-fetch)
6977 (uri (bioconductor-uri "BiocParallel" version))
6978 (sha256
6979 (base32
a044c7f4 6980 "13ng3n2wsgl3fh0v6jnz3vg51k5c1sh44pqdvblcrcd1qyjmmqhd"))))
e91d362e
RW		 6981 (properties
6982 `((upstream-name . "BiocParallel")))
6983 (build-system r-build-system)
6984 (propagated-inputs
6985 `(("r-futile-logger" ,r-futile-logger)
a044c7f4
RW		 6986 ("r-snow" ,r-snow)
6987 ("r-bh" ,r-bh)))
5713bbf1 6988 (home-page "https://bioconductor.org/packages/BiocParallel")
e91d362e
RW		 6989 (synopsis "Bioconductor facilities for parallel evaluation")
6990 (description
6991 "This package provides modified versions and novel implementation of
6992functions for parallel evaluation, tailored to use with Bioconductor
6993objects.")
6994 (license (list license:gpl2+ license:gpl3+))))
6995
bf159353
RW		 6996(define-public r-biostrings
6997 (package
6998 (name "r-biostrings")
b719435e 6999 (version "2.46.0")
bf159353
RW		 7000 (source (origin
7001 (method url-fetch)
7002 (uri (bioconductor-uri "Biostrings" version))
7003 (sha256
7004 (base32
b719435e 7005 "0vg50qdlxqcm2d6axjnzg8wh8pr4c5gz03l8bdl0llmwzp0zclzk"))))
bf159353
RW		 7006 (properties
7007 `((upstream-name . "Biostrings")))
7008 (build-system r-build-system)
7009 (propagated-inputs
7010 `(("r-biocgenerics" ,r-biocgenerics)
7011 ("r-iranges" ,r-iranges)
7012 ("r-s4vectors" ,r-s4vectors)
7013 ("r-xvector" ,r-xvector)))
5713bbf1 7014 (home-page "https://bioconductor.org/packages/Biostrings")
bf159353
RW		 7015 (synopsis "String objects and algorithms for biological sequences")
7016 (description
7017 "This package provides memory efficient string containers, string
7018matching algorithms, and other utilities, for fast manipulation of large
7019biological sequences or sets of sequences.")
7020 (license license:artistic2.0)))
7021
f8d74f70
RW		 7022(define-public r-rsamtools
7023 (package
7024 (name "r-rsamtools")
f8068419 7025 (version "1.30.0")
f8d74f70
RW		 7026 (source (origin
7027 (method url-fetch)
7028 (uri (bioconductor-uri "Rsamtools" version))
7029 (sha256
7030 (base32
f8068419 7031 "0pjny5fjvbnfdyhl3bwxin678sha2drvs00sivxh3l772cn6yams"))))
f8d74f70
RW		 7032 (properties
7033 `((upstream-name . "Rsamtools")))
7034 (build-system r-build-system)
7035 (arguments
7036 `(#:phases
7037 (modify-phases %standard-phases
7038 (add-after 'unpack 'use-system-zlib
7039 (lambda _
7040 (substitute* "DESCRIPTION"
7041 (("zlibbioc, ") ""))
7042 (substitute* "NAMESPACE"
7043 (("import\\(zlibbioc\\)") ""))
7044 #t)))))
7045 (inputs
7046 `(("zlib" ,zlib)))
7047 (propagated-inputs
7048 `(("r-biocgenerics" ,r-biocgenerics)
7049 ("r-biocparallel" ,r-biocparallel)
7050 ("r-biostrings" ,r-biostrings)
7051 ("r-bitops" ,r-bitops)
7052 ("r-genomeinfodb" ,r-genomeinfodb)
7053 ("r-genomicranges" ,r-genomicranges)
7054 ("r-iranges" ,r-iranges)
7055 ("r-s4vectors" ,r-s4vectors)
7056 ("r-xvector" ,r-xvector)))
5713bbf1 7057 (home-page "https://bioconductor.org/packages/release/bioc/html/Rsamtools.html")
f8d74f70
RW		 7058 (synopsis "Interface to samtools, bcftools, and tabix")
7059 (description
7060 "This package provides an interface to the 'samtools', 'bcftools', and
7061'tabix' utilities for manipulating SAM (Sequence Alignment / Map), FASTA,
7062binary variant call (BCF) and compressed indexed tab-delimited (tabix)
7063files.")
7064 (license license:expat)))
7065
71e34e6b
RJ		 7066(define-public r-delayedarray
7067 (package
7068 (name "r-delayedarray")
cb0a9a74 7069 (version "0.4.1")
71e34e6b
RJ		 7070 (source (origin
7071 (method url-fetch)
7072 (uri (bioconductor-uri "DelayedArray" version))
7073 (sha256
7074 (base32
cb0a9a74 7075 "0s7h2giyvz04cg6248kbbzpwhxdrpnsvl2s8k5c8ricisd9aaz4b"))))
71e34e6b
RJ		 7076 (properties
7077 `((upstream-name . "DelayedArray")))
7078 (build-system r-build-system)
7079 (propagated-inputs
7080 `(("r-biocgenerics" ,r-biocgenerics)
7081 ("r-s4vectors" ,r-s4vectors)
7082 ("r-iranges" ,r-iranges)
7083 ("r-matrixstats" ,r-matrixstats)))
5713bbf1 7084 (home-page "https://bioconductor.org/packages/DelayedArray")
71e34e6b
RJ		 7085 (synopsis "Delayed operations on array-like objects")
7086 (description
7087 "Wrapping an array-like object (typically an on-disk object) in a
7088@code{DelayedArray} object allows one to perform common array operations on it
7089without loading the object in memory. In order to reduce memory usage and
7090optimize performance, operations on the object are either delayed or executed
7091using a block processing mechanism. Note that this also works on in-memory
7092array-like objects like @code{DataFrame} objects (typically with Rle columns),
7093@code{Matrix} objects, and ordinary arrays and data frames.")
7094 (license license:artistic2.0)))
7095
6e76dda2
RW		 7096(define-public r-summarizedexperiment
7097 (package
7098 (name "r-summarizedexperiment")
6fcbed73 7099 (version "1.8.0")
6e76dda2
RW		 7100 (source (origin
7101 (method url-fetch)
7102 (uri (bioconductor-uri "SummarizedExperiment" version))
7103 (sha256
7104 (base32
6fcbed73 7105 "1011r8l0k8420j31bmh4xdcp6ka5bzf4bqhip84v5b6alpkcbvmf"))))
6e76dda2
RW		 7106 (properties
7107 `((upstream-name . "SummarizedExperiment")))
7108 (build-system r-build-system)
7109 (propagated-inputs
7110 `(("r-biobase" ,r-biobase)
7111 ("r-biocgenerics" ,r-biocgenerics)
d006ee31 7112 ("r-delayedarray" ,r-delayedarray)
6e76dda2
RW		 7113 ("r-genomeinfodb" ,r-genomeinfodb)
7114 ("r-genomicranges" ,r-genomicranges)
7115 ("r-iranges" ,r-iranges)
aeb64f3c 7116 ("r-matrix" ,r-matrix)
6e76dda2 7117 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 7118 (home-page "https://bioconductor.org/packages/SummarizedExperiment")
6e76dda2
RW		 7119 (synopsis "Container for representing genomic ranges by sample")
7120 (description
7121 "The SummarizedExperiment container contains one or more assays, each
7122represented by a matrix-like object of numeric or other mode. The rows
7123typically represent genomic ranges of interest and the columns represent
7124samples.")
7125 (license license:artistic2.0)))
7126
d8a828af
RW		 7127(define-public r-genomicalignments
7128 (package
7129 (name "r-genomicalignments")
96fb4651 7130 (version "1.14.0")
d8a828af
RW		 7131 (source (origin
7132 (method url-fetch)
7133 (uri (bioconductor-uri "GenomicAlignments" version))
7134 (sha256
7135 (base32
96fb4651 7136 "0sw30lj11wv7ifzypqm04lcah987crqwvj48wz3flaw3biw41zfi"))))
d8a828af
RW		 7137 (properties
7138 `((upstream-name . "GenomicAlignments")))
7139 (build-system r-build-system)
7140 (propagated-inputs
7141 `(("r-biocgenerics" ,r-biocgenerics)
7142 ("r-biocparallel" ,r-biocparallel)
7143 ("r-biostrings" ,r-biostrings)
7144 ("r-genomeinfodb" ,r-genomeinfodb)
7145 ("r-genomicranges" ,r-genomicranges)
7146 ("r-iranges" ,r-iranges)
7147 ("r-rsamtools" ,r-rsamtools)
7148 ("r-s4vectors" ,r-s4vectors)
7149 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 7150 (home-page "https://bioconductor.org/packages/GenomicAlignments")
d8a828af
RW		 7151 (synopsis "Representation and manipulation of short genomic alignments")
7152 (description
7153 "This package provides efficient containers for storing and manipulating
7154short genomic alignments (typically obtained by aligning short reads to a
7155reference genome). This includes read counting, computing the coverage,
7156junction detection, and working with the nucleotide content of the
7157alignments.")
7158 (license license:artistic2.0)))
7159
317755ff
RW		 7160(define-public r-rtracklayer
7161 (package
7162 (name "r-rtracklayer")
4dd469ef 7163 (version "1.38.0")
317755ff
RW		 7164 (source (origin
7165 (method url-fetch)
7166 (uri (bioconductor-uri "rtracklayer" version))
7167 (sha256
7168 (base32
4dd469ef 7169 "12al1ygzy9p4myxa1fd817m28x2fj6f863znk9bw3hp7knbi98dh"))))
317755ff
RW		 7170 (build-system r-build-system)
7171 (arguments
7172 `(#:phases
7173 (modify-phases %standard-phases
7174 (add-after 'unpack 'use-system-zlib
7175 (lambda _
7176 (substitute* "DESCRIPTION"
4dd469ef 7177 ((" zlibbioc,") ""))
317755ff
RW		 7178 (substitute* "NAMESPACE"
7179 (("import\\(zlibbioc\\)") ""))
7180 #t)))))
7181 (inputs
7182 `(("zlib" ,zlib)))
7183 (propagated-inputs
7184 `(("r-biocgenerics" ,r-biocgenerics)
7185 ("r-biostrings" ,r-biostrings)
7186 ("r-genomeinfodb" ,r-genomeinfodb)
7187 ("r-genomicalignments" ,r-genomicalignments)
7188 ("r-genomicranges" ,r-genomicranges)
7189 ("r-iranges" ,r-iranges)
7190 ("r-rcurl" ,r-rcurl)
7191 ("r-rsamtools" ,r-rsamtools)
7192 ("r-s4vectors" ,r-s4vectors)
7193 ("r-xml" ,r-xml)
7194 ("r-xvector" ,r-xvector)))
5713bbf1 7195 (home-page "https://bioconductor.org/packages/rtracklayer")
317755ff
RW		 7196 (synopsis "R interface to genome browsers and their annotation tracks")
7197 (description
7198 "rtracklayer is an extensible framework for interacting with multiple
7199genome browsers (currently UCSC built-in) and manipulating annotation tracks
7200in various formats (currently GFF, BED, bedGraph, BED15, WIG, BigWig and 2bit
7201built-in). The user may export/import tracks to/from the supported browsers,
7202as well as query and modify the browser state, such as the current viewport.")
7203 (license license:artistic2.0)))
7204
2fd7c049
RW		 7205(define-public r-genomicfeatures
7206 (package
7207 (name "r-genomicfeatures")
158b6743 7208 (version "1.30.0")
2fd7c049
RW		 7209 (source (origin
7210 (method url-fetch)
7211 (uri (bioconductor-uri "GenomicFeatures" version))
7212 (sha256
7213 (base32
158b6743 7214 "1khjvq1ffhqavkwf8n7bilknci60lxbg52icrcf2vnb9k8rlpghs"))))
2fd7c049
RW		 7215 (properties
7216 `((upstream-name . "GenomicFeatures")))
7217 (build-system r-build-system)
7218 (propagated-inputs
7219 `(("r-annotationdbi" ,r-annotationdbi)
7220 ("r-biobase" ,r-biobase)
7221 ("r-biocgenerics" ,r-biocgenerics)
7222 ("r-biomart" ,r-biomart)
7223 ("r-biostrings" ,r-biostrings)
7224 ("r-dbi" ,r-dbi)
7225 ("r-genomeinfodb" ,r-genomeinfodb)
7226 ("r-genomicranges" ,r-genomicranges)
7227 ("r-iranges" ,r-iranges)
7228 ("r-rcurl" ,r-rcurl)
7229 ("r-rsqlite" ,r-rsqlite)
158b6743 7230 ("r-rmysql" ,r-rmysql)
2fd7c049
RW		 7231 ("r-rtracklayer" ,r-rtracklayer)
7232 ("r-s4vectors" ,r-s4vectors)
7233 ("r-xvector" ,r-xvector)))
5713bbf1 7234 (home-page "https://bioconductor.org/packages/GenomicFeatures")
2fd7c049
RW		 7235 (synopsis "Tools for working with transcript centric annotations")
7236 (description
7237 "This package provides a set of tools and methods for making and
7238manipulating transcript centric annotations. With these tools the user can
7239easily download the genomic locations of the transcripts, exons and cds of a
7240given organism, from either the UCSC Genome Browser or a BioMart
7241database (more sources will be supported in the future). This information is
7242then stored in a local database that keeps track of the relationship between
7243transcripts, exons, cds and genes. Flexible methods are provided for
7244extracting the desired features in a convenient format.")
7245 (license license:artistic2.0)))
7246
fb25d880
RW		 7247(define-public r-go-db
7248 (package
7249 (name "r-go-db")
592f4a94 7250 (version "3.5.0")
fb25d880
RW		 7251 (source (origin
7252 (method url-fetch)
5713bbf1 7253 (uri (string-append "https://www.bioconductor.org/packages/"
f82c8c3c
PP		 7254 "release/data/annotation/src/contrib/GO.db_"
7255 version ".tar.gz"))
fb25d880
RW		 7256 (sha256
7257 (base32
592f4a94 7258 "02d1mn1al3q7qvhx1ylrr3ar4w4iw0qyi5d89v2336rzwk9maq35"))))
fb25d880
RW		 7259 (properties
7260 `((upstream-name . "GO.db")))
7261 (build-system r-build-system)
3141b83d
RW		 7262 (propagated-inputs
7263 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7264 (home-page "https://bioconductor.org/packages/GO.db")
fb25d880
RW		 7265 (synopsis "Annotation maps describing the entire Gene Ontology")
7266 (description
7267 "The purpose of this GO.db annotation package is to provide detailed
7268information about the latest version of the Gene Ontologies.")
7269 (license license:artistic2.0)))
7270
d1dbde6a
RW		 7271(define-public r-graph
7272 (package
7273 (name "r-graph")
aeb73879 7274 (version "1.56.0")
d1dbde6a
RW		 7275 (source (origin
7276 (method url-fetch)
7277 (uri (bioconductor-uri "graph" version))
7278 (sha256
7279 (base32
aeb73879 7280 "15aajjp8h2z14p80c8hyd4rrmr9vqsm7bvwb989jxjl4k6g52an1"))))
d1dbde6a
RW		 7281 (build-system r-build-system)
7282 (propagated-inputs
7283 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 7284 (home-page "https://bioconductor.org/packages/graph")
d1dbde6a
RW		 7285 (synopsis "Handle graph data structures in R")
7286 (description
7287 "This package implements some simple graph handling capabilities for R.")
7288 (license license:artistic2.0)))
7289
d547ce5e
RW		 7290(define-public r-topgo
7291 (package
7292 (name "r-topgo")
d13a3fea 7293 (version "2.30.0")
d547ce5e
RW		 7294 (source (origin
7295 (method url-fetch)
7296 (uri (bioconductor-uri "topGO" version))
7297 (sha256
7298 (base32
d13a3fea 7299 "1hqffz5qp7glxdvjp37005g8qk5nam3f9wpf6d1wjnzpar04f3dz"))))
d547ce5e
RW		 7300 (properties
7301 `((upstream-name . "topGO")))
7302 (build-system r-build-system)
7303 (propagated-inputs
7304 `(("r-annotationdbi" ,r-annotationdbi)
30ec4de7 7305 ("r-dbi" ,r-dbi)
d547ce5e
RW		 7306 ("r-biobase" ,r-biobase)
7307 ("r-biocgenerics" ,r-biocgenerics)
7308 ("r-go-db" ,r-go-db)
6d415db2 7309 ("r-graph" ,r-graph)
aeb64f3c
RW		 7310 ("r-lattice" ,r-lattice)
7311 ("r-matrixstats" ,r-matrixstats)
d547ce5e 7312 ("r-sparsem" ,r-sparsem)))
5713bbf1 7313 (home-page "https://bioconductor.org/packages/topGO")
d547ce5e
RW		 7314 (synopsis "Enrichment analysis for gene ontology")
7315 (description
7316 "The topGO package provides tools for testing @dfn{gene ontology} (GO)
7317terms while accounting for the topology of the GO graph. Different test
7318statistics and different methods for eliminating local similarities and
7319dependencies between GO terms can be implemented and applied.")
7320 ;; Any version of the LGPL applies.
7321 (license license:lgpl2.1+)))
7322
c63cef66
RW		 7323(define-public r-bsgenome
7324 (package
7325 (name "r-bsgenome")
e67850b4 7326 (version "1.46.0")
c63cef66
RW		 7327 (source (origin
7328 (method url-fetch)
7329 (uri (bioconductor-uri "BSgenome" version))
7330 (sha256
7331 (base32
e67850b4 7332 "1jbzq7lm2iajajn2bifxnkss0k9fdvgqr30mral17cbhp5f6w4lq"))))
c63cef66
RW		 7333 (properties
7334 `((upstream-name . "BSgenome")))
7335 (build-system r-build-system)
7336 (propagated-inputs
7337 `(("r-biocgenerics" ,r-biocgenerics)
7338 ("r-biostrings" ,r-biostrings)
7339 ("r-genomeinfodb" ,r-genomeinfodb)
7340 ("r-genomicranges" ,r-genomicranges)
7341 ("r-iranges" ,r-iranges)
7342 ("r-rsamtools" ,r-rsamtools)
7343 ("r-rtracklayer" ,r-rtracklayer)
7344 ("r-s4vectors" ,r-s4vectors)
7345 ("r-xvector" ,r-xvector)))
5713bbf1 7346 (home-page "https://bioconductor.org/packages/BSgenome")
c63cef66
RW		 7347 (synopsis "Infrastructure for Biostrings-based genome data packages")
7348 (description
7349 "This package provides infrastructure shared by all Biostrings-based
7350genome data packages and support for efficient SNP representation.")
7351 (license license:artistic2.0)))
7352
aa3eeeb5
RJ		 7353(define-public r-bsgenome-hsapiens-1000genomes-hs37d5
7354 (package
7355 (name "r-bsgenome-hsapiens-1000genomes-hs37d5")
7356 (version "0.99.1")
7357 (source (origin
7358 (method url-fetch)
7359 ;; We cannot use bioconductor-uri here because this tarball is
7360 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7361 (uri (string-append "https://www.bioconductor.org/packages/"
aa3eeeb5
RJ		 7362 "release/data/annotation/src/contrib/"
7363 "BSgenome.Hsapiens.1000genomes.hs37d5_"
7364 version ".tar.gz"))
7365 (sha256
7366 (base32
7367 "1cg0g5fqmsvwyw2p9hp2yy4ilk21jkbbrnpgqvb5c36ihjwvc7sr"))))
7368 (properties
7369 `((upstream-name . "BSgenome.Hsapiens.1000genomes.hs37d5")))
7370 (build-system r-build-system)
7371 ;; As this package provides little more than a very large data file it
7372 ;; doesn't make sense to build substitutes.
7373 (arguments `(#:substitutable? #f))
7374 (propagated-inputs
7375 `(("r-bsgenome" ,r-bsgenome)))
7376 (home-page
5713bbf1 7377 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.1000genomes.hs37d5/")
aa3eeeb5
RJ		 7378 (synopsis "Full genome sequences for Homo sapiens")
7379 (description
7380 "This package provides full genome sequences for Homo sapiens from
73811000genomes phase2 reference genome sequence (hs37d5), based on NCBI GRCh37.")
7382 (license license:artistic2.0)))
7383
c43a011d
RW		 7384(define-public r-impute
7385 (package
7386 (name "r-impute")
e6ce4bf0 7387 (version "1.52.0")
c43a011d
RW		 7388 (source (origin
7389 (method url-fetch)
7390 (uri (bioconductor-uri "impute" version))
7391 (sha256
7392 (base32
e6ce4bf0 7393 "0b8r4swvyx3cjcc2ky8yn0ncpzlbi1pgfsn3wpbjmhh7sqrffm2n"))))
c43a011d
RW		 7394 (inputs
7395 `(("gfortran" ,gfortran)))
7396 (build-system r-build-system)
5713bbf1 7397 (home-page "https://bioconductor.org/packages/impute")
c43a011d
RW		 7398 (synopsis "Imputation for microarray data")
7399 (description
7400 "This package provides a function to impute missing gene expression
7401microarray data, using nearest neighbor averaging.")
7402 (license license:gpl2+)))
7403
03ea5a35
RW		 7404(define-public r-seqpattern
7405 (package
7406 (name "r-seqpattern")
0f948b11 7407 (version "1.10.0")
03ea5a35
RW		 7408 (source (origin
7409 (method url-fetch)
7410 (uri (bioconductor-uri "seqPattern" version))
7411 (sha256
7412 (base32
0f948b11 7413 "1kcm5w83q7w0v0vs7nyp4gq5z86c6n6pqy9zmyyhxcrns7f597pm"))))
03ea5a35
RW		 7414 (properties
7415 `((upstream-name . "seqPattern")))
7416 (build-system r-build-system)
7417 (propagated-inputs
7418 `(("r-biostrings" ,r-biostrings)
7419 ("r-genomicranges" ,r-genomicranges)
7420 ("r-iranges" ,r-iranges)
e92dd6f5 7421 ("r-kernsmooth" ,r-kernsmooth)
03ea5a35 7422 ("r-plotrix" ,r-plotrix)))
5713bbf1 7423 (home-page "https://bioconductor.org/packages/seqPattern")
03ea5a35
RW		 7424 (synopsis "Visualising oligonucleotide patterns and motif occurrences")
7425 (description
7426 "This package provides tools to visualize oligonucleotide patterns and
7427sequence motif occurrences across a large set of sequences centred at a common
7428reference point and sorted by a user defined feature.")
7429 (license license:gpl3+)))
7430
cb933df6
RW		 7431(define-public r-genomation
7432 (package
7433 (name "r-genomation")
e127c918 7434 (version "1.10.0")
cb933df6
RW		 7435 (source (origin
7436 (method url-fetch)
7437 (uri (bioconductor-uri "genomation" version))
7438 (sha256
7439 (base32
e127c918 7440 "1ddd8c9w1f1i1ga9rpbwiic8rsaws1chdxx4j38bpyaiy4zhz1ca"))))
cb933df6
RW		 7441 (build-system r-build-system)
7442 (propagated-inputs
7443 `(("r-biostrings" ,r-biostrings)
7444 ("r-bsgenome" ,r-bsgenome)
7445 ("r-data-table" ,r-data-table)
7446 ("r-genomeinfodb" ,r-genomeinfodb)
7447 ("r-genomicalignments" ,r-genomicalignments)
7448 ("r-genomicranges" ,r-genomicranges)
7449 ("r-ggplot2" ,r-ggplot2)
7450 ("r-gridbase" ,r-gridbase)
7451 ("r-impute" ,r-impute)
7452 ("r-iranges" ,r-iranges)
7453 ("r-matrixstats" ,r-matrixstats)
7454 ("r-plotrix" ,r-plotrix)
7455 ("r-plyr" ,r-plyr)
51c3c490 7456 ("r-rcpp" ,r-rcpp)
cb933df6
RW		 7457 ("r-readr" ,r-readr)
7458 ("r-reshape2" ,r-reshape2)
7459 ("r-rsamtools" ,r-rsamtools)
7460 ("r-rtracklayer" ,r-rtracklayer)
51c3c490
RW		 7461 ("r-runit" ,r-runit)
7462 ("r-s4vectors" ,r-s4vectors)
cb933df6
RW		 7463 ("r-seqpattern" ,r-seqpattern)))
7464 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7465 (synopsis "Summary, annotation and visualization of genomic data")
7466 (description
7467 "This package provides a package for summary and annotation of genomic
7468intervals. Users can visualize and quantify genomic intervals over
7469pre-defined functional regions, such as promoters, exons, introns, etc. The
7470genomic intervals represent regions with a defined chromosome position, which
7471may be associated with a score, such as aligned reads from HT-seq experiments,
7472TF binding sites, methylation scores, etc. The package can use any tabular
7473genomic feature data as long as it has minimal information on the locations of
7474genomic intervals. In addition, it can use BAM or BigWig files as input.")
7475 (license license:artistic2.0)))
7476
64efa307
RW		 7477(define-public r-genomationdata
7478 (package
7479 (name "r-genomationdata")
57dc9b58 7480 (version "1.10.0")
64efa307
RW		 7481 (source (origin
7482 (method url-fetch)
7483 ;; We cannot use bioconductor-uri here because this tarball is
7484 ;; located under "data/annotation/" instead of "bioc/".
7485 (uri (string-append "https://bioconductor.org/packages/"
7486 "release/data/experiment/src/contrib/"
7487 "genomationData_" version ".tar.gz"))
7488 (sha256
7489 (base32
57dc9b58 7490 "0h7g5x3kyb50qlblz5hc85lfm6n6f5nb68i146way3ggs04sqvla"))))
64efa307
RW		 7491 (build-system r-build-system)
7492 ;; As this package provides little more than large data files, it doesn't
7493 ;; make sense to build substitutes.
7494 (arguments `(#:substitutable? #f))
7495 (native-inputs
7496 `(("r-knitr" ,r-knitr)))
7497 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7498 (synopsis "Experimental data for use with the genomation package")
7499 (description
7500 "This package contains experimental genetic data for use with the
7501genomation package. Included are Chip Seq, Methylation and Cage data,
7502downloaded from Encode.")
7503 (license license:gpl3+)))
7504
486da491
RW		 7505(define-public r-org-hs-eg-db
7506 (package
7507 (name "r-org-hs-eg-db")
d595fed3 7508 (version "3.5.0")
486da491
RW		 7509 (source (origin
7510 (method url-fetch)
7511 ;; We cannot use bioconductor-uri here because this tarball is
7512 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7513 (uri (string-append "https://www.bioconductor.org/packages/"
486da491
RW		 7514 "release/data/annotation/src/contrib/"
7515 "org.Hs.eg.db_" version ".tar.gz"))
7516 (sha256
7517 (base32
d595fed3 7518 "1v6wa5613cjq59xd7x1qz8lr9nb2abm9abl2cci1khrnrlpla927"))))
486da491
RW		 7519 (properties
7520 `((upstream-name . "org.Hs.eg.db")))
7521 (build-system r-build-system)
7522 (propagated-inputs
7523 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7524 (home-page "https://www.bioconductor.org/packages/org.Hs.eg.db/")
486da491
RW		 7525 (synopsis "Genome wide annotation for Human")
7526 (description
676507e3
RW		 7527 "This package contains genome-wide annotations for Human, primarily based
7528on mapping using Entrez Gene identifiers.")
486da491
RW		 7529 (license license:artistic2.0)))
7530
fefedf98
RW		 7531(define-public r-org-ce-eg-db
7532 (package
7533 (name "r-org-ce-eg-db")
e4664290 7534 (version "3.5.0")
fefedf98
RW		 7535 (source (origin
7536 (method url-fetch)
7537 ;; We cannot use bioconductor-uri here because this tarball is
7538 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7539 (uri (string-append "https://www.bioconductor.org/packages/"
fefedf98
RW		 7540 "release/data/annotation/src/contrib/"
7541 "org.Ce.eg.db_" version ".tar.gz"))
7542 (sha256
7543 (base32
e4664290 7544 "02ggchixlmzywhsbr0h2ms4dravv7m5964cjxqcjxqs16vjwlbk9"))))
fefedf98
RW		 7545 (properties
7546 `((upstream-name . "org.Ce.eg.db")))
7547 (build-system r-build-system)
7548 (propagated-inputs
7549 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7550 (home-page "https://www.bioconductor.org/packages/org.Ce.eg.db/")
fefedf98
RW		 7551 (synopsis "Genome wide annotation for Worm")
7552 (description
7553 "This package provides mappings from Entrez gene identifiers to various
7554annotations for the genome of the model worm Caenorhabditis elegans.")
7555 (license license:artistic2.0)))
7556
16c53a1e
RW		 7557(define-public r-org-dm-eg-db
7558 (package
7559 (name "r-org-dm-eg-db")
19fc299f 7560 (version "3.5.0")
16c53a1e
RW		 7561 (source (origin
7562 (method url-fetch)
7563 ;; We cannot use bioconductor-uri here because this tarball is
7564 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7565 (uri (string-append "https://www.bioconductor.org/packages/"
16c53a1e
RW		 7566 "release/data/annotation/src/contrib/"
7567 "org.Dm.eg.db_" version ".tar.gz"))
7568 (sha256
7569 (base32
19fc299f 7570 "033qak1d3wwz17va0bh8z8p8arx0aw2va6gm1qfwsvdkj9cd9d7d"))))
16c53a1e
RW		 7571 (properties
7572 `((upstream-name . "org.Dm.eg.db")))
7573 (build-system r-build-system)
7574 (propagated-inputs
7575 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7576 (home-page "https://www.bioconductor.org/packages/org.Dm.eg.db/")
16c53a1e
RW		 7577 (synopsis "Genome wide annotation for Fly")
7578 (description
7579 "This package provides mappings from Entrez gene identifiers to various
7580annotations for the genome of the model fruit fly Drosophila melanogaster.")
7581 (license license:artistic2.0)))
7582
e761beb9
RW		 7583(define-public r-org-mm-eg-db
7584 (package
7585 (name "r-org-mm-eg-db")
f3569f52 7586 (version "3.5.0")
e761beb9
RW		 7587 (source (origin
7588 (method url-fetch)
7589 ;; We cannot use bioconductor-uri here because this tarball is
7590 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7591 (uri (string-append "https://www.bioconductor.org/packages/"
e761beb9
RW		 7592 "release/data/annotation/src/contrib/"
7593 "org.Mm.eg.db_" version ".tar.gz"))
7594 (sha256
7595 (base32
f3569f52 7596 "11q21p3ki4bn4hb3aix0g775l45l66jmas6m94nfhqqnpjhv4d6g"))))
e761beb9
RW		 7597 (properties
7598 `((upstream-name . "org.Mm.eg.db")))
7599 (build-system r-build-system)
7600 (propagated-inputs
7601 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7602 (home-page "https://www.bioconductor.org/packages/org.Mm.eg.db/")
e761beb9
RW		 7603 (synopsis "Genome wide annotation for Mouse")
7604 (description
7605 "This package provides mappings from Entrez gene identifiers to various
7606annotations for the genome of the model mouse Mus musculus.")
7607 (license license:artistic2.0)))
7608
936e7d67
RW		 7609(define-public r-seqlogo
7610 (package
7611 (name "r-seqlogo")
61770089 7612 (version "1.44.0")
936e7d67
RW		 7613 (source
7614 (origin
7615 (method url-fetch)
7616 (uri (bioconductor-uri "seqLogo" version))
7617 (sha256
7618 (base32
61770089 7619 "1ql4q4vx0j61a893dqc3c8zxmgs8sqhy3j1qhyfdvbd01vw9w1kq"))))
936e7d67
RW		 7620 (properties `((upstream-name . "seqLogo")))
7621 (build-system r-build-system)
5713bbf1 7622 (home-page "https://bioconductor.org/packages/seqLogo")
936e7d67
RW		 7623 (synopsis "Sequence logos for DNA sequence alignments")
7624 (description
7625 "seqLogo takes the position weight matrix of a DNA sequence motif and
7626plots the corresponding sequence logo as introduced by Schneider and
7627Stephens (1990).")
7628 (license license:lgpl2.0+)))
7629
c90a4baf
RW		 7630(define-public r-bsgenome-hsapiens-ucsc-hg19
7631 (package
7632 (name "r-bsgenome-hsapiens-ucsc-hg19")
7633 (version "1.4.0")
7634 (source (origin
7635 (method url-fetch)
7636 ;; We cannot use bioconductor-uri here because this tarball is
7637 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7638 (uri (string-append "https://www.bioconductor.org/packages/"
c90a4baf
RW		 7639 "release/data/annotation/src/contrib/"
7640 "BSgenome.Hsapiens.UCSC.hg19_"
7641 version ".tar.gz"))
7642 (sha256
7643 (base32
9d217d27 7644 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
c90a4baf
RW		 7645 (properties
7646 `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
7647 (build-system r-build-system)
7648 ;; As this package provides little more than a very large data file it
7649 ;; doesn't make sense to build substitutes.
7650 (arguments `(#:substitutable? #f))
7651 (propagated-inputs
7652 `(("r-bsgenome" ,r-bsgenome)))
7653 (home-page
5713bbf1 7654 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
c90a4baf
RW		 7655 (synopsis "Full genome sequences for Homo sapiens")
7656 (description
7657 "This package provides full genome sequences for Homo sapiens as provided
7658by UCSC (hg19, February 2009) and stored in Biostrings objects.")
7659 (license license:artistic2.0)))
7660
a3e90287
RW		 7661(define-public r-bsgenome-mmusculus-ucsc-mm9
7662 (package
7663 (name "r-bsgenome-mmusculus-ucsc-mm9")
7664 (version "1.4.0")
7665 (source (origin
7666 (method url-fetch)
7667 ;; We cannot use bioconductor-uri here because this tarball is
7668 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7669 (uri (string-append "https://www.bioconductor.org/packages/"
a3e90287
RW		 7670 "release/data/annotation/src/contrib/"
7671 "BSgenome.Mmusculus.UCSC.mm9_"
7672 version ".tar.gz"))
7673 (sha256
7674 (base32
7675 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
7676 (properties
7677 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
7678 (build-system r-build-system)
7679 ;; As this package provides little more than a very large data file it
7680 ;; doesn't make sense to build substitutes.
7681 (arguments `(#:substitutable? #f))
7682 (propagated-inputs
7683 `(("r-bsgenome" ,r-bsgenome)))
7684 (home-page
5713bbf1 7685 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
a3e90287
RW		 7686 (synopsis "Full genome sequences for Mouse")
7687 (description
7688 "This package provides full genome sequences for Mus musculus (Mouse) as
7689provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
7690 (license license:artistic2.0)))
7691
4714d521
RW		 7692(define-public r-bsgenome-mmusculus-ucsc-mm10
7693 (package
7694 (name "r-bsgenome-mmusculus-ucsc-mm10")
7695 (version "1.4.0")
7696 (source (origin
7697 (method url-fetch)
7698 ;; We cannot use bioconductor-uri here because this tarball is
7699 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7700 (uri (string-append "https://www.bioconductor.org/packages/"
4714d521
RW		 7701 "release/data/annotation/src/contrib/"
7702 "BSgenome.Mmusculus.UCSC.mm10_"
7703 version ".tar.gz"))
7704 (sha256
7705 (base32
7706 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
7707 (properties
7708 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
7709 (build-system r-build-system)
7710 ;; As this package provides little more than a very large data file it
7711 ;; doesn't make sense to build substitutes.
7712 (arguments `(#:substitutable? #f))
7713 (propagated-inputs
7714 `(("r-bsgenome" ,r-bsgenome)))
7715 (home-page
5713bbf1 7716 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
4714d521
RW		 7717 (synopsis "Full genome sequences for Mouse")
7718 (description
7719 "This package provides full genome sequences for Mus
7720musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
7721in Biostrings objects.")
7722 (license license:artistic2.0)))
7723
c5173d74
RJ		 7724(define-public r-txdb-mmusculus-ucsc-mm10-knowngene
7725 (package
7726 (name "r-txdb-mmusculus-ucsc-mm10-knowngene")
7727 (version "3.4.0")
7728 (source (origin
7729 (method url-fetch)
7730 ;; We cannot use bioconductor-uri here because this tarball is
7731 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7732 (uri (string-append "https://www.bioconductor.org/packages/"
c5173d74
RJ		 7733 "release/data/annotation/src/contrib/"
7734 "TxDb.Mmusculus.UCSC.mm10.knownGene_"
7735 version ".tar.gz"))
7736 (sha256
7737 (base32
7738 "08gava9wsvpcqz51k2sni3pj03n5155v32d9riqbf305nbirqbkb"))))
7739 (properties
7740 `((upstream-name . "TxDb.Mmusculus.UCSC.mm10.knownGene")))
7741 (build-system r-build-system)
7742 ;; As this package provides little more than a very large data file it
7743 ;; doesn't make sense to build substitutes.
7744 (arguments `(#:substitutable? #f))
7745 (propagated-inputs
7746 `(("r-bsgenome" ,r-bsgenome)
7747 ("r-genomicfeatures" ,r-genomicfeatures)
7748 ("r-annotationdbi" ,r-annotationdbi)))
7749 (home-page
5713bbf1 7750 "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm10.knownGene/")
c5173d74
RJ		 7751 (synopsis "Annotation package for TxDb knownGene object(s) for Mouse")
7752 (description
7753 "This package loads a TxDb object, which is an R interface to
7754prefabricated databases contained in this package. This package provides
7755the TxDb object of Mouse data as provided by UCSC (mm10, December 2011)
7756based on the knownGene track.")
7757 (license license:artistic2.0)))
7758
943bd627
RW		 7759(define-public r-bsgenome-celegans-ucsc-ce6
7760 (package
7761 (name "r-bsgenome-celegans-ucsc-ce6")
7762 (version "1.4.0")
7763 (source (origin
7764 (method url-fetch)
7765 ;; We cannot use bioconductor-uri here because this tarball is
7766 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7767 (uri (string-append "https://www.bioconductor.org/packages/"
943bd627
RW		 7768 "release/data/annotation/src/contrib/"
7769 "BSgenome.Celegans.UCSC.ce6_"
7770 version ".tar.gz"))
7771 (sha256
7772 (base32
7773 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
7774 (properties
7775 `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
7776 (build-system r-build-system)
7777 ;; As this package provides little more than a very large data file it
7778 ;; doesn't make sense to build substitutes.
7779 (arguments `(#:substitutable? #f))
7780 (propagated-inputs
7781 `(("r-bsgenome" ,r-bsgenome)))
7782 (home-page
5713bbf1 7783 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
943bd627
RW		 7784 (synopsis "Full genome sequences for Worm")
7785 (description
7786 "This package provides full genome sequences for Caenorhabditis
7787elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
7788objects.")
7789 (license license:artistic2.0)))
7790
fc47c7d6
RW		 7791(define-public r-bsgenome-celegans-ucsc-ce10
7792 (package
7793 (name "r-bsgenome-celegans-ucsc-ce10")
7794 (version "1.4.0")
7795 (source (origin
7796 (method url-fetch)
7797 ;; We cannot use bioconductor-uri here because this tarball is
7798 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7799 (uri (string-append "https://www.bioconductor.org/packages/"
fc47c7d6
RW		 7800 "release/data/annotation/src/contrib/"
7801 "BSgenome.Celegans.UCSC.ce10_"
7802 version ".tar.gz"))
7803 (sha256
7804 (base32
7805 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
7806 (properties
7807 `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
7808 (build-system r-build-system)
7809 ;; As this package provides little more than a very large data file it
7810 ;; doesn't make sense to build substitutes.
7811 (arguments `(#:substitutable? #f))
7812 (propagated-inputs
7813 `(("r-bsgenome" ,r-bsgenome)))
7814 (home-page
5713bbf1 7815 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
fc47c7d6
RW		 7816 (synopsis "Full genome sequences for Worm")
7817 (description
7818 "This package provides full genome sequences for Caenorhabditis
7819elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
7820objects.")
7821 (license license:artistic2.0)))
7822
6dc60998
RW		 7823(define-public r-bsgenome-dmelanogaster-ucsc-dm3
7824 (package
7825 (name "r-bsgenome-dmelanogaster-ucsc-dm3")
7826 (version "1.4.0")
7827 (source (origin
7828 (method url-fetch)
7829 ;; We cannot use bioconductor-uri here because this tarball is
7830 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7831 (uri (string-append "https://www.bioconductor.org/packages/"
6dc60998
RW		 7832 "release/data/annotation/src/contrib/"
7833 "BSgenome.Dmelanogaster.UCSC.dm3_"
7834 version ".tar.gz"))
7835 (sha256
7836 (base32
7837 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
7838 (properties
7839 `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
7840 (build-system r-build-system)
7841 ;; As this package provides little more than a very large data file it
7842 ;; doesn't make sense to build substitutes.
7843 (arguments `(#:substitutable? #f))
7844 (propagated-inputs
7845 `(("r-bsgenome" ,r-bsgenome)))
7846 (home-page
5713bbf1 7847 "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
6dc60998
RW		 7848 (synopsis "Full genome sequences for Fly")
7849 (description
7850 "This package provides full genome sequences for Drosophila
7851melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
7852Biostrings objects.")
7853 (license license:artistic2.0)))
7854
ae2462f7
RW		 7855(define-public r-motifrg
7856 (package
7857 (name "r-motifrg")
ce59d625 7858 (version "1.22.0")
ae2462f7
RW		 7859 (source
7860 (origin
7861 (method url-fetch)
7862 (uri (bioconductor-uri "motifRG" version))
7863 (sha256
7864 (base32
ce59d625 7865 "193zl2rlzwxv9p9q5i7rilj3w05ndqfyp9bdpvagp5s5cin4hf44"))))
ae2462f7
RW		 7866 (properties `((upstream-name . "motifRG")))
7867 (build-system r-build-system)
7868 (propagated-inputs
7869 `(("r-biostrings" ,r-biostrings)
7870 ("r-bsgenome" ,r-bsgenome)
007424b5 7871 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
ae2462f7
RW		 7872 ("r-iranges" ,r-iranges)
7873 ("r-seqlogo" ,r-seqlogo)
7874 ("r-xvector" ,r-xvector)))
5713bbf1 7875 (home-page "https://bioconductor.org/packages/motifRG")
ae2462f7
RW		 7876 (synopsis "Discover motifs in high throughput sequencing data")
7877 (description
7878 "This package provides tools for discriminative motif discovery in high
7879throughput genetic sequencing data sets using regression methods.")
7880 (license license:artistic2.0)))
7881
a5002ae7
AE		 7882(define-public r-qtl
7883 (package
7884 (name "r-qtl")
c8a9b2bc 7885 (version "1.41-6")
a5002ae7
AE		 7886 (source
7887 (origin
7888 (method url-fetch)
7889 (uri (string-append "mirror://cran/src/contrib/qtl_"
7890 version ".tar.gz"))
7891 (sha256
7892 (base32
c8a9b2bc 7893 "067az4v432zxp6lxck8d7vlh9w4r13r0mvw5zsglyaqwsh3d9sad"))))
a5002ae7
AE		 7894 (build-system r-build-system)
7895 (home-page "http://rqtl.org/")
7896 (synopsis "R package for analyzing QTL experiments in genetics")
7897 (description "R/qtl is an extension library for the R statistics
7898system. It is used to analyze experimental crosses for identifying
7899genes contributing to variation in quantitative traits (so-called
7900quantitative trait loci, QTLs).
7901
7902Using a hidden Markov model, R/qtl allows to estimate genetic maps, to
7903identify genotyping errors, and to perform single-QTL and two-QTL,
7904two-dimensional genome scans.")
7905 (license license:gpl3)))
d1e32822 7906
9e3ba31c
RJ		 7907(define-public r-zlibbioc
7908 (package
7909 (name "r-zlibbioc")
5c184700 7910 (version "1.24.0")
9e3ba31c
RJ		 7911 (source (origin
7912 (method url-fetch)
7913 (uri (bioconductor-uri "zlibbioc" version))
7914 (sha256
7915 (base32
5c184700 7916 "1zr9hbh55hglfpy15cpxwmddxblhyb0an15953l3rbhmlh2vpy92"))))
9e3ba31c
RJ		 7917 (properties
7918 `((upstream-name . "zlibbioc")))
7919 (build-system r-build-system)
7920 (home-page "https://bioconductor.org/packages/zlibbioc")
7921 (synopsis "Provider for zlib-1.2.5 to R packages")
7922 (description "This package uses the source code of zlib-1.2.5 to create
7923libraries for systems that do not have these available via other means.")
7924 (license license:artistic2.0)))
7925
e619a5c2
RW		 7926(define-public r-r4rna
7927 (package
7928 (name "r-r4rna")
7929 (version "0.1.4")
7930 (source
7931 (origin
7932 (method url-fetch)
7933 (uri (string-append "http://www.e-rna.org/r-chie/files/R4RNA_"
7934 version ".tar.gz"))
7935 (sha256
7936 (base32
7937 "1p0i78wh76jfgmn9jphbwwaz6yy6pipzfg08xs54cxavxg2j81p5"))))
7938 (build-system r-build-system)
7939 (propagated-inputs
7940 `(("r-optparse" ,r-optparse)
7941 ("r-rcolorbrewer" ,r-rcolorbrewer)))
7942 (home-page "http://www.e-rna.org/r-chie/index.cgi")
7943 (synopsis "Analysis framework for RNA secondary structure")
7944 (description
7945 "The R4RNA package aims to be a general framework for the analysis of RNA
7946secondary structure and comparative analysis in R.")
7947 (license license:gpl3+)))
7948
52765a63
RW		 7949(define-public r-rhtslib
7950 (package
7951 (name "r-rhtslib")
1d0263b4 7952 (version "1.10.0")
52765a63
RW		 7953 (source
7954 (origin
7955 (method url-fetch)
7956 (uri (bioconductor-uri "Rhtslib" version))
7957 (sha256
7958 (base32
1d0263b4 7959 "1dw3p44bfr0m7w39ckc2k37sjcp1zz0b9g12mr8am15jaj6v0q2j"))))
52765a63
RW		 7960 (properties `((upstream-name . "Rhtslib")))
7961 (build-system r-build-system)
7962 (propagated-inputs
7963 `(("r-zlibbioc" ,r-zlibbioc)))
7964 (inputs
7965 `(("zlib" ,zlib)))
53ca52f0
RW		 7966 (native-inputs
7967 `(("autoconf" ,autoconf)))
52765a63
RW		 7968 (home-page "https://github.com/nhayden/Rhtslib")
7969 (synopsis "High-throughput sequencing library as an R package")
7970 (description
7971 "This package provides the HTSlib C library for high-throughput
7972nucleotide sequence analysis. The package is primarily useful to developers
7973of other R packages who wish to make use of HTSlib.")
7974 (license license:lgpl2.0+)))
7975
fe02c4c9
RW		 7976(define-public r-bamsignals
7977 (package
7978 (name "r-bamsignals")
da153b7f 7979 (version "1.10.0")
fe02c4c9
RW		 7980 (source
7981 (origin
7982 (method url-fetch)
7983 (uri (bioconductor-uri "bamsignals" version))
7984 (sha256
7985 (base32
da153b7f 7986 "15id6mkj95skb4kfafvfs2j7ylydal60c3pspcl7llhwpq6vcqvl"))))
fe02c4c9
RW		 7987 (build-system r-build-system)
7988 (propagated-inputs
7989 `(("r-biocgenerics" ,r-biocgenerics)
7990 ("r-genomicranges" ,r-genomicranges)
7991 ("r-iranges" ,r-iranges)
7992 ("r-rcpp" ,r-rcpp)
7993 ("r-rhtslib" ,r-rhtslib)
7994 ("r-zlibbioc" ,r-zlibbioc)))
7995 (inputs
7996 `(("zlib" ,zlib)))
5713bbf1 7997 (home-page "https://bioconductor.org/packages/bamsignals")
fe02c4c9
RW		 7998 (synopsis "Extract read count signals from bam files")
7999 (description
8000 "This package allows to efficiently obtain count vectors from indexed bam
8001files. It counts the number of nucleotide sequence reads in given genomic
8002ranges and it computes reads profiles and coverage profiles. It also handles
8003paired-end data.")
8004 (license license:gpl2+)))
8005
89984be4
RW		 8006(define-public r-rcas
8007 (package
8008 (name "r-rcas")
d82937fb 8009 (version "1.3.4")
89984be4
RW		 8010 (source (origin
8011 (method url-fetch)
8012 (uri (string-append "https://github.com/BIMSBbioinfo/RCAS/archive/v"
8013 version ".tar.gz"))
8014 (file-name (string-append name "-" version ".tar.gz"))
8015 (sha256
8016 (base32
d82937fb 8017 "1qgc7vi6fpzl440yg7jhiycg5q336kd4pxqzx10yx2zcq3bq3msg"))))
89984be4
RW		 8018 (build-system r-build-system)
8019 (native-inputs
8020 `(("r-knitr" ,r-knitr)
8021 ("r-testthat" ,r-testthat)
8022 ;; During vignette building knitr checks that "pandoc-citeproc"
8023 ;; is in the PATH.
8024 ("ghc-pandoc-citeproc" ,ghc-pandoc-citeproc)))
8025 (propagated-inputs
8026 `(("r-data-table" ,r-data-table)
8027 ("r-biomart" ,r-biomart)
8028 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
8029 ("r-org-ce-eg-db" ,r-org-ce-eg-db)
8030 ("r-org-dm-eg-db" ,r-org-dm-eg-db)
8031 ("r-org-mm-eg-db" ,r-org-mm-eg-db)
8032 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8033 ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
8034 ("r-bsgenome-celegans-ucsc-ce10" ,r-bsgenome-celegans-ucsc-ce10)
8035 ("r-bsgenome-dmelanogaster-ucsc-dm3" ,r-bsgenome-dmelanogaster-ucsc-dm3)
8036 ("r-topgo" ,r-topgo)
8037 ("r-dt" ,r-dt)
ebfd6a71 8038 ("r-pbapply" ,r-pbapply)
89984be4 8039 ("r-plotly" ,r-plotly)
2c8d6c0b 8040 ("r-plotrix" ,r-plotrix)
89984be4
RW		 8041 ("r-motifrg" ,r-motifrg)
8042 ("r-genomation" ,r-genomation)
8043 ("r-genomicfeatures" ,r-genomicfeatures)
8044 ("r-rtracklayer" ,r-rtracklayer)
8045 ("r-rmarkdown" ,r-rmarkdown)))
8046 (synopsis "RNA-centric annotation system")
8047 (description
8048 "RCAS aims to be a standalone RNA-centric annotation system that provides
8049intuitive reports and publication-ready graphics. This package provides the R
8050library implementing most of the pipeline's features.")
8051 (home-page "https://github.com/BIMSBbioinfo/RCAS")
75690c9f 8052 (license license:artistic2.0)))
89984be4 8053
50937297
RW		 8054(define-public rcas-web
8055 (package
8056 (name "rcas-web")
01d87d3c 8057 (version "0.0.4")
50937297
RW		 8058 (source
8059 (origin
8060 (method url-fetch)
8061 (uri (string-append "https://github.com/BIMSBbioinfo/rcas-web/"
8062 "releases/download/v" version
8063 "/rcas-web-" version ".tar.gz"))
8064 (sha256
8065 (base32
01d87d3c 8066 "1p16frfys41a8yaa4gkm457nzkqhqs2pc3lkac0ds457w9w5j1gm"))))
50937297
RW		 8067 (build-system gnu-build-system)
8068 (arguments
8069 `(#:phases
8070 (modify-phases %standard-phases
8071 (add-after 'install 'wrap-executable
8072 (lambda* (#:key inputs outputs #:allow-other-keys)
8073 (let* ((out (assoc-ref outputs "out"))
8074 (json (assoc-ref inputs "guile-json"))
8075 (redis (assoc-ref inputs "guile-redis"))
8076 (path (string-append
8077 json "/share/guile/site/2.2:"
8078 redis "/share/guile/site/2.2")))
8079 (wrap-program (string-append out "/bin/rcas-web")
8080 `("GUILE_LOAD_PATH" ":" = (,path))
8081 `("GUILE_LOAD_COMPILED_PATH" ":" = (,path))
8082 `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
8083 #t)))))
8084 (inputs
2d7c4ae3 8085 `(("r-minimal" ,r-minimal)
50937297 8086 ("r-rcas" ,r-rcas)
f6396d86 8087 ("guile-next" ,guile-2.2)
2252f087 8088 ("guile-json" ,guile-json)
50937297
RW		 8089 ("guile-redis" ,guile2.2-redis)))
8090 (native-inputs
8091 `(("pkg-config" ,pkg-config)))
8092 (home-page "https://github.com/BIMSBbioinfo/rcas-web")
8093 (synopsis "Web interface for RNA-centric annotation system (RCAS)")
8094 (description "This package provides a simple web interface for the
8095@dfn{RNA-centric annotation system} (RCAS).")
8096 (license license:agpl3+)))
8097
7500e42b
RJ		 8098(define-public r-mutationalpatterns
8099 (package
8100 (name "r-mutationalpatterns")
39d9098d 8101 (version "1.4.0")
7500e42b
RJ		 8102 (source
8103 (origin
8104 (method url-fetch)
8105 (uri (bioconductor-uri "MutationalPatterns" version))
8106 (sha256
8107 (base32
39d9098d 8108 "0sqbrswg8ylkjb9q3vqcb5ggwixynwj6hyv2n4sk7snyk61z3fq9"))))
7500e42b
RJ		 8109 (build-system r-build-system)
8110 (propagated-inputs
8111 `(("r-biocgenerics" ,r-biocgenerics)
8112 ("r-biostrings" ,r-biostrings)
cf4ac4e4
RJ		 8113 ("r-bsgenome-hsapiens-1000g" ,r-bsgenome-hsapiens-1000genomes-hs37d5)
8114 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7500e42b
RJ		 8115 ("r-genomicranges" ,r-genomicranges)
8116 ("r-genomeinfodb" ,r-genomeinfodb)
8117 ("r-ggplot2" ,r-ggplot2)
8118 ("r-gridextra" ,r-gridextra)
8119 ("r-iranges" ,r-iranges)
8120 ("r-nmf" ,r-nmf)
8121 ("r-plyr" ,r-plyr)
8122 ("r-pracma" ,r-pracma)
8123 ("r-reshape2" ,r-reshape2)
39d9098d
RW		 8124 ("r-cowplot" ,r-cowplot)
8125 ("r-ggdendro" ,r-ggdendro)
8126 ("r-s4vectors" ,r-s4vectors)
7500e42b
RJ		 8127 ("r-summarizedexperiment" ,r-summarizedexperiment)
8128 ("r-variantannotation" ,r-variantannotation)))
5713bbf1 8129 (home-page "https://bioconductor.org/packages/MutationalPatterns/")
7500e42b
RJ		 8130 (synopsis "Extract and visualize mutational patterns in genomic data")
8131 (description "This package provides an extensive toolset for the
8132characterization and visualization of a wide range of mutational patterns
8133in SNV base substitution data.")
8134 (license license:expat)))
8135
d7160529
RW		 8136(define-public r-wgcna
8137 (package
8138 (name "r-wgcna")
1b22ecda 8139 (version "1.61")
d7160529
RW		 8140 (source
8141 (origin
8142 (method url-fetch)
8143 (uri (cran-uri "WGCNA" version))
8144 (sha256
8145 (base32
1b22ecda 8146 "1vrc2k33a196hrrl7k0z534fp96vv0shmigcr65ny1q0v6lq0h6i"))))
d7160529
RW		 8147 (properties `((upstream-name . "WGCNA")))
8148 (build-system r-build-system)
8149 (propagated-inputs
8150 `(("r-annotationdbi" ,r-annotationdbi)
8151 ("r-doparallel" ,r-doparallel)
8152 ("r-dynamictreecut" ,r-dynamictreecut)
8153 ("r-fastcluster" ,r-fastcluster)
8154 ("r-foreach" ,r-foreach)
8155 ("r-go-db" ,r-go-db)
8156 ("r-hmisc" ,r-hmisc)
8157 ("r-impute" ,r-impute)
1b22ecda
RW		 8158 ("r-rcpp" ,r-rcpp)
8159 ("r-robust" ,r-robust)
8160 ("r-survival" ,r-survival)
d7160529
RW		 8161 ("r-matrixstats" ,r-matrixstats)
8162 ("r-preprocesscore" ,r-preprocesscore)))
8163 (home-page
8164 "http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/Rpackages/WGCNA/")
8165 (synopsis "Weighted correlation network analysis")
8166 (description
8167 "This package provides functions necessary to perform Weighted
8168Correlation Network Analysis on high-dimensional data. It includes functions
8169for rudimentary data cleaning, construction and summarization of correlation
8170networks, module identification and functions for relating both variables and
8171modules to sample traits. It also includes a number of utility functions for
8172data manipulation and visualization.")
8173 (license license:gpl2+)))
8174
c827f202
RW		 8175(define-public r-chipkernels
8176 (let ((commit "c9cfcacb626b1221094fb3490ea7bac0fd625372")
8177 (revision "1"))
8178 (package
8179 (name "r-chipkernels")
8180 (version (string-append "1.1-" revision "." (string-take commit 9)))
8181 (source
8182 (origin
8183 (method git-fetch)
8184 (uri (git-reference
8185 (url "https://github.com/ManuSetty/ChIPKernels.git")
8186 (commit commit)))
8187 (file-name (string-append name "-" version))
8188 (sha256
8189 (base32
8190 "14bj5qhjm1hsm9ay561nfbqi9wxsa7y487df2idsaaf6z10nw4v0"))))
8191 (build-system r-build-system)
8192 (propagated-inputs
8193 `(("r-iranges" ,r-iranges)
8194 ("r-xvector" ,r-xvector)
8195 ("r-biostrings" ,r-biostrings)
8196 ("r-bsgenome" ,r-bsgenome)
8197 ("r-gtools" ,r-gtools)
8198 ("r-genomicranges" ,r-genomicranges)
8199 ("r-sfsmisc" ,r-sfsmisc)
8200 ("r-kernlab" ,r-kernlab)
8201 ("r-s4vectors" ,r-s4vectors)
8202 ("r-biocgenerics" ,r-biocgenerics)))
8203 (home-page "https://github.com/ManuSetty/ChIPKernels")
8204 (synopsis "Build string kernels for DNA Sequence analysis")
8205 (description "ChIPKernels is an R package for building different string
8206kernels used for DNA Sequence analysis. A dictionary of the desired kernel
8207must be built and this dictionary can be used for determining kernels for DNA
8208Sequences.")
8209 (license license:gpl2+))))
8210
2d9fb170
RW		 8211(define-public r-seqgl
8212 (package
8213 (name "r-seqgl")
8214 (version "1.1.4")
8215 (source
8216 (origin
8217 (method url-fetch)
8218 (uri (string-append "https://github.com/ManuSetty/SeqGL/"
8219 "archive/" version ".tar.gz"))
8220 (file-name (string-append name "-" version ".tar.gz"))
8221 (sha256
8222 (base32
8223 "0pnk1p3sci5yipyc8xnb6jbmydpl80fld927xgnbcv104hy8h8yh"))))
8224 (build-system r-build-system)
8225 (propagated-inputs
8226 `(("r-biostrings" ,r-biostrings)
8227 ("r-chipkernels" ,r-chipkernels)
8228 ("r-genomicranges" ,r-genomicranges)
8229 ("r-spams" ,r-spams)
8230 ("r-wgcna" ,r-wgcna)
8231 ("r-fastcluster" ,r-fastcluster)))
8232 (home-page "https://github.com/ManuSetty/SeqGL")
8233 (synopsis "Group lasso for Dnase/ChIP-seq data")
8234 (description "SeqGL is a group lasso based algorithm to extract
8235transcription factor sequence signals from ChIP, DNase and ATAC-seq profiles.
8236This package presents a method which uses group lasso to discriminate between
8237bound and non bound genomic regions to accurately identify transcription
8238factors bound at the specific regions.")
8239 (license license:gpl2+)))
8240
bd3be46e
RW		 8241(define-public r-gkmsvm
8242 (package
8243 (name "r-gkmsvm")
8244 (version "0.71.0")
8245 (source
8246 (origin
8247 (method url-fetch)
8248 (uri (cran-uri "gkmSVM" version))
8249 (sha256
8250 (base32
8251 "1zpxgxmf2nd5j5wn00ps6kfxr8wxh7d1swr1rr4spq7sj5z5z0k0"))))
8252 (properties `((upstream-name . "gkmSVM")))
8253 (build-system r-build-system)
8254 (propagated-inputs
8255 `(("r-biocgenerics" ,r-biocgenerics)
8256 ("r-biostrings" ,r-biostrings)
8257 ("r-genomeinfodb" ,r-genomeinfodb)
8258 ("r-genomicranges" ,r-genomicranges)
8259 ("r-iranges" ,r-iranges)
8260 ("r-kernlab" ,r-kernlab)
8261 ("r-rcpp" ,r-rcpp)
8262 ("r-rocr" ,r-rocr)
8263 ("r-rtracklayer" ,r-rtracklayer)
8264 ("r-s4vectors" ,r-s4vectors)
8265 ("r-seqinr" ,r-seqinr)))
8266 (home-page "http://cran.r-project.org/web/packages/gkmSVM")
8267 (synopsis "Gapped-kmer support vector machine")
8268 (description
8269 "This R package provides tools for training gapped-kmer SVM classifiers
8270for DNA and protein sequences. This package supports several sequence
8271kernels, including: gkmSVM, kmer-SVM, mismatch kernel and wildcard kernel.")
8272 (license license:gpl2+)))
8273
d4af25b5
RJPB		 8274(define-public r-tximport
8275 (package
8276 (name "r-tximport")
19e8929c 8277 (version "1.6.0")
d4af25b5
RJPB		 8278 (source (origin
8279 (method url-fetch)
8280 (uri (bioconductor-uri "tximport" version))
8281 (sha256
8282 (base32
19e8929c 8283 "1gyqcm91hxg1kgjqcz2qw1n56yp9pymjzs50rwcpb2893dr8sp2h"))))
d4af25b5 8284 (build-system r-build-system)
5713bbf1 8285 (home-page "https://bioconductor.org/packages/tximport")
d4af25b5
RJPB		 8286 (synopsis "Import and summarize transcript-level estimates for gene-level analysis")
8287 (description
8288 "This package provides tools to import transcript-level abundance,
8289estimated counts and transcript lengths, and to summarize them into matrices
8290for use with downstream gene-level analysis packages. Average transcript
8291length, weighted by sample-specific transcript abundance estimates, is
8292provided as a matrix which can be used as an offset for different expression
8293of gene-level counts.")
8294 (license license:gpl2+)))
8295
69f2b3bd
RJPB		 8296(define-public r-rhdf5
8297 (package
8298 (name "r-rhdf5")
e6b332d4 8299 (version "2.22.0")
69f2b3bd
RJPB		 8300 (source (origin
8301 (method url-fetch)
8302 (uri (bioconductor-uri "rhdf5" version))
8303 (sha256
8304 (base32
e6b332d4 8305 "145858qg1xan6imxcbprzq3yn3mdf532aahdr6cibvdjg47hs4c1"))))
69f2b3bd
RJPB		 8306 (build-system r-build-system)
8307 (arguments
8308 `(#:phases
8309 (modify-phases %standard-phases
8310 (add-after 'unpack 'unpack-smallhdf5
8311 (lambda* (#:key outputs #:allow-other-keys)
8312 (system* "tar" "-xzvf"
8313 "src/hdf5source/hdf5small.tgz" "-C" "src/" )
8314 (substitute* "src/Makevars"
8315 (("^.*cd hdf5source &&.*$") "")
8316 (("^.*gunzip -dc hdf5small.tgz.*$") "")
8317 (("^.*rm -rf hdf5.*$") "")
8318 (("^.*mv hdf5source/hdf5 ..*$") ""))
8319 (substitute* "src/hdf5/configure"
8320 (("/bin/mv") "mv"))
8321 #t)))))
8322 (propagated-inputs
8323 `(("r-zlibbioc" ,r-zlibbioc)))
8324 (inputs
8325 `(("perl" ,perl)
8326 ("zlib" ,zlib)))
5713bbf1 8327 (home-page "https://bioconductor.org/packages/rhdf5")
69f2b3bd
RJPB		 8328 (synopsis "HDF5 interface to R")
8329 (description
8330 "This R/Bioconductor package provides an interface between HDF5 and R.
8331HDF5's main features are the ability to store and access very large and/or
8332complex datasets and a wide variety of metadata on mass storage (disk) through
8333a completely portable file format. The rhdf5 package is thus suited for the
8334exchange of large and/or complex datasets between R and other software
8335package, and for letting R applications work on datasets that are larger than
8336the available RAM.")
8337 (license license:artistic2.0)))
8338
17cddc17
RW		 8339(define-public r-annotationfilter
8340 (package
8341 (name "r-annotationfilter")
598dacf6 8342 (version "1.2.0")
17cddc17
RW		 8343 (source (origin
8344 (method url-fetch)
8345 (uri (bioconductor-uri "AnnotationFilter" version))
8346 (sha256
8347 (base32
598dacf6 8348 "04zf864c1fvdlaay2r5cn30fc1n5i3czh31fs62qlrvs61wjiscs"))))
17cddc17
RW		 8349 (properties
8350 `((upstream-name . "AnnotationFilter")))
8351 (build-system r-build-system)
8352 (propagated-inputs
8353 `(("r-genomicranges" ,r-genomicranges)
8354 ("r-lazyeval" ,r-lazyeval)))
8355 (home-page "https://github.com/Bioconductor/AnnotationFilter")
8356 (synopsis "Facilities for filtering Bioconductor annotation resources")
8357 (description
8358 "This package provides classes and other infrastructure to implement
8359filters for manipulating Bioconductor annotation resources. The filters are
8360used by @code{ensembldb}, @code{Organism.dplyr}, and other packages.")
8361 (license license:artistic2.0)))
8362
66e40e00
RW		 8363(define-public emboss
8364 (package
8365 (name "emboss")
8366 (version "6.5.7")
8367 (source (origin
8368 (method url-fetch)
8369 (uri (string-append "ftp://emboss.open-bio.org/pub/EMBOSS/old/"
8370 (version-major+minor version) ".0/"
8371 "EMBOSS-" version ".tar.gz"))
8372 (sha256
8373 (base32
8374 "0vsmz96gc411yj2iyzdrsmg4l2n1nhgmp7vrgzlxx3xixv9xbf0q"))))
8375 (build-system gnu-build-system)
8376 (arguments
8377 `(#:configure-flags
8378 (list (string-append "--with-hpdf="
8379 (assoc-ref %build-inputs "libharu")))
8380 #:phases
8381 (modify-phases %standard-phases
8382 (add-after 'unpack 'fix-checks
8383 (lambda _
8384 ;; The PNGDRIVER tests check for the presence of libgd, libpng
8385 ;; and zlib, but assume that they are all found at the same
8386 ;; prefix.
8387 (substitute* "configure.in"
8388 (("CHECK_PNGDRIVER")
8389 "LIBS=\"$LIBS -lgd -lpng -lz -lm\"
8390AC_DEFINE([PLD_png], [1], [Define to 1 if PNG support is available])
8391AM_CONDITIONAL(AMPNG, true)"))
8392 #t))
d10092b8 8393 (add-after 'fix-checks 'disable-update-check
66e40e00
RW		 8394 (lambda _
8395 ;; At build time there is no connection to the Internet, so
8396 ;; looking for updates will not work.
8397 (substitute* "Makefile.am"
8398 (("\\$\\(bindir\\)/embossupdate") ""))
8399 #t))
d10092b8 8400 (add-after 'disable-update-check 'autogen
66e40e00
RW		 8401 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
8402 (inputs
8403 `(("perl" ,perl)
8404 ("libpng" ,libpng)
8405 ("gd" ,gd)
8406 ("libx11" ,libx11)
8407 ("libharu" ,libharu)
8408 ("zlib" ,zlib)))
8409 (native-inputs
8410 `(("autoconf" ,autoconf)
8411 ("automake" ,automake)
8412 ("libtool" ,libtool)
8413 ("pkg-config" ,pkg-config)))
8414 (home-page "http://emboss.sourceforge.net")
8415 (synopsis "Molecular biology analysis suite")
8416 (description "EMBOSS is the \"European Molecular Biology Open Software
8417Suite\". EMBOSS is an analysis package specially developed for the needs of
8418the molecular biology (e.g. EMBnet) user community. The software
8419automatically copes with data in a variety of formats and even allows
8420transparent retrieval of sequence data from the web. It also provides a
8421number of libraries for the development of software in the field of molecular
8422biology. EMBOSS also integrates a range of currently available packages and
8423tools for sequence analysis into a seamless whole.")
8424 (license license:gpl2+)))
8425
1f1b20b8
RW		 8426(define-public bits
8427 (let ((revision "1")
8428 (commit "3cc4567896d9d6442923da944beb704750a08d2d"))
8429 (package
8430 (name "bits")
8431 ;; The version is 2.13.0 even though no release archives have been
8432 ;; published as yet.
8433 (version (string-append "2.13.0-" revision "." (string-take commit 9)))
8434 (source (origin
8435 (method git-fetch)
8436 (uri (git-reference
8437 (url "https://github.com/arq5x/bits.git")
8438 (commit commit)))
8439 (file-name (string-append name "-" version "-checkout"))
8440 (sha256
8441 (base32
8442 "17n2kffk4kmhivd8c98g2vr6y1s23vbg4sxlxs689wni66797hbs"))))
8443 (build-system gnu-build-system)
8444 (arguments
8445 `(#:tests? #f ;no tests included
8446 #:phases
8447 (modify-phases %standard-phases
8448 (delete 'configure)
8449 (add-after 'unpack 'remove-cuda
8450 (lambda _
8451 (substitute* "Makefile"
8452 ((".*_cuda") "")
8453 (("(bits_test_intersections) \\\\" _ match) match))
8454 #t))
8455 (replace 'install
8456 (lambda* (#:key outputs #:allow-other-keys)
8457 (copy-recursively
8458 "bin" (string-append (assoc-ref outputs "out") "/bin"))
8459 #t)))))
8460 (inputs
8461 `(("gsl" ,gsl)
8462 ("zlib" ,zlib)))
8463 (home-page "https://github.com/arq5x/bits")
8464 (synopsis "Implementation of binary interval search algorithm")
8465 (description "This package provides an implementation of the
8466BITS (Binary Interval Search) algorithm, an approach to interval set
8467intersection. It is especially suited for the comparison of diverse genomic
8468datasets and the exploration of large datasets of genome
8469intervals (e.g. genes, sequence alignments).")
8470 (license license:gpl2))))
8471
e62ffce5 8472(define-public piranha
883302da
RW		 8473 ;; There is no release tarball for the latest version. The latest commit is
8474 ;; older than one year at the time of this writing.
8475 (let ((revision "1")
8476 (commit "0466d364b71117d01e4471b74c514436cc281233"))
8477 (package
8478 (name "piranha")
8479 (version (string-append "1.2.1-" revision "." (string-take commit 9)))
8480 (source (origin
8481 (method git-fetch)
8482 (uri (git-reference
8483 (url "https://github.com/smithlabcode/piranha.git")
8484 (commit commit)))
8485 (sha256
8486 (base32
8487 "117dc0zf20c61jam69sk4abl57ah6yi6i7qra7d7y5zrbgk12q5n"))))
8488 (build-system gnu-build-system)
8489 (arguments
8490 `(#:test-target "test"
8491 #:phases
8492 (modify-phases %standard-phases
8493 (add-after 'unpack 'copy-smithlab-cpp
8494 (lambda* (#:key inputs #:allow-other-keys)
e62ffce5 8495 (for-each (lambda (file)
883302da
RW		 8496 (install-file file "./src/smithlab_cpp/"))
8497 (find-files (assoc-ref inputs "smithlab-cpp")))
8498 #t))
8499 (add-after 'install 'install-to-store
8500 (lambda* (#:key outputs #:allow-other-keys)
8501 (let* ((out (assoc-ref outputs "out"))
8502 (bin (string-append out "/bin")))
883302da
RW		 8503 (for-each (lambda (file)
8504 (install-file file bin))
8505 (find-files "bin" ".*")))
8506 #t)))
8507 #:configure-flags
8508 (list (string-append "--with-bam_tools_headers="
8509 (assoc-ref %build-inputs "bamtools") "/include/bamtools")
8510 (string-append "--with-bam_tools_library="
8511 (assoc-ref %build-inputs "bamtools") "/lib/bamtools"))))
8512 (inputs
8513 `(("bamtools" ,bamtools)
8514 ("samtools" ,samtools-0.1)
8515 ("gsl" ,gsl)
8516 ("smithlab-cpp"
8517 ,(let ((commit "3723e2db438c51501d0423429ff396c3035ba46a"))
8518 (origin
8519 (method git-fetch)
8520 (uri (git-reference
8521 (url "https://github.com/smithlabcode/smithlab_cpp.git")
8522 (commit commit)))
8523 (file-name (string-append "smithlab_cpp-" commit "-checkout"))
8524 (sha256
8525 (base32
8526 "0l4gvbwslw5ngziskja41c00x1r06l3yidv7y0xw9djibhykzy0g")))))))
8527 (native-inputs
8528 `(("python" ,python-2)))
8529 (home-page "https://github.com/smithlabcode/piranha")
8530 (synopsis "Peak-caller for CLIP-seq and RIP-seq data")
8531 (description
8532 "Piranha is a peak-caller for genomic data produced by CLIP-seq and
e62ffce5
RW		 8533RIP-seq experiments. It takes input in BED or BAM format and identifies
8534regions of statistically significant read enrichment. Additional covariates
8535may optionally be provided to further inform the peak-calling process.")
883302da 8536 (license license:gpl3+))))
e62ffce5 8537
d1e32822
RW		 8538(define-public pepr
8539 (package
8540 (name "pepr")
8541 (version "1.0.9")
8542 (source (origin
8543 (method url-fetch)
8544 (uri (string-append "https://pypi.python.org/packages/source/P"
8545 "/PePr/PePr-" version ".tar.gz"))
8546 (sha256
8547 (base32
8548 "0qxjfdpl1b1y53nccws2d85f6k74zwmx8y8sd9rszcqhfayx6gdx"))))
8549 (build-system python-build-system)
8550 (arguments
8551 `(#:python ,python-2 ; python2 only
b41a05ce 8552 #:tests? #f)) ; no tests included
d1e32822
RW		 8553 (propagated-inputs
8554 `(("python2-numpy" ,python2-numpy)
8555 ("python2-scipy" ,python2-scipy)
8556 ("python2-pysam" ,python2-pysam)))
0c6c9c00 8557 (home-page "https://github.com/shawnzhangyx/PePr")
d1e32822
RW		 8558 (synopsis "Peak-calling and prioritization pipeline for ChIP-Seq data")
8559 (description
8560 "PePr is a ChIP-Seq peak calling or differential binding analysis tool
8561that is primarily designed for data with biological replicates. It uses a
8562negative binomial distribution to model the read counts among the samples in
8563the same group, and look for consistent differences between ChIP and control
8564group or two ChIP groups run under different conditions.")
8565 (license license:gpl3+)))
6b49a37e
RJ		 8566
8567(define-public filevercmp
8568 (let ((commit "1a9b779b93d0b244040274794d402106907b71b7"))
8569 (package
8570 (name "filevercmp")
8571 (version (string-append "0-1." (string-take commit 7)))
8572 (source (origin
8573 (method url-fetch)
8574 (uri (string-append "https://github.com/ekg/filevercmp/archive/"
8575 commit ".tar.gz"))
8576 (file-name (string-append name "-" version ".tar.gz"))
8577 (sha256
8578 (base32 "0yp5jswf5j2pqc6517x277s4s6h1ss99v57kxw9gy0jkfl3yh450"))))
8579 (build-system gnu-build-system)
8580 (arguments
8581 `(#:tests? #f ; There are no tests to run.
8582 #:phases
8583 (modify-phases %standard-phases
8584 (delete 'configure) ; There is no configure phase.
8585 (replace 'install
8586 (lambda* (#:key outputs #:allow-other-keys)
8587 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
8588 (install-file "filevercmp" bin)))))))
8589 (home-page "https://github.com/ekg/filevercmp")
8590 (synopsis "This program compares version strings")
8591 (description "This program compares version strings. It intends to be a
8592replacement for strverscmp.")
8593 (license license:gpl3+))))
5fb5dffb
RW		 8594
8595(define-public multiqc
8596 (package
8597 (name "multiqc")
66865ab8 8598 (version "1.3")
5fb5dffb
RW		 8599 (source
8600 (origin
8601 (method url-fetch)
8602 (uri (pypi-uri "multiqc" version))
8603 (sha256
8604 (base32
66865ab8 8605 "0fx1sx53znbgzfhbbiyd8j6cg5llpcsl5q5c45jy2c81d12piqfd"))))
5fb5dffb
RW		 8606 (build-system python-build-system)
8607 (propagated-inputs
66865ab8
RW		 8608 `(("python-enum34" ,python-enum34)
8609 ("python-jinja2" ,python-jinja2)
5fb5dffb
RW		 8610 ("python-simplejson" ,python-simplejson)
8611 ("python-pyyaml" ,python-pyyaml)
8612 ("python-click" ,python-click)
799247d2
RW		 8613 ("python-spectra" ,python-spectra)
8614 ("python-requests" ,python-requests)
8615 ("python-markdown" ,python-markdown)
8616 ("python-lzstring" ,python-lzstring)
5fb5dffb 8617 ("python-matplotlib" ,python-matplotlib)
1778ea74
RW		 8618 ("python-numpy" ,python-numpy)
8619 ;; MultQC checks for the presence of nose at runtime.
8620 ("python-nose" ,python-nose)))
5fb5dffb
RW		 8621 (home-page "http://multiqc.info")
8622 (synopsis "Aggregate bioinformatics analysis reports")
8623 (description
8624 "MultiQC is a tool to aggregate bioinformatics results across many
8625samples into a single report. It contains modules for a large number of
8626common bioinformatics tools.")
66865ab8 8627 (license license:gpl3+)))
6193563a
RW		 8628
8629(define-public r-chipseq
8630 (package
8631 (name "r-chipseq")
eeaa6967 8632 (version "1.28.0")
6193563a
RW		 8633 (source
8634 (origin
8635 (method url-fetch)
8636 (uri (bioconductor-uri "chipseq" version))
8637 (sha256
8638 (base32
eeaa6967 8639 "1ymcq77krwjzrkzzcw7i9909cmkqa7c0675z9wzvrrk81hgdssfq"))))
6193563a
RW		 8640 (build-system r-build-system)
8641 (propagated-inputs
8642 `(("r-biocgenerics" ,r-biocgenerics)
8643 ("r-genomicranges" ,r-genomicranges)
8644 ("r-iranges" ,r-iranges)
eeaa6967 8645 ("r-lattice" ,r-lattice)
6193563a
RW		 8646 ("r-s4vectors" ,r-s4vectors)
8647 ("r-shortread" ,r-shortread)))
5713bbf1 8648 (home-page "https://bioconductor.org/packages/chipseq")
6193563a
RW		 8649 (synopsis "Package for analyzing ChIPseq data")
8650 (description
8651 "This package provides tools for processing short read data from ChIPseq
8652experiments.")
8653 (license license:artistic2.0)))
d407bdb9
RW		 8654
8655(define-public r-copyhelper
8656 (package
8657 (name "r-copyhelper")
8658 (version "1.6.0")
8659 (source
8660 (origin
8661 (method url-fetch)
5713bbf1 8662 (uri (string-append "https://bioconductor.org/packages/release/"
d407bdb9
RW		 8663 "data/experiment/src/contrib/CopyhelpeR_"
8664 version ".tar.gz"))
8665 (sha256
8666 (base32
8667 "0x7cyynjmxls9as2gg0iyp9x5fpalxmdjq914ss7i84i9zyk5bhq"))))
8668 (properties `((upstream-name . "CopyhelpeR")))
8669 (build-system r-build-system)
5713bbf1 8670 (home-page "https://bioconductor.org/packages/CopyhelpeR/")
d407bdb9
RW		 8671 (synopsis "Helper files for CopywriteR")
8672 (description
8673 "This package contains the helper files that are required to run the
8674Bioconductor package CopywriteR. It contains pre-assembled 1kb bin GC-content
8675and mappability files for the reference genomes hg18, hg19, hg38, mm9 and
8676mm10. In addition, it contains a blacklist filter to remove regions that
8677display copy number variation. Files are stored as GRanges objects from the
8678GenomicRanges Bioconductor package.")
8679 (license license:gpl2)))
3a3bf2f8
RW		 8680
8681(define-public r-copywriter
8682 (package
8683 (name "r-copywriter")
c126ba69 8684 (version "2.10.0")
3a3bf2f8
RW		 8685 (source
8686 (origin
8687 (method url-fetch)
8688 (uri (bioconductor-uri "CopywriteR" version))
8689 (sha256
8690 (base32
c126ba69 8691 "17fy2lc5yf3nh6v077kv87h53n263hqz2540lzrl0vjiqrl2plca"))))
3a3bf2f8
RW		 8692 (properties `((upstream-name . "CopywriteR")))
8693 (build-system r-build-system)
8694 (propagated-inputs
8695 `(("r-biocparallel" ,r-biocparallel)
8696 ("r-chipseq" ,r-chipseq)
8697 ("r-copyhelper" ,r-copyhelper)
8698 ("r-data-table" ,r-data-table)
8699 ("r-dnacopy" ,r-dnacopy)
8700 ("r-futile-logger" ,r-futile-logger)
8701 ("r-genomeinfodb" ,r-genomeinfodb)
8702 ("r-genomicalignments" ,r-genomicalignments)
8703 ("r-genomicranges" ,r-genomicranges)
8704 ("r-gtools" ,r-gtools)
8705 ("r-iranges" ,r-iranges)
8706 ("r-matrixstats" ,r-matrixstats)
8707 ("r-rsamtools" ,r-rsamtools)
8708 ("r-s4vectors" ,r-s4vectors)))
8709 (home-page "https://github.com/PeeperLab/CopywriteR")
8710 (synopsis "Copy number information from targeted sequencing")
8711 (description
8712 "CopywriteR extracts DNA copy number information from targeted sequencing
8713by utilizing off-target reads. It allows for extracting uniformly distributed
8714copy number information, can be used without reference, and can be applied to
8715sequencing data obtained from various techniques including chromatin
8716immunoprecipitation and target enrichment on small gene panels. Thereby,
8717CopywriteR constitutes a widely applicable alternative to available copy
8718number detection tools.")
8719 (license license:gpl2)))
dd42a330 8720
bf3fa996
RW		 8721(define-public r-methylkit
8722 (package
8723 (name "r-methylkit")
91e18679 8724 (version "1.4.0")
bf3fa996
RW		 8725 (source (origin
8726 (method url-fetch)
8727 (uri (bioconductor-uri "methylKit" version))
8728 (sha256
8729 (base32
91e18679 8730 "0h53w2mrjrg2n0ndi12k9j6cwclgwcgpy25nz7nyj971aisw02xn"))))
bf3fa996
RW		 8731 (properties `((upstream-name . "methylKit")))
8732 (build-system r-build-system)
8733 (propagated-inputs
8734 `(("r-data-table" ,r-data-table)
8735 ("r-emdbook" ,r-emdbook)
8736 ("r-fastseg" ,r-fastseg)
8737 ("r-genomeinfodb" ,r-genomeinfodb)
8738 ("r-genomicranges" ,r-genomicranges)
8739 ("r-gtools" ,r-gtools)
8740 ("r-iranges" ,r-iranges)
8741 ("r-kernsmooth" ,r-kernsmooth)
8742 ("r-limma" ,r-limma)
8743 ("r-mclust" ,r-mclust)
8744 ("r-qvalue" ,r-qvalue)
8745 ("r-r-utils" ,r-r-utils)
8746 ("r-rcpp" ,r-rcpp)
8747 ("r-rhtslib" ,r-rhtslib)
8748 ("r-rsamtools" ,r-rsamtools)
8749 ("r-rtracklayer" ,r-rtracklayer)
8750 ("r-s4vectors" ,r-s4vectors)
8751 ("r-zlibbioc" ,r-zlibbioc)))
8752 (inputs
8753 `(("zlib" ,zlib)))
11ed4108 8754 (home-page "https://github.com/al2na/methylKit")
bf3fa996
RW		 8755 (synopsis
8756 "DNA methylation analysis from high-throughput bisulfite sequencing results")
8757 (description
8758 "MethylKit is an R package for DNA methylation analysis and annotation
8759from high-throughput bisulfite sequencing. The package is designed to deal
8760with sequencing data from @dfn{Reduced representation bisulfite
8761sequencing} (RRBS) and its variants, but also target-capture methods and whole
8762genome bisulfite sequencing. It also has functions to analyze base-pair
8763resolution 5hmC data from experimental protocols such as oxBS-Seq and
8764TAB-Seq.")
8765 (license license:artistic2.0)))
8766
dd42a330
RJPB		 8767(define-public r-sva
8768 (package
8769 (name "r-sva")
fad7c0c7 8770 (version "3.26.0")
dd42a330
RJPB		 8771 (source
8772 (origin
8773 (method url-fetch)
8774 (uri (bioconductor-uri "sva" version))
8775 (sha256
8776 (base32
fad7c0c7 8777 "0q5xb68wfcnchy8rkv5ma67pmz1i91lsnvmwmj8f1c3w4xan3pgw"))))
dd42a330
RJPB		 8778 (build-system r-build-system)
8779 (propagated-inputs
aeb64f3c 8780 `(("r-genefilter" ,r-genefilter)
bfa0d1e7
RJ		 8781 ("r-mgcv" ,r-mgcv)
8782 ("r-biocparallel" ,r-biocparallel)
8783 ("r-matrixstats" ,r-matrixstats)
8784 ("r-limma" ,r-limma)))
5713bbf1 8785 (home-page "https://bioconductor.org/packages/sva")
dd42a330
RJPB		 8786 (synopsis "Surrogate variable analysis")
8787 (description
8788 "This package contains functions for removing batch effects and other
8789unwanted variation in high-throughput experiment. It also contains functions
8790for identifying and building surrogate variables for high-dimensional data
8791sets. Surrogate variables are covariates constructed directly from
8792high-dimensional data like gene expression/RNA sequencing/methylation/brain
8793imaging data that can be used in subsequent analyses to adjust for unknown,
8794unmodeled, or latent sources of noise.")
8795 (license license:artistic2.0)))
fb1e528e
RW		 8796
8797(define-public r-seqminer
8798 (package
8799 (name "r-seqminer")
5e0de298 8800 (version "6.0")
fb1e528e
RW		 8801 (source
8802 (origin
8803 (method url-fetch)
8804 (uri (cran-uri "seqminer" version))
8805 (sha256
8806 (base32
5e0de298 8807 "057j1l6dip35l1aivilapl2zv9db677b3di2pb3sfgq2sxg0ps3l"))))
fb1e528e
RW		 8808 (build-system r-build-system)
8809 (inputs
8810 `(("zlib" ,zlib)))
8811 (home-page "http://seqminer.genomic.codes")
8812 (synopsis "Read nucleotide sequence data (VCF, BCF, and METAL formats)")
8813 (description
8814 "This package provides tools to integrate nucleotide sequencing
8815data (variant call format, e.g. VCF or BCF) or meta-analysis results in R.")
8816 ;; Any version of the GPL is acceptable
8817 (license (list license:gpl2+ license:gpl3+))))
1f40e73c
RW		 8818
8819(define-public r-raremetals2
8820 (package
8821 (name "r-raremetals2")
8822 (version "0.1")
8823 (source
8824 (origin
8825 (method url-fetch)
8826 (uri (string-append "http://genome.sph.umich.edu/w/images/"
8827 "b/b7/RareMETALS2_" version ".tar.gz"))
8828 (sha256
8829 (base32
8830 "0z5ljcgvnm06ja9lm85a3cniq7slxcy37aqqkxrdidr79an5fs4s"))))
8831 (properties `((upstream-name . "RareMETALS2")))
8832 (build-system r-build-system)
8833 (propagated-inputs
8834 `(("r-seqminer" ,r-seqminer)
8835 ("r-mvtnorm" ,r-mvtnorm)
aeb64f3c 8836 ("r-mass" ,r-mass)
1f40e73c
RW		 8837 ("r-compquadform" ,r-compquadform)
8838 ("r-getopt" ,r-getopt)))
8839 (home-page "http://genome.sph.umich.edu/wiki/RareMETALS2")
8840 (synopsis "Analyze gene-level association tests for binary trait")
8841 (description
8842 "The R package rareMETALS2 is an extension of the R package rareMETALS.
8843It was designed to meta-analyze gene-level association tests for binary trait.
8844While rareMETALS offers a near-complete solution for meta-analysis of
8845gene-level tests for quantitative trait, it does not offer the optimal
8846solution for binary trait. The package rareMETALS2 offers improved features
8847for analyzing gene-level association tests in meta-analyses for binary
8848trait.")
8849 (license license:gpl3)))
0e7d058e
RW		 8850
8851(define-public r-maldiquant
8852 (package
8853 (name "r-maldiquant")
7b92acdf 8854 (version "1.16.4")
0e7d058e
RW		 8855 (source
8856 (origin
8857 (method url-fetch)
8858 (uri (cran-uri "MALDIquant" version))
8859 (sha256
8860 (base32
7b92acdf 8861 "1pmhsfvd45a44xdiml4zx3zd5fhygqyziqvygahkk9yibnyhv4cv"))))
0e7d058e
RW		 8862 (properties `((upstream-name . "MALDIquant")))
8863 (build-system r-build-system)
8864 (home-page "http://cran.r-project.org/web/packages/MALDIquant")
8865 (synopsis "Quantitative analysis of mass spectrometry data")
8866 (description
8867 "This package provides a complete analysis pipeline for matrix-assisted
8868laser desorption/ionization-time-of-flight (MALDI-TOF) and other
8869two-dimensional mass spectrometry data. In addition to commonly used plotting
8870and processing methods it includes distinctive features, namely baseline
8871subtraction methods such as morphological filters (TopHat) or the
8872statistics-sensitive non-linear iterative peak-clipping algorithm (SNIP), peak
8873alignment using warping functions, handling of replicated measurements as well
8874as allowing spectra with different resolutions.")
8875 (license license:gpl3+)))
ae262174
RW		 8876
8877(define-public r-protgenerics
8878 (package
8879 (name "r-protgenerics")
9a665dfb 8880 (version "1.10.0")
ae262174
RW		 8881 (source
8882 (origin
8883 (method url-fetch)
8884 (uri (bioconductor-uri "ProtGenerics" version))
8885 (sha256
8886 (base32
9a665dfb 8887 "16ijp50448wnabp43klx943rhdvh7x45hvy7cnpq1s4dckxhhyni"))))
ae262174
RW		 8888 (properties `((upstream-name . "ProtGenerics")))
8889 (build-system r-build-system)
8890 (home-page "https://github.com/lgatto/ProtGenerics")
8891 (synopsis "S4 generic functions for proteomics infrastructure")
8892 (description
8893 "This package provides S4 generic functions needed by Bioconductor
8894proteomics packages.")
8895 (license license:artistic2.0)))
71676a1f
RW		 8896
8897(define-public r-mzr
8898 (package
8899 (name "r-mzr")
34b96754 8900 (version "2.12.0")
71676a1f
RW		 8901 (source
8902 (origin
8903 (method url-fetch)
8904 (uri (bioconductor-uri "mzR" version))
8905 (sha256
8906 (base32
15ca1959
RW		 8907 "1x3gp30sfxz2v3k3swih9kff9b2rvk7hzhnlkp6ywlnn2wgb0q8c"))
8908 (modules '((guix build utils)))
8909 (snippet
8910 '(begin
8911 (delete-file-recursively "src/boost")
8912 #t))))
71676a1f
RW		 8913 (properties `((upstream-name . "mzR")))
8914 (build-system r-build-system)
15ca1959
RW		 8915 (arguments
8916 `(#:phases
8917 (modify-phases %standard-phases
8918 (add-after 'unpack 'use-system-boost
8919 (lambda _
8920 (substitute* "src/Makevars"
8921 (("\\./boost/libs.*") "")
8922 (("ARCH_OBJS=" line)
8923 (string-append line
8924 "\nARCH_LIBS=-lboost_system -lboost_regex \
8925-lboost_iostreams -lboost_thread -lboost_filesystem -lboost_chrono\n")))
8926 #t)))))
71676a1f 8927 (inputs
34b96754 8928 `(("boost" ,boost) ; use this instead of the bundled boost sources
d534f3a4 8929 ("netcdf" ,netcdf)))
71676a1f
RW		 8930 (propagated-inputs
8931 `(("r-biobase" ,r-biobase)
8932 ("r-biocgenerics" ,r-biocgenerics)
8933 ("r-protgenerics" ,r-protgenerics)
8934 ("r-rcpp" ,r-rcpp)
8935 ("r-zlibbioc" ,r-zlibbioc)))
8936 (home-page "https://github.com/sneumann/mzR/")
8937 (synopsis "Parser for mass spectrometry data files")
8938 (description
8939 "The mzR package provides a unified API to the common file formats and
8940parsers available for mass spectrometry data. It comes with a wrapper for the
8941ISB random access parser for mass spectrometry mzXML, mzData and mzML files.
8942The package contains the original code written by the ISB, and a subset of the
8943proteowizard library for mzML and mzIdentML. The netCDF reading code has
8944previously been used in XCMS.")
8945 (license license:artistic2.0)))
5c9d1505
RW		 8946
8947(define-public r-affyio
8948 (package
8949 (name "r-affyio")
8a8138d2 8950 (version "1.48.0")
5c9d1505
RW		 8951 (source
8952 (origin
8953 (method url-fetch)
8954 (uri (bioconductor-uri "affyio" version))
8955 (sha256
8956 (base32
8a8138d2 8957 "1pzzp3d3dbmyf34gvivfiprkpscn36rgvhrq853a1d3avcwr5ak9"))))
5c9d1505
RW		 8958 (build-system r-build-system)
8959 (propagated-inputs
8960 `(("r-zlibbioc" ,r-zlibbioc)))
8961 (inputs
8962 `(("zlib" ,zlib)))
8963 (home-page "https://github.com/bmbolstad/affyio")
8964 (synopsis "Tools for parsing Affymetrix data files")
8965 (description
8966 "This package provides routines for parsing Affymetrix data files based
8967upon file format information. The primary focus is on accessing the CEL and
8968CDF file formats.")
8969 (license license:lgpl2.0+)))
05c7e5fb
RW		 8970
8971(define-public r-affy
8972 (package
8973 (name "r-affy")
dad5f838 8974 (version "1.56.0")
05c7e5fb
RW		 8975 (source
8976 (origin
8977 (method url-fetch)
8978 (uri (bioconductor-uri "affy" version))
8979 (sha256
8980 (base32
dad5f838 8981 "0jmbkimma5ffsdkk3xp03g4lpz84gd95nkqakif2nqq6wmx0syrj"))))
05c7e5fb
RW		 8982 (build-system r-build-system)
8983 (propagated-inputs
8984 `(("r-affyio" ,r-affyio)
8985 ("r-biobase" ,r-biobase)
8986 ("r-biocgenerics" ,r-biocgenerics)
8987 ("r-biocinstaller" ,r-biocinstaller)
8988 ("r-preprocesscore" ,r-preprocesscore)
8989 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 8990 (home-page "https://bioconductor.org/packages/affy")
05c7e5fb
RW		 8991 (synopsis "Methods for affymetrix oligonucleotide arrays")
8992 (description
8993 "This package contains functions for exploratory oligonucleotide array
8994analysis.")
8995 (license license:lgpl2.0+)))
7c08afaf
RW		 8996
8997(define-public r-vsn
8998 (package
8999 (name "r-vsn")
1be0f536 9000 (version "3.46.0")
7c08afaf
RW		 9001 (source
9002 (origin
9003 (method url-fetch)
9004 (uri (bioconductor-uri "vsn" version))
9005 (sha256
9006 (base32
1be0f536 9007 "18y62phzirj75gg6v5l41jwybmk23ia6w7qhch0kxc4bl2rysw6j"))))
7c08afaf
RW		 9008 (build-system r-build-system)
9009 (propagated-inputs
9010 `(("r-affy" ,r-affy)
9011 ("r-biobase" ,r-biobase)
9012 ("r-ggplot2" ,r-ggplot2)
aeb64f3c 9013 ("r-lattice" ,r-lattice)
7c08afaf 9014 ("r-limma" ,r-limma)))
5713bbf1 9015 (home-page "https://bioconductor.org/packages/release/bioc/html/vsn.html")
7c08afaf
RW		 9016 (synopsis "Variance stabilization and calibration for microarray data")
9017 (description
9018 "The package implements a method for normalising microarray intensities,
9019and works for single- and multiple-color arrays. It can also be used for data
9020from other technologies, as long as they have similar format. The method uses
9021a robust variant of the maximum-likelihood estimator for an
9022additive-multiplicative error model and affine calibration. The model
9023incorporates data calibration step (a.k.a. normalization), a model for the
9024dependence of the variance on the mean intensity and a variance stabilizing
9025data transformation. Differences between transformed intensities are
9026analogous to \"normalized log-ratios\". However, in contrast to the latter,
9027their variance is independent of the mean, and they are usually more sensitive
9028and specific in detecting differential transcription.")
9029 (license license:artistic2.0)))
4aa7d592
RW		 9030
9031(define-public r-mzid
9032 (package
9033 (name "r-mzid")
b500dbcd 9034 (version "1.16.0")
4aa7d592
RW		 9035 (source
9036 (origin
9037 (method url-fetch)
9038 (uri (bioconductor-uri "mzID" version))
9039 (sha256
9040 (base32
b500dbcd 9041 "0yk70dka56zd8w62f03ggx3mandj91gfa767h9ajj0sd3mjmfqb9"))))
4aa7d592
RW		 9042 (properties `((upstream-name . "mzID")))
9043 (build-system r-build-system)
9044 (propagated-inputs
9045 `(("r-doparallel" ,r-doparallel)
9046 ("r-foreach" ,r-foreach)
9047 ("r-iterators" ,r-iterators)
9048 ("r-plyr" ,r-plyr)
9049 ("r-protgenerics" ,r-protgenerics)
9050 ("r-rcpp" ,r-rcpp)
9051 ("r-xml" ,r-xml)))
5713bbf1 9052 (home-page "https://bioconductor.org/packages/mzID")
4aa7d592
RW		 9053 (synopsis "Parser for mzIdentML files")
9054 (description
9055 "This package provides a parser for mzIdentML files implemented using the
9056XML package. The parser tries to be general and able to handle all types of
9057mzIdentML files with the drawback of having less pretty output than a vendor
9058specific parser.")
9059 (license license:gpl2+)))
6a67e181
RW		 9060
9061(define-public r-pcamethods
9062 (package
9063 (name "r-pcamethods")
a998c9a4 9064 (version "1.70.0")
6a67e181
RW		 9065 (source
9066 (origin
9067 (method url-fetch)
9068 (uri (bioconductor-uri "pcaMethods" version))
9069 (sha256
9070 (base32
a998c9a4 9071 "0ii235g0x0492kh8cfrf28ni0b6vd6fh7kizkqmczzqggd6b1bk8"))))
6a67e181
RW		 9072 (properties `((upstream-name . "pcaMethods")))
9073 (build-system r-build-system)
9074 (propagated-inputs
9075 `(("r-biobase" ,r-biobase)
9076 ("r-biocgenerics" ,r-biocgenerics)
aeb64f3c 9077 ("r-mass" ,r-mass)
6a67e181
RW		 9078 ("r-rcpp" ,r-rcpp)))
9079 (home-page "https://github.com/hredestig/pcamethods")
9080 (synopsis "Collection of PCA methods")
9081 (description
9082 "This package provides Bayesian PCA, Probabilistic PCA, Nipals PCA,
9083Inverse Non-Linear PCA and the conventional SVD PCA. A cluster based method
9084for missing value estimation is included for comparison. BPCA, PPCA and
9085NipalsPCA may be used to perform PCA on incomplete data as well as for
9086accurate missing value estimation. A set of methods for printing and plotting
9087the results is also provided. All PCA methods make use of the same data
9088structure (pcaRes) to provide a common interface to the PCA results.")
9089 (license license:gpl3+)))
11879284
RW		 9090
9091(define-public r-msnbase
9092 (package
9093 (name "r-msnbase")
baa00ee9 9094 (version "2.4.0")
11879284
RW		 9095 (source
9096 (origin
9097 (method url-fetch)
9098 (uri (bioconductor-uri "MSnbase" version))
9099 (sha256
9100 (base32
baa00ee9 9101 "0dqfimljhrx3gac8d1k72gppx27lz8yckyb12v4604nbviw7xd3r"))))
11879284
RW		 9102 (properties `((upstream-name . "MSnbase")))
9103 (build-system r-build-system)
9104 (propagated-inputs
9105 `(("r-affy" ,r-affy)
9106 ("r-biobase" ,r-biobase)
9107 ("r-biocgenerics" ,r-biocgenerics)
9108 ("r-biocparallel" ,r-biocparallel)
9109 ("r-digest" ,r-digest)
9110 ("r-ggplot2" ,r-ggplot2)
9111 ("r-impute" ,r-impute)
9112 ("r-iranges" ,r-iranges)
baa00ee9 9113 ("r-lattice" ,r-lattice)
11879284
RW		 9114 ("r-maldiquant" ,r-maldiquant)
9115 ("r-mzid" ,r-mzid)
9116 ("r-mzr" ,r-mzr)
9117 ("r-pcamethods" ,r-pcamethods)
9118 ("r-plyr" ,r-plyr)
9119 ("r-preprocesscore" ,r-preprocesscore)
9120 ("r-protgenerics" ,r-protgenerics)
9121 ("r-rcpp" ,r-rcpp)
11879284
RW		 9122 ("r-s4vectors" ,r-s4vectors)
9123 ("r-vsn" ,r-vsn)
9124 ("r-xml" ,r-xml)))
9125 (home-page "https://github.com/lgatto/MSnbase")
9126 (synopsis "Base functions and classes for MS-based proteomics")
9127 (description
9128 "This package provides basic plotting, data manipulation and processing
e614d639 9129of mass spectrometry based proteomics data.")
11879284 9130 (license license:artistic2.0)))
66da3a9f
RW		 9131
9132(define-public r-msnid
9133 (package
9134 (name "r-msnid")
22752bef 9135 (version "1.11.0")
66da3a9f
RW		 9136 (source
9137 (origin
9138 (method url-fetch)
9139 (uri (bioconductor-uri "MSnID" version))
9140 (sha256
9141 (base32
22752bef 9142 "1vi4ngwbayrv2jkfb4pbmdp37xn04y07rh1jcklqfh0fcrm1jdig"))))
66da3a9f
RW		 9143 (properties `((upstream-name . "MSnID")))
9144 (build-system r-build-system)
9145 (propagated-inputs
9146 `(("r-biobase" ,r-biobase)
9147 ("r-data-table" ,r-data-table)
9148 ("r-doparallel" ,r-doparallel)
9149 ("r-dplyr" ,r-dplyr)
9150 ("r-foreach" ,r-foreach)
9151 ("r-iterators" ,r-iterators)
9152 ("r-msnbase" ,r-msnbase)
9153 ("r-mzid" ,r-mzid)
9154 ("r-mzr" ,r-mzr)
9155 ("r-protgenerics" ,r-protgenerics)
9156 ("r-r-cache" ,r-r-cache)
9157 ("r-rcpp" ,r-rcpp)
9158 ("r-reshape2" ,r-reshape2)))
5713bbf1 9159 (home-page "https://bioconductor.org/packages/MSnID")
66da3a9f
RW		 9160 (synopsis "Utilities for LC-MSn proteomics identifications")
9161 (description
2923f3e5
RW		 9162 "This package extracts @dfn{tandem mass spectrometry} (MS/MS) ID data
9163from mzIdentML (leveraging the mzID package) or text files. After collating
9164the search results from multiple datasets it assesses their identification
9165quality and optimize filtering criteria to achieve the maximum number of
9166identifications while not exceeding a specified false discovery rate. It also
9167contains a number of utilities to explore the MS/MS results and assess missed
9168and irregular enzymatic cleavages, mass measurement accuracy, etc.")
66da3a9f 9169 (license license:artistic2.0)))
5ded35d8 9170
2441c284
RJPB		 9171(define-public r-seurat
9172 ;; Source releases are only made for new x.0 versions. All newer versions
9173 ;; are only released as pre-built binaries. At the time of this writing the
9174 ;; latest binary release is 1.4.0.12, which is equivalent to this commit.
9175 (let ((commit "fccb77d1452c35ee47e47ebf8e87bddb59f3b08d")
9176 (revision "1"))
9177 (package
9178 (name "r-seurat")
9179 (version (string-append "1.4.0.12-" revision "." (string-take commit 7)))
9180 (source (origin
9181 (method git-fetch)
9182 (uri (git-reference
9183 (url "https://github.com/satijalab/seurat")
9184 (commit commit)))
9185 (file-name (string-append name "-" version "-checkout"))
9186 (sha256
9187 (base32
9188 "101wq3aqrdmbfi3lqmq4iivk9iwbf10d4z216ss25hf7n9091cyl"))
9189 ;; Delete pre-built jar.
9190 (snippet
9191 '(begin (delete-file "inst/java/ModularityOptimizer.jar")
9192 #t))))
9193 (build-system r-build-system)
9194 (arguments
9195 `(#:phases
9196 (modify-phases %standard-phases
9197 (add-after 'unpack 'build-jar
9198 (lambda* (#:key inputs #:allow-other-keys)
9199 (let ((classesdir "tmp-classes"))
9200 (setenv "JAVA_HOME" (assoc-ref inputs "jdk"))
9201 (mkdir classesdir)
9202 (and (zero? (apply system* `("javac" "-d" ,classesdir
9203 ,@(find-files "java" "\\.java$"))))
9204 (zero? (system* "jar"
9205 "-cf" "inst/java/ModularityOptimizer.jar"
9206 "-C" classesdir ".")))))))))
9207 (native-inputs
9208 `(("jdk" ,icedtea "jdk")))
9209 (propagated-inputs
9210 `(("r-ape" ,r-ape)
9211 ("r-caret" ,r-caret)
9212 ("r-cowplot" ,r-cowplot)
9213 ("r-dplyr" ,r-dplyr)
9214 ("r-fastica" ,r-fastica)
9215 ("r-fnn" ,r-fnn)
9216 ("r-fpc" ,r-fpc)
9217 ("r-gdata" ,r-gdata)
9218 ("r-ggplot2" ,r-ggplot2)
9219 ("r-gplots" ,r-gplots)
9220 ("r-gridextra" ,r-gridextra)
9221 ("r-igraph" ,r-igraph)
9222 ("r-irlba" ,r-irlba)
9223 ("r-lars" ,r-lars)
9224 ("r-mixtools" ,r-mixtools)
9225 ("r-pbapply" ,r-pbapply)
9226 ("r-plyr" ,r-plyr)
9227 ("r-ranger" ,r-ranger)
9228 ("r-rcolorbrewer" ,r-rcolorbrewer)
9229 ("r-rcpp" ,r-rcpp)
9230 ("r-rcppeigen" ,r-rcppeigen)
9231 ("r-rcppprogress" ,r-rcppprogress)
9232 ("r-reshape2" ,r-reshape2)
9233 ("r-rocr" ,r-rocr)
9234 ("r-rtsne" ,r-rtsne)
9235 ("r-stringr" ,r-stringr)
9236 ("r-tclust" ,r-tclust)
9237 ("r-tsne" ,r-tsne)
9238 ("r-vgam" ,r-vgam)))
9239 (home-page "http://www.satijalab.org/seurat")
9240 (synopsis "Seurat is an R toolkit for single cell genomics")
9241 (description
9242 "This package is an R package designed for QC, analysis, and
9243exploration of single cell RNA-seq data. It easily enables widely-used
9244analytical techniques, including the identification of highly variable genes,
9245dimensionality reduction; PCA, ICA, t-SNE, standard unsupervised clustering
9246algorithms; density clustering, hierarchical clustering, k-means, and the
9247discovery of differentially expressed genes and markers.")
9248 (license license:gpl3))))
9249
d4b83651
RW		 9250(define-public r-aroma-light
9251 (package
9252 (name "r-aroma-light")
531e8090 9253 (version "3.8.0")
d4b83651
RW		 9254 (source
9255 (origin
9256 (method url-fetch)
9257 (uri (bioconductor-uri "aroma.light" version))
9258 (sha256
9259 (base32
531e8090 9260 "0crnk6851jwypqr5l5jcbbay0vi5vvdjyisaf6z2d69c39wmr6sc"))))
d4b83651
RW		 9261 (properties `((upstream-name . "aroma.light")))
9262 (build-system r-build-system)
9263 (propagated-inputs
9264 `(("r-matrixstats" ,r-matrixstats)
9265 ("r-r-methodss3" ,r-r-methodss3)
9266 ("r-r-oo" ,r-r-oo)
9267 ("r-r-utils" ,r-r-utils)))
9268 (home-page "https://github.com/HenrikBengtsson/aroma.light")
9269 (synopsis "Methods for normalization and visualization of microarray data")
9270 (description
9271 "This package provides methods for microarray analysis that take basic
9272data types such as matrices and lists of vectors. These methods can be used
9273standalone, be utilized in other packages, or be wrapped up in higher-level
9274classes.")
9275 (license license:gpl2+)))
9276
7b465327
RW		 9277(define-public r-deseq
9278 (package
9279 (name "r-deseq")
8140f680 9280 (version "1.30.0")
7b465327
RW		 9281 (source
9282 (origin
9283 (method url-fetch)
9284 (uri (bioconductor-uri "DESeq" version))
9285 (sha256
9286 (base32
8140f680 9287 "0mn5w3cy16iwwk8zxs7za6aa6cnrca75z0g45zd5zh1py5d7nfv9"))))
7b465327
RW		 9288 (properties `((upstream-name . "DESeq")))
9289 (build-system r-build-system)
9290 (propagated-inputs
9291 `(("r-biobase" ,r-biobase)
9292 ("r-biocgenerics" ,r-biocgenerics)
9293 ("r-genefilter" ,r-genefilter)
9294 ("r-geneplotter" ,r-geneplotter)
9295 ("r-lattice" ,r-lattice)
9296 ("r-locfit" ,r-locfit)
9297 ("r-mass" ,r-mass)
9298 ("r-rcolorbrewer" ,r-rcolorbrewer)))
9299 (home-page "http://www-huber.embl.de/users/anders/DESeq")
9300 (synopsis "Differential gene expression analysis")
9301 (description
9302 "This package provides tools for estimating variance-mean dependence in
9303count data from high-throughput genetic sequencing assays and for testing for
9304differential expression based on a model using the negative binomial
9305distribution.")
9306 (license license:gpl3+)))
9307
296b3c63
RW		 9308(define-public r-edaseq
9309 (package
9310 (name "r-edaseq")
00961279 9311 (version "2.12.0")
296b3c63
RW		 9312 (source
9313 (origin
9314 (method url-fetch)
9315 (uri (bioconductor-uri "EDASeq" version))
9316 (sha256
9317 (base32
00961279 9318 "07zm89zcivyn2261aq9grqmly8ji482kr9h9dyfknfdfrpv7jpwv"))))
296b3c63
RW		 9319 (properties `((upstream-name . "EDASeq")))
9320 (build-system r-build-system)
9321 (propagated-inputs
9322 `(("r-annotationdbi" ,r-annotationdbi)
9323 ("r-aroma-light" ,r-aroma-light)
9324 ("r-biobase" ,r-biobase)
9325 ("r-biocgenerics" ,r-biocgenerics)
9326 ("r-biomart" ,r-biomart)
9327 ("r-biostrings" ,r-biostrings)
9328 ("r-deseq" ,r-deseq)
9329 ("r-genomicfeatures" ,r-genomicfeatures)
9330 ("r-genomicranges" ,r-genomicranges)
9331 ("r-iranges" ,r-iranges)
9332 ("r-rsamtools" ,r-rsamtools)
9333 ("r-shortread" ,r-shortread)))
9334 (home-page "https://github.com/drisso/EDASeq")
9335 (synopsis "Exploratory data analysis and normalization for RNA-Seq")
9336 (description
9337 "This package provides support for numerical and graphical summaries of
9338RNA-Seq genomic read data. Provided within-lane normalization procedures to
9339adjust for GC-content effect (or other gene-level effects) on read counts:
9340loess robust local regression, global-scaling, and full-quantile
9341normalization. Between-lane normalization procedures to adjust for
9342distributional differences between lanes (e.g., sequencing depth):
9343global-scaling and full-quantile normalization.")
9344 (license license:artistic2.0)))
9345
5f96f30f
RW		 9346(define-public r-interactivedisplaybase
9347 (package
9348 (name "r-interactivedisplaybase")
1567c8d9 9349 (version "1.16.0")
5f96f30f
RW		 9350 (source
9351 (origin
9352 (method url-fetch)
9353 (uri (bioconductor-uri "interactiveDisplayBase" version))
9354 (sha256
9355 (base32
1567c8d9 9356 "01yb945jqqimwjgriza6yy4dnp303cdirxrhl4hjyprfdlmnz5p5"))))
5f96f30f
RW		 9357 (properties
9358 `((upstream-name . "interactiveDisplayBase")))
9359 (build-system r-build-system)
9360 (propagated-inputs
9361 `(("r-biocgenerics" ,r-biocgenerics)
9362 ("r-shiny" ,r-shiny)))
5713bbf1 9363 (home-page "https://bioconductor.org/packages/interactiveDisplayBase")
5f96f30f
RW		 9364 (synopsis "Base package for web displays of Bioconductor objects")
9365 (description
9366 "This package contains the basic methods needed to generate interactive
9367Shiny-based display methods for Bioconductor objects.")
9368 (license license:artistic2.0)))
9369
4a007ffa
RW		 9370(define-public r-annotationhub
9371 (package
9372 (name "r-annotationhub")
14b3c719 9373 (version "2.10.0")
4a007ffa
RW		 9374 (source
9375 (origin
9376 (method url-fetch)
9377 (uri (bioconductor-uri "AnnotationHub" version))
9378 (sha256
9379 (base32
14b3c719 9380 "1arfka3czw8hkv6n2d85bgibq81s2rgkwhmpaxzhy6nw39vv7y8b"))))
4a007ffa
RW		 9381 (properties `((upstream-name . "AnnotationHub")))
9382 (build-system r-build-system)
9383 (propagated-inputs
9384 `(("r-annotationdbi" ,r-annotationdbi)
9385 ("r-biocgenerics" ,r-biocgenerics)
9386 ("r-biocinstaller" ,r-biocinstaller)
14b3c719 9387 ("r-curl" ,r-curl)
4a007ffa
RW		 9388 ("r-httr" ,r-httr)
9389 ("r-interactivedisplaybase" ,r-interactivedisplaybase)
9390 ("r-rsqlite" ,r-rsqlite)
9391 ("r-s4vectors" ,r-s4vectors)
9392 ("r-yaml" ,r-yaml)))
5713bbf1 9393 (home-page "https://bioconductor.org/packages/AnnotationHub")
4a007ffa
RW		 9394 (synopsis "Client to access AnnotationHub resources")
9395 (description
9396 "This package provides a client for the Bioconductor AnnotationHub web
9397resource. The AnnotationHub web resource provides a central location where
9398genomic files (e.g. VCF, bed, wig) and other resources from standard
9399locations (e.g. UCSC, Ensembl) can be discovered. The resource includes
9400metadata about each resource, e.g., a textual description, tags, and date of
9401modification. The client creates and manages a local cache of files retrieved
9402by the user, helping with quick and reproducible access.")
9403 (license license:artistic2.0)))
9404
d4a1e918
RW		 9405(define-public r-fastseg
9406 (package
9407 (name "r-fastseg")
87fa5afa 9408 (version "1.24.0")
d4a1e918
RW		 9409 (source
9410 (origin
9411 (method url-fetch)
9412 (uri (bioconductor-uri "fastseg" version))
9413 (sha256
9414 (base32
87fa5afa 9415 "0dd7nr3klwz9ailwshnbynhd62lwb8zbbpj6jf3igpb94yi6x2jp"))))
d4a1e918
RW		 9416 (build-system r-build-system)
9417 (propagated-inputs
9418 `(("r-biobase" ,r-biobase)
9419 ("r-biocgenerics" ,r-biocgenerics)
9420 ("r-genomicranges" ,r-genomicranges)
9421 ("r-iranges" ,r-iranges)
9422 ("r-s4vectors" ,r-s4vectors)))
9423 (home-page "http://www.bioinf.jku.at/software/fastseg/index.html")
9424 (synopsis "Fast segmentation algorithm for genetic sequencing data")
9425 (description
9426 "Fastseg implements a very fast and efficient segmentation algorithm.
9427It can segment data from DNA microarrays and data from next generation
9428sequencing for example to detect copy number segments. Further it can segment
9429data from RNA microarrays like tiling arrays to identify transcripts. Most
9430generally, it can segment data given as a matrix or as a vector. Various data
9431formats can be used as input to fastseg like expression set objects for
9432microarrays or GRanges for sequencing data.")
9433 (license license:lgpl2.0+)))
9434
3121814e
RW		 9435(define-public r-keggrest
9436 (package
9437 (name "r-keggrest")
9438 (version "1.18.0")
9439 (source
9440 (origin
9441 (method url-fetch)
9442 (uri (bioconductor-uri "KEGGREST" version))
9443 (sha256
9444 (base32
9445 "1i3i88lj57wvpgjf75a23msgfsjv8pr2b4j1faga276p4fsblkhj"))))
9446 (properties `((upstream-name . "KEGGREST")))
9447 (build-system r-build-system)
9448 (propagated-inputs
9449 `(("r-biostrings" ,r-biostrings)
9450 ("r-httr" ,r-httr)
9451 ("r-png" ,r-png)))
9452 (home-page "https://bioconductor.org/packages/KEGGREST")
9453 (synopsis "Client-side REST access to KEGG")
9454 (description
9455 "This package provides a package that provides a client interface to the
9456@dfn{Kyoto Encyclopedia of Genes and Genomes} (KEGG) REST server.")
9457 (license license:artistic2.0)))
9458
e160aa1f
RW		 9459(define-public r-gage
9460 (package
9461 (name "r-gage")
9462 (version "2.28.0")
9463 (source
9464 (origin
9465 (method url-fetch)
9466 (uri (bioconductor-uri "gage" version))
9467 (sha256
9468 (base32
9469 "1r14p88q3y736pkqm4pdimf1izy1xy3xgivmj3cr4dv65kjny1zk"))))
9470 (build-system r-build-system)
9471 (propagated-inputs
9472 `(("r-annotationdbi" ,r-annotationdbi)
9473 ("r-graph" ,r-graph)
9474 ("r-keggrest" ,r-keggrest)))
9475 (home-page "http://www.biomedcentral.com/1471-2105/10/161")
9476 (synopsis "Generally applicable gene-set enrichment for pathway analysis")
9477 (description
9478 "GAGE is a published method for gene set (enrichment or GSEA) or pathway
9479analysis. GAGE is generally applicable independent of microarray or RNA-Seq
9480data attributes including sample sizes, experimental designs, assay platforms,
9481and other types of heterogeneity. The gage package provides functions for
9482basic GAGE analysis, result processing and presentation. In addition, it
9483provides demo microarray data and commonly used gene set data based on KEGG
9484pathways and GO terms. These funtions and data are also useful for gene set
9485analysis using other methods.")
9486 (license license:gpl2+)))
9487
902fb15d
RW		 9488(define-public r-genomicfiles
9489 (package
9490 (name "r-genomicfiles")
9491 (version "1.14.0")
9492 (source
9493 (origin
9494 (method url-fetch)
9495 (uri (bioconductor-uri "GenomicFiles" version))
9496 (sha256
9497 (base32
9498 "0r0wmrs5jycf1kckhnc2sgjmp336srlcjdkpbb1ymm7kazdd0s9n"))))
9499 (properties `((upstream-name . "GenomicFiles")))
9500 (build-system r-build-system)
9501 (propagated-inputs
9502 `(("r-biocgenerics" ,r-biocgenerics)
9503 ("r-biocparallel" ,r-biocparallel)
9504 ("r-genomeinfodb" ,r-genomeinfodb)
9505 ("r-genomicalignments" ,r-genomicalignments)
9506 ("r-genomicranges" ,r-genomicranges)
9507 ("r-iranges" ,r-iranges)
9508 ("r-rsamtools" ,r-rsamtools)
9509 ("r-rtracklayer" ,r-rtracklayer)
9510 ("r-s4vectors" ,r-s4vectors)
9511 ("r-summarizedexperiment" ,r-summarizedexperiment)
9512 ("r-variantannotation" ,r-variantannotation)))
9513 (home-page "https://bioconductor.org/packages/GenomicFiles")
9514 (synopsis "Distributed computing by file or by range")
9515 (description
9516 "This package provides infrastructure for parallel computations
9517distributed by file or by range. User defined mapper and reducer functions
9518provide added flexibility for data combination and manipulation.")
9519 (license license:artistic2.0)))
9520
3af149f5
RW		 9521(define-public r-complexheatmap
9522 (package
9523 (name "r-complexheatmap")
9524 (version "1.17.1")
9525 (source
9526 (origin
9527 (method url-fetch)
9528 (uri (bioconductor-uri "ComplexHeatmap" version))
9529 (sha256
9530 (base32
9531 "1x6kp55iqqsd8bhdl3qch95nfiy2y46ldbbsx1sj1v8f0b0ywwcy"))))
9532 (properties
9533 `((upstream-name . "ComplexHeatmap")))
9534 (build-system r-build-system)
9535 (propagated-inputs
9536 `(("r-circlize" ,r-circlize)
9537 ("r-colorspace" ,r-colorspace)
9538 ("r-getoptlong" ,r-getoptlong)
9539 ("r-globaloptions" ,r-globaloptions)
9540 ("r-rcolorbrewer" ,r-rcolorbrewer)))
9541 (home-page
9542 "https://github.com/jokergoo/ComplexHeatmap")
9543 (synopsis "Making Complex Heatmaps")
9544 (description
9545 "Complex heatmaps are efficient to visualize associations between
9546different sources of data sets and reveal potential structures. This package
9547provides a highly flexible way to arrange multiple heatmaps and supports
9548self-defined annotation graphics.")
9549 (license license:gpl2+)))
9550
04e2a2e9
RW		 9551(define-public r-dirichletmultinomial
9552 (package
9553 (name "r-dirichletmultinomial")
9554 (version "1.20.0")
9555 (source
9556 (origin
9557 (method url-fetch)
9558 (uri (bioconductor-uri "DirichletMultinomial" version))
9559 (sha256
9560 (base32
9561 "1c4s6x0qm20556grcd1xys9kkpnlzpasaai474malwcg6qvgi4x1"))))
9562 (properties
9563 `((upstream-name . "DirichletMultinomial")))
9564 (build-system r-build-system)
9565 (inputs
9566 `(("gsl" ,gsl)))
9567 (propagated-inputs
9568 `(("r-biocgenerics" ,r-biocgenerics)
9569 ("r-iranges" ,r-iranges)
9570 ("r-s4vectors" ,r-s4vectors)))
9571 (home-page "https://bioconductor.org/packages/DirichletMultinomial")
9572 (synopsis "Dirichlet-Multinomial mixture models for microbiome data")
9573 (description
9574 "Dirichlet-multinomial mixture models can be used to describe variability
9575in microbial metagenomic data. This package is an interface to code
9576originally made available by Holmes, Harris, and Quince, 2012, PLoS ONE 7(2):
95771-15.")
9578 (license license:lgpl3)))
9579
aa30ab5a
RW		 9580(define-public r-ensembldb
9581 (package
9582 (name "r-ensembldb")
9583 (version "2.2.0")
9584 (source
9585 (origin
9586 (method url-fetch)
9587 (uri (bioconductor-uri "ensembldb" version))
9588 (sha256
9589 (base32
9590 "1w0lca3ws5j770bmls91cn93lznvv2pc8s42nybdzz3vdxjvb4m1"))))
9591 (build-system r-build-system)
9592 (propagated-inputs
9593 `(("r-annotationdbi" ,r-annotationdbi)
9594 ("r-annotationfilter" ,r-annotationfilter)
9595 ("r-annotationhub" ,r-annotationhub)
9596 ("r-biobase" ,r-biobase)
9597 ("r-biocgenerics" ,r-biocgenerics)
9598 ("r-biostrings" ,r-biostrings)
9599 ("r-curl" ,r-curl)
9600 ("r-dbi" ,r-dbi)
9601 ("r-genomeinfodb" ,r-genomeinfodb)
9602 ("r-genomicfeatures" ,r-genomicfeatures)
9603 ("r-genomicranges" ,r-genomicranges)
9604 ("r-iranges" ,r-iranges)
9605 ("r-protgenerics" ,r-protgenerics)
9606 ("r-rsamtools" ,r-rsamtools)
9607 ("r-rsqlite" ,r-rsqlite)
9608 ("r-rtracklayer" ,r-rtracklayer)
9609 ("r-s4vectors" ,r-s4vectors)))
9610 (home-page "https://github.com/jotsetung/ensembldb")
9611 (synopsis "Utilities to create and use Ensembl-based annotation databases")
9612 (description
9613 "The package provides functions to create and use transcript-centric
9614annotation databases/packages. The annotation for the databases are directly
9615fetched from Ensembl using their Perl API. The functionality and data is
9616similar to that of the TxDb packages from the @code{GenomicFeatures} package,
9617but, in addition to retrieve all gene/transcript models and annotations from
9618the database, the @code{ensembldb} package also provides a filter framework
9619allowing to retrieve annotations for specific entries like genes encoded on a
9620chromosome region or transcript models of lincRNA genes.")
9621 ;; No version specified
9622 (license license:lgpl3+)))
9623
2f6f61fd
RW		 9624(define-public r-organismdbi
9625 (package
9626 (name "r-organismdbi")
9627 (version "1.20.0")
9628 (source
9629 (origin
9630 (method url-fetch)
9631 (uri (bioconductor-uri "OrganismDbi" version))
9632 (sha256
9633 (base32
9634 "0yxvhwn0m53wfwp0zi81x96argdf7cf1lpymc2as51apvfcnjdl8"))))
9635 (properties `((upstream-name . "OrganismDbi")))
9636 (build-system r-build-system)
9637 (propagated-inputs
9638 `(("r-annotationdbi" ,r-annotationdbi)
9639 ("r-biobase" ,r-biobase)
9640 ("r-biocgenerics" ,r-biocgenerics)
9641 ("r-biocinstaller" ,r-biocinstaller)
9642 ("r-dbi" ,r-dbi)
9643 ("r-genomicfeatures" ,r-genomicfeatures)
9644 ("r-genomicranges" ,r-genomicranges)
9645 ("r-graph" ,r-graph)
9646 ("r-iranges" ,r-iranges)
9647 ("r-rbgl" ,r-rbgl)
9648 ("r-s4vectors" ,r-s4vectors)))
9649 (home-page "https://bioconductor.org/packages/OrganismDbi")
9650 (synopsis "Software to enable the smooth interfacing of database packages")
9651 (description "The package enables a simple unified interface to several
9652annotation packages each of which has its own schema by taking advantage of
9653the fact that each of these packages implements a select methods.")
9654 (license license:artistic2.0)))
9655
7af1f403
RW		 9656(define-public r-biovizbase
9657 (package
9658 (name "r-biovizbase")
9659 (version "1.26.0")
9660 (source
9661 (origin
9662 (method url-fetch)
9663 (uri (bioconductor-uri "biovizBase" version))
9664 (sha256
9665 (base32
9666 "14l4vhj0a4ssr9m9zdzz3qpd4qw1mhgq5bmxq7jhrq3j9kmd6i2f"))))
9667 (properties `((upstream-name . "biovizBase")))
9668 (build-system r-build-system)
9669 (propagated-inputs
9670 `(("r-annotationdbi" ,r-annotationdbi)
9671 ("r-annotationfilter" ,r-annotationfilter)
9672 ("r-biocgenerics" ,r-biocgenerics)
9673 ("r-biostrings" ,r-biostrings)
9674 ("r-dichromat" ,r-dichromat)
9675 ("r-ensembldb" ,r-ensembldb)
9676 ("r-genomeinfodb" ,r-genomeinfodb)
9677 ("r-genomicalignments" ,r-genomicalignments)
9678 ("r-genomicfeatures" ,r-genomicfeatures)
9679 ("r-genomicranges" ,r-genomicranges)
9680 ("r-hmisc" ,r-hmisc)
9681 ("r-iranges" ,r-iranges)
9682 ("r-rcolorbrewer" ,r-rcolorbrewer)
9683 ("r-rsamtools" ,r-rsamtools)
9684 ("r-s4vectors" ,r-s4vectors)
9685 ("r-scales" ,r-scales)
9686 ("r-summarizedexperiment" ,r-summarizedexperiment)
9687 ("r-variantannotation" ,r-variantannotation)))
9688 (home-page "https://bioconductor.org/packages/biovizBase")
9689 (synopsis "Basic graphic utilities for visualization of genomic data")
9690 (description
9691 "The biovizBase package is designed to provide a set of utilities, color
9692schemes and conventions for genomic data. It serves as the base for various
9693high-level packages for biological data visualization. This saves development
9694effort and encourages consistency.")
9695 (license license:artistic2.0)))
9696
60faf945
RW		 9697(define-public r-ggbio
9698 (package
9699 (name "r-ggbio")
9700 (version "1.26.0")
9701 (source
9702 (origin
9703 (method url-fetch)
9704 (uri (bioconductor-uri "ggbio" version))
9705 (sha256
9706 (base32
9707 "1bqxfqy0hff87ax92z4lfbjz01ndrz7x8pzm6dlkdmi52p30krm9"))))
9708 (build-system r-build-system)
9709 (propagated-inputs
9710 `(("r-annotationdbi" ,r-annotationdbi)
9711 ("r-annotationfilter" ,r-annotationfilter)
9712 ("r-biobase" ,r-biobase)
9713 ("r-biocgenerics" ,r-biocgenerics)
9714 ("r-biostrings" ,r-biostrings)
9715 ("r-biovizbase" ,r-biovizbase)
9716 ("r-bsgenome" ,r-bsgenome)
9717 ("r-ensembldb" ,r-ensembldb)
9718 ("r-genomeinfodb" ,r-genomeinfodb)
9719 ("r-genomicalignments" ,r-genomicalignments)
9720 ("r-genomicfeatures" ,r-genomicfeatures)
9721 ("r-genomicranges" ,r-genomicranges)
9722 ("r-ggally" ,r-ggally)
9723 ("r-ggplot2" ,r-ggplot2)
9724 ("r-gridextra" ,r-gridextra)
9725 ("r-gtable" ,r-gtable)
9726 ("r-hmisc" ,r-hmisc)
9727 ("r-iranges" ,r-iranges)
9728 ("r-organismdbi" ,r-organismdbi)
9729 ("r-reshape2" ,r-reshape2)
9730 ("r-rsamtools" ,r-rsamtools)
9731 ("r-rtracklayer" ,r-rtracklayer)
9732 ("r-s4vectors" ,r-s4vectors)
9733 ("r-scales" ,r-scales)
9734 ("r-summarizedexperiment" ,r-summarizedexperiment)
9735 ("r-variantannotation" ,r-variantannotation)))
9736 (home-page "http://www.tengfei.name/ggbio/")
9737 (synopsis "Visualization tools for genomic data")
9738 (description
9739 "The ggbio package extends and specializes the grammar of graphics for
9740biological data. The graphics are designed to answer common scientific
9741questions, in particular those often asked of high throughput genomics data.
9742All core Bioconductor data structures are supported, where appropriate. The
9743package supports detailed views of particular genomic regions, as well as
9744genome-wide overviews. Supported overviews include ideograms and grand linear
9745views. High-level plots include sequence fragment length, edge-linked
9746interval to data view, mismatch pileup, and several splicing summaries.")
9747 (license license:artistic2.0)))
9748
96a392a0
RW		 9749(define-public r-gprofiler
9750 (package
9751 (name "r-gprofiler")
9752 (version "0.6.1")
9753 (source
9754 (origin
9755 (method url-fetch)
9756 (uri (cran-uri "gProfileR" version))
9757 (sha256
9758 (base32
9759 "1qix15d0wa9nspdclcawml94mng4qmr2jciv7d24py315wfsvv8p"))))
9760 (properties `((upstream-name . "gProfileR")))
9761 (build-system r-build-system)
9762 (propagated-inputs
9763 `(("r-plyr" ,r-plyr)
9764 ("r-rcurl" ,r-rcurl)))
9765 (home-page "http://cran.r-project.org/web/packages/gProfileR/")
9766 (synopsis "Interface to the g:Profiler toolkit")
9767 (description
9768 "This package provides tools for functional enrichment analysis,
9769gene identifier conversion and mapping homologous genes across related
9770organisms via the @code{g:Profiler} toolkit.")
9771 (license license:gpl2+)))
9772
e2b92799
RW		 9773(define-public r-gqtlbase
9774 (package
9775 (name "r-gqtlbase")
9776 (version "1.10.0")
9777 (source
9778 (origin
9779 (method url-fetch)
9780 (uri (bioconductor-uri "gQTLBase" version))
9781 (sha256
9782 (base32
9783 "1756vfcj2dkkgcmfkkg7qdaig36dv9gfvpypn9rbrky56wm1p035"))))
9784 (properties `((upstream-name . "gQTLBase")))
9785 (build-system r-build-system)
9786 (propagated-inputs
9787 `(("r-batchjobs" ,r-batchjobs)
9788 ("r-bbmisc" ,r-bbmisc)
9789 ("r-biocgenerics" ,r-biocgenerics)
9790 ("r-bit" ,r-bit)
9791 ("r-doparallel" ,r-doparallel)
9792 ("r-ff" ,r-ff)
9793 ("r-ffbase" ,r-ffbase)
9794 ("r-foreach" ,r-foreach)
9795 ("r-genomicfiles" ,r-genomicfiles)
9796 ("r-genomicranges" ,r-genomicranges)
9797 ("r-rtracklayer" ,r-rtracklayer)
9798 ("r-s4vectors" ,r-s4vectors)
9799 ("r-summarizedexperiment" ,r-summarizedexperiment)))
9800 (home-page "https://bioconductor.org/packages/gQTLBase")
9801 (synopsis "Infrastructure for eQTL, mQTL and similar studies")
9802 (description
9803 "The purpose of this package is to simplify the storage and interrogation
9804of @dfn{quantitative trait loci} (QTL) archives, such as eQTL, mQTL, dsQTL,
9805and more.")
9806 (license license:artistic2.0)))
9807
dd5fc8b7
RW		 9808(define-public r-snpstats
9809 (package
9810 (name "r-snpstats")
9811 (version "1.28.0")
9812 (source
9813 (origin
9814 (method url-fetch)
9815 (uri (bioconductor-uri "snpStats" version))
9816 (sha256
9817 (base32
9818 "1x9qwynh2hwl24vq02naf4mchpch7xi2pkdrlgw896k28kx0lvir"))))
9819 (properties `((upstream-name . "snpStats")))
9820 (build-system r-build-system)
9821 (inputs `(("zlib" ,zlib)))
9822 (propagated-inputs
9823 `(("r-biocgenerics" ,r-biocgenerics)
9824 ("r-matrix" ,r-matrix)
9825 ("r-survival" ,r-survival)
9826 ("r-zlibbioc" ,r-zlibbioc)))
9827 (home-page "https://bioconductor.org/packages/snpStats")
9828 (synopsis "Methods for SNP association studies")
9829 (description
9830 "This package provides classes and statistical methods for large
9831@dfn{single-nucleotide polymorphism} (SNP) association studies. This extends
9832the earlier snpMatrix package, allowing for uncertainty in genotypes.")
9833 (license license:gpl3)))
9834
421c68e3
RW		 9835(define-public r-homo-sapiens
9836 (package
9837 (name "r-homo-sapiens")
9838 (version "1.3.1")
9839 (source (origin
9840 (method url-fetch)
9841 ;; We cannot use bioconductor-uri here because this tarball is
9842 ;; located under "data/annotation/" instead of "bioc/".
9843 (uri (string-append "http://www.bioconductor.org/packages/"
9844 "release/data/annotation/src/contrib/"
9845 "Homo.sapiens_"
9846 version ".tar.gz"))
9847 (sha256
9848 (base32
9849 "151vj7h5p1c8yd5swrchk46z469p135wk50hvkl0nhgndvy0jj01"))))
9850 (properties
9851 `((upstream-name . "Homo.sapiens")))
9852 (build-system r-build-system)
9853 (propagated-inputs
9854 `(("r-genomicfeatures" ,r-genomicfeatures)
9855 ("r-go-db" ,r-go-db)
9856 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
9857 ("r-txdb-hsapiens-ucsc-hg19-knowngene" ,r-txdb-hsapiens-ucsc-hg19-knowngene)
9858 ("r-organismdbi" ,r-organismdbi)
9859 ("r-annotationdbi" ,r-annotationdbi)))
9860 (home-page "https://bioconductor.org/packages/Homo.sapiens/")
9861 (synopsis "Annotation package for the Homo.sapiens object")
9862 (description
9863 "This package contains the Homo.sapiens object to access data from
9864several related annotation packages.")
9865 (license license:artistic2.0)))
9866
15f98b0d
RW		 9867(define-public r-erma
9868 (package
9869 (name "r-erma")
9870 (version "0.10.0")
9871 (source
9872 (origin
9873 (method url-fetch)
9874 (uri (bioconductor-uri "erma" version))
9875 (sha256
9876 (base32
9877 "0gcfs9g8vvdv5vmq9b21kd8sq5mizjj49nfzd4in9zvp4b9v7x1g"))))
9878 (build-system r-build-system)
9879 (propagated-inputs
9880 `(("r-annotationdbi" ,r-annotationdbi)
9881 ("r-biobase" ,r-biobase)
9882 ("r-biocgenerics" ,r-biocgenerics)
9883 ("r-foreach" ,r-foreach)
9884 ("r-genomicfiles" ,r-genomicfiles)
9885 ("r-genomicranges" ,r-genomicranges)
9886 ("r-ggplot2" ,r-ggplot2)
9887 ("r-homo-sapiens" ,r-homo-sapiens)
9888 ("r-rtracklayer" ,r-rtracklayer)
9889 ("r-s4vectors" ,r-s4vectors)
9890 ("r-shiny" ,r-shiny)
9891 ("r-summarizedexperiment" ,r-summarizedexperiment)))
9892 (home-page "https://bioconductor.org/packages/erma")
9893 (synopsis "Epigenomic road map adventures")
9894 (description
9895 "The epigenomics road map describes locations of epigenetic marks in DNA
9896from a variety of cell types. Of interest are locations of histone
9897modifications, sites of DNA methylation, and regions of accessible chromatin.
9898This package presents a selection of elements of the road map including
9899metadata and outputs of the ChromImpute procedure applied to ENCODE cell lines
9900by Ernst and Kellis.")
9901 (license license:artistic2.0)))
9902
66c39102
RW		 9903(define-public r-ldblock
9904 (package
9905 (name "r-ldblock")
9906 (version "1.8.0")
9907 (source
9908 (origin
9909 (method url-fetch)
9910 (uri (bioconductor-uri "ldblock" version))
9911 (sha256
9912 (base32
9913 "18nfsixh6d2wfrb9laqsgly5w1frzihhak683k0p8fdf51h4aqba"))))
9914 (build-system r-build-system)
9915 (propagated-inputs
9916 `(("r-erma" ,r-erma)
9917 ("r-genomeinfodb" ,r-genomeinfodb)
9918 ("r-genomicfiles" ,r-genomicfiles)
9919 ("r-go-db" ,r-go-db)
9920 ("r-homo-sapiens" ,r-homo-sapiens)
9921 ("r-matrix" ,r-matrix)
9922 ("r-rsamtools" ,r-rsamtools)
9923 ("r-snpstats" ,r-snpstats)
9924 ("r-variantannotation" ,r-variantannotation)))
9925 (home-page "https://bioconductor.org/packages/ldblock")
9926 (synopsis "Data structures for linkage disequilibrium measures in populations")
9927 (description
9928 "This package defines data structures for @dfn{linkage
9929disequilibrium} (LD) measures in populations. Its purpose is to simplify
9930handling of existing population-level data for the purpose of flexibly
9931defining LD blocks.")
9932 (license license:artistic2.0)))
9933
794ff347
RW		 9934(define-public r-gqtlstats
9935 (package
9936 (name "r-gqtlstats")
9937 (version "1.10.0")
9938 (source
9939 (origin
9940 (method url-fetch)
9941 (uri (bioconductor-uri "gQTLstats" version))
9942 (sha256
9943 (base32
9944 "1cbdqawxzgna8rrgj3siph5sw4d2pb57qc0gn6ibfkhyk45f8gdv"))))
9945 (properties `((upstream-name . "gQTLstats")))
9946 (build-system r-build-system)
9947 (propagated-inputs
9948 `(("r-annotationdbi" ,r-annotationdbi)
9949 ("r-batchjobs" ,r-batchjobs)
9950 ("r-bbmisc" ,r-bbmisc)
9951 ("r-beeswarm" ,r-beeswarm)
9952 ("r-biobase" ,r-biobase)
9953 ("r-biocgenerics" ,r-biocgenerics)
9954 ("r-doparallel" ,r-doparallel)
9955 ("r-dplyr" ,r-dplyr)
9956 ("r-erma" ,r-erma)
9957 ("r-ffbase" ,r-ffbase)
9958 ("r-foreach" ,r-foreach)
9959 ("r-genomeinfodb" ,r-genomeinfodb)
9960 ("r-genomicfeatures" ,r-genomicfeatures)
9961 ("r-genomicfiles" ,r-genomicfiles)
9962 ("r-genomicranges" ,r-genomicranges)
9963 ("r-ggbeeswarm" ,r-ggbeeswarm)
9964 ("r-ggplot2" ,r-ggplot2)
9965 ("r-gqtlbase" ,r-gqtlbase)
9966 ("r-hardyweinberg" ,r-hardyweinberg)
9967 ("r-iranges" ,r-iranges)
9968 ("r-ldblock" ,r-ldblock)
9969 ("r-limma" ,r-limma)
9970 ("r-mgcv" ,r-mgcv)
9971 ("r-plotly" ,r-plotly)
9972 ("r-reshape2" ,r-reshape2)
9973 ("r-s4vectors" ,r-s4vectors)
9974 ("r-shiny" ,r-shiny)
9975 ("r-snpstats" ,r-snpstats)
9976 ("r-summarizedexperiment" ,r-summarizedexperiment)
9977 ("r-variantannotation" ,r-variantannotation)))
9978 (home-page "https://bioconductor.org/packages/gQTLstats")
9979 (synopsis "Computationally efficient analysis for eQTL and allied studies")
9980 (description
9981 "This package provides tools for the computationally efficient analysis
9982of @dfn{quantitative trait loci} (QTL) data, including eQTL, mQTL, dsQTL, etc.
9983The software in this package aims to support refinements and functional
9984interpretation of members of a collection of association statistics on a
9985family of feature/genome hypotheses.")
9986 (license license:artistic2.0)))
9987
42720006
RW		 9988(define-public r-gviz
9989 (package
9990 (name "r-gviz")
9991 (version "1.22.0")
9992 (source
9993 (origin
9994 (method url-fetch)
9995 (uri (bioconductor-uri "Gviz" version))
9996 (sha256
9997 (base32
9998 "1lrw65a8426wpxw975wjcaiacpp6fqa00nif1yxigyankbfs23c8"))))
9999 (properties `((upstream-name . "Gviz")))
10000 (build-system r-build-system)
10001 (propagated-inputs
10002 `(("r-annotationdbi" ,r-annotationdbi)
10003 ("r-biobase" ,r-biobase)
10004 ("r-biocgenerics" ,r-biocgenerics)
10005 ("r-biomart" ,r-biomart)
10006 ("r-biostrings" ,r-biostrings)
10007 ("r-biovizbase" ,r-biovizbase)
10008 ("r-bsgenome" ,r-bsgenome)
10009 ("r-digest" ,r-digest)
10010 ("r-genomeinfodb" ,r-genomeinfodb)
10011 ("r-genomicalignments" ,r-genomicalignments)
10012 ("r-genomicfeatures" ,r-genomicfeatures)
10013 ("r-genomicranges" ,r-genomicranges)
10014 ("r-iranges" ,r-iranges)
10015 ("r-lattice" ,r-lattice)
10016 ("r-latticeextra" ,r-latticeextra)
10017 ("r-matrixstats" ,r-matrixstats)
10018 ("r-rcolorbrewer" ,r-rcolorbrewer)
10019 ("r-rsamtools" ,r-rsamtools)
10020 ("r-rtracklayer" ,r-rtracklayer)
10021 ("r-s4vectors" ,r-s4vectors)
10022 ("r-xvector" ,r-xvector)))
10023 (home-page "https://bioconductor.org/packages/Gviz")
10024 (synopsis "Plotting data and annotation information along genomic coordinates")
10025 (description
10026 "Genomic data analyses requires integrated visualization of known genomic
10027information and new experimental data. Gviz uses the biomaRt and the
10028rtracklayer packages to perform live annotation queries to Ensembl and UCSC
10029and translates this to e.g. gene/transcript structures in viewports of the
10030grid graphics package. This results in genomic information plotted together
10031with your data.")
10032 (license license:artistic2.0)))
10033
2e4ce60e
RW		 10034(define-public r-gwascat
10035 (package
10036 (name "r-gwascat")
10037 (version "2.10.0")
10038 (source
10039 (origin
10040 (method url-fetch)
10041 (uri (bioconductor-uri "gwascat" version))
10042 (sha256
10043 (base32
10044 "0n5x5i5v6a8wpn5mxmlpkl34b4kyypmymiwww6g61zch7xqrgywi"))))
10045 (build-system r-build-system)
10046 (propagated-inputs
10047 `(("r-annotationdbi" ,r-annotationdbi)
10048 ("r-annotationhub" ,r-annotationhub)
10049 ("r-biocgenerics" ,r-biocgenerics)
10050 ("r-biostrings" ,r-biostrings)
10051 ("r-genomeinfodb" ,r-genomeinfodb)
10052 ("r-genomicfeatures" ,r-genomicfeatures)
10053 ("r-genomicranges" ,r-genomicranges)
10054 ("r-ggbio" ,r-ggbio)
10055 ("r-ggplot2" ,r-ggplot2)
10056 ("r-gqtlstats" ,r-gqtlstats)
10057 ("r-graph" ,r-graph)
10058 ("r-gviz" ,r-gviz)
10059 ("r-homo-sapiens" ,r-homo-sapiens)
10060 ("r-iranges" ,r-iranges)
10061 ("r-rsamtools" ,r-rsamtools)
10062 ("r-rtracklayer" ,r-rtracklayer)
10063 ("r-s4vectors" ,r-s4vectors)
10064 ("r-snpstats" ,r-snpstats)
10065 ("r-summarizedexperiment" ,r-summarizedexperiment)
10066 ("r-variantannotation" ,r-variantannotation)))
10067 (home-page "https://bioconductor.org/packages/gwascat")
10068 (synopsis "Tools for data in the EMBL-EBI GWAS catalog")
10069 (description
10070 "This package provides tools for representing and modeling data in the
10071EMBL-EBI GWAS catalog.")
10072 (license license:artistic2.0)))
10073
fbf54406
RW		 10074(define-public r-sushi
10075 (package
10076 (name "r-sushi")
10077 (version "1.16.0")
10078 (source (origin
10079 (method url-fetch)
10080 (uri (bioconductor-uri "Sushi" version))
10081 (sha256
10082 (base32
10083 "0axaqm480z8d0b2ldgxwm0swava1p4irc62bpl17p2k8k78g687g"))))
10084 (properties `((upstream-name . "Sushi")))
10085 (build-system r-build-system)
10086 (propagated-inputs
10087 `(("r-biomart" ,r-biomart)
10088 ("r-zoo" ,r-zoo)))
10089 (home-page "https://bioconductor.org/packages/Sushi")
10090 (synopsis "Tools for visualizing genomics data")
10091 (description
10092 "This package provides flexible, quantitative, and integrative genomic
10093visualizations for publication-quality multi-panel figures.")
10094 (license license:gpl2+)))
10095
30eb4e37
RW		 10096(define-public r-fithic
10097 (package
10098 (name "r-fithic")
10099 (version "1.4.0")
10100 (source (origin
10101 (method url-fetch)
10102 (uri (bioconductor-uri "FitHiC" version))
10103 (sha256
10104 (base32
10105 "12ylhrppi051m7nqsgq95kzd9g9wmp34i0zzfi55cjqawlpx7c6n"))))
10106 (properties `((upstream-name . "FitHiC")))
10107 (build-system r-build-system)
10108 (propagated-inputs
10109 `(("r-data-table" ,r-data-table)
10110 ("r-fdrtool" ,r-fdrtool)
10111 ("r-rcpp" ,r-rcpp)))
10112 (home-page "https://bioconductor.org/packages/FitHiC")
10113 (synopsis "Confidence estimation for intra-chromosomal contact maps")
10114 (description
10115 "Fit-Hi-C is a tool for assigning statistical confidence estimates to
10116intra-chromosomal contact maps produced by genome-wide genome architecture
10117assays such as Hi-C.")
10118 (license license:gpl2+)))
10119
c49102ec
RW		 10120(define-public r-hitc
10121 (package
10122 (name "r-hitc")
10123 (version "1.22.0")
10124 (source (origin
10125 (method url-fetch)
10126 (uri (bioconductor-uri "HiTC" version))
10127 (sha256
10128 (base32
10129 "0288xa1jy6nzvz2ha07csmp6dirjw5r7p9vy69q2wsbyzr02ymkp"))))
10130 (properties `((upstream-name . "HiTC")))
10131 (build-system r-build-system)
10132 (propagated-inputs
10133 `(("r-biostrings" ,r-biostrings)
10134 ("r-genomeinfodb" ,r-genomeinfodb)
10135 ("r-genomicranges" ,r-genomicranges)
10136 ("r-iranges" ,r-iranges)
10137 ("r-matrix" ,r-matrix)
10138 ("r-rcolorbrewer" ,r-rcolorbrewer)
10139 ("r-rtracklayer" ,r-rtracklayer)))
10140 (home-page "https://bioconductor.org/packages/HiTC")
10141 (synopsis "High throughput chromosome conformation capture analysis")
10142 (description
10143 "The HiTC package was developed to explore high-throughput \"C\" data
10144such as 5C or Hi-C. Dedicated R classes as well as standard methods for
10145quality controls, normalization, visualization, and further analysis are also
10146provided.")
10147 (license license:artistic2.0)))
10148
212f6120
RW		 10149(define-public r-qvalue
10150 (package
10151 (name "r-qvalue")
98791c25 10152 (version "2.10.0")
212f6120
RW		 10153 (source
10154 (origin
10155 (method url-fetch)
10156 (uri (bioconductor-uri "qvalue" version))
10157 (sha256
10158 (base32
98791c25 10159 "1rd9rnf16kh8wc076kahd9hsb9rfwsbzmz3kjmp0pj6rbiq0051i"))))
212f6120
RW		 10160 (build-system r-build-system)
10161 (propagated-inputs
10162 `(("r-ggplot2" ,r-ggplot2)
10163 ("r-reshape2" ,r-reshape2)))
10164 (home-page "http://github.com/jdstorey/qvalue")
10165 (synopsis "Q-value estimation for false discovery rate control")
10166 (description
10167 "This package takes a list of p-values resulting from the simultaneous
10168testing of many hypotheses and estimates their q-values and local @dfn{false
10169discovery rate} (FDR) values. The q-value of a test measures the proportion
10170of false positives incurred when that particular test is called significant.
10171The local FDR measures the posterior probability the null hypothesis is true
10172given the test's p-value. Various plots are automatically generated, allowing
10173one to make sensible significance cut-offs. The software can be applied to
10174problems in genomics, brain imaging, astrophysics, and data mining.")
10175 ;; Any version of the LGPL.
10176 (license license:lgpl3+)))
10177
5ded35d8
RW		 10178(define htslib-for-sambamba
10179 (let ((commit "2f3c3ea7b301f9b45737a793c0b2dcf0240e5ee5"))
10180 (package
10181 (inherit htslib)
10182 (name "htslib-for-sambamba")
10183 (version (string-append "1.3.1-1." (string-take commit 9)))
10184 (source
10185 (origin
10186 (method git-fetch)
10187 (uri (git-reference
10188 (url "https://github.com/lomereiter/htslib.git")
10189 (commit commit)))
10190 (file-name (string-append "htslib-" version "-checkout"))
10191 (sha256
10192 (base32
10193 "0g38g8s3npr0gjm9fahlbhiskyfws9l5i0x1ml3rakzj7az5l9c9"))))
10194 (arguments
10195 (substitute-keyword-arguments (package-arguments htslib)
10196 ((#:phases phases)
10197 `(modify-phases ,phases
d10092b8 10198 (add-after 'unpack 'bootstrap
5ded35d8
RW		 10199 (lambda _
10200 (zero? (system* "autoreconf" "-vif"))))))))
10201 (native-inputs
10202 `(("autoconf" ,autoconf)
10203 ("automake" ,automake)
10204 ,@(package-native-inputs htslib))))))
10205
10206(define-public sambamba
10207 (package
10208 (name "sambamba")
10209 (version "0.6.5")
10210 (source
10211 (origin
10212 (method url-fetch)
10213 (uri (string-append "https://github.com/lomereiter/sambamba/"
10214 "archive/v" version ".tar.gz"))
10215 (file-name (string-append name "-" version ".tar.gz"))
10216 (sha256
10217 (base32
10218 "17076gijd65a3f07zns2gvbgahiz5lriwsa6dq353ss3jl85d8vy"))))
10219 (build-system gnu-build-system)
10220 (arguments
10221 `(#:tests? #f ; there is no test target
10222 #:make-flags
10223 '("D_COMPILER=ldc2"
10224 ;; Override "--compiler" flag only.
10225 "D_FLAGS=--compiler=ldc2 -IBioD -g -d"
10226 "sambamba-ldmd2-64")
10227 #:phases
10228 (modify-phases %standard-phases
10229 (delete 'configure)
10230 (add-after 'unpack 'place-biod
10231 (lambda* (#:key inputs #:allow-other-keys)
10232 (copy-recursively (assoc-ref inputs "biod") "BioD")
10233 #t))
10234 (add-after 'unpack 'unbundle-prerequisites
10235 (lambda _
10236 (substitute* "Makefile"
10237 ((" htslib-static lz4-static") ""))
10238 #t))
10239 (replace 'install
10240 (lambda* (#:key outputs #:allow-other-keys)
10241 (let* ((out (assoc-ref outputs "out"))
10242 (bin (string-append out "/bin")))
10243 (mkdir-p bin)
10244 (install-file "build/sambamba" bin)
10245 #t))))))
10246 (native-inputs
10247 `(("ldc" ,ldc)
10248 ("rdmd" ,rdmd)
10249 ("biod"
10250 ,(let ((commit "1248586b54af4bd4dfb28ebfebfc6bf012e7a587"))
10251 (origin
10252 (method git-fetch)
10253 (uri (git-reference
10254 (url "https://github.com/biod/BioD.git")
10255 (commit commit)))
10256 (file-name (string-append "biod-"
10257 (string-take commit 9)
10258 "-checkout"))
10259 (sha256
10260 (base32
10261 "1m8hi1n7x0ri4l6s9i0x6jg4z4v94xrfdzp7mbizdipfag0m17g3")))))))
10262 (inputs
10263 `(("lz4" ,lz4)
10264 ("htslib" ,htslib-for-sambamba)))
10265 (home-page "http://lomereiter.github.io/sambamba")
10266 (synopsis "Tools for working with SAM/BAM data")
10267 (description "Sambamba is a high performance modern robust and
10268fast tool (and library), written in the D programming language, for
10269working with SAM and BAM files. Current parallelised functionality is
10270an important subset of samtools functionality, including view, index,
10271sort, markdup, and depth.")
10272 (license license:gpl2+)))
b8fffd3a
RW		 10273
10274(define-public ritornello
10275 (package
10276 (name "ritornello")
10277 (version "1.0.0")
10278 (source (origin
10279 (method url-fetch)
10280 (uri (string-append "https://github.com/KlugerLab/"
10281 "Ritornello/archive/v"
10282 version ".tar.gz"))
10283 (file-name (string-append name "-" version ".tar.gz"))
10284 (sha256
10285 (base32
10286 "02nik86gq9ljjriv6pamwlmqnfky3ads1fpklx6mc3hx6k40pg38"))))
10287 (build-system gnu-build-system)
10288 (arguments
10289 `(#:tests? #f ; there are no tests
10290 #:phases
10291 (modify-phases %standard-phases
10292 (add-after 'unpack 'patch-samtools-references
10293 (lambda* (#:key inputs #:allow-other-keys)
10294 (substitute* '("src/SamStream.h"
10295 "src/BufferedGenomeReader.h")
10296 (("<sam.h>") "<samtools/sam.h>"))
10297 #t))
10298 (delete 'configure)
10299 (replace 'install
10300 (lambda* (#:key inputs outputs #:allow-other-keys)
10301 (let* ((out (assoc-ref outputs "out"))
10302 (bin (string-append out "/bin/")))
10303 (mkdir-p bin)
10304 (install-file "bin/Ritornello" bin)
10305 #t))))))
10306 (inputs
10307 `(("samtools" ,samtools-0.1)
10308 ("fftw" ,fftw)
10309 ("boost" ,boost)
10310 ("zlib" ,zlib)))
10311 (home-page "https://github.com/KlugerLab/Ritornello")
10312 (synopsis "Control-free peak caller for ChIP-seq data")
10313 (description "Ritornello is a ChIP-seq peak calling algorithm based on
10314signal processing that can accurately call binding events without the need to
10315do a pair total DNA input or IgG control sample. It has been tested for use
10316with narrow binding events such as transcription factor ChIP-seq.")
10317 (license license:gpl3+)))
44f6c889
RW		 10318
10319(define-public trim-galore
10320 (package
10321 (name "trim-galore")
10322 (version "0.4.2")
10323 (source
10324 (origin
10325 (method url-fetch)
10326 (uri (string-append "http://www.bioinformatics.babraham.ac.uk/"
10327 "projects/trim_galore/trim_galore_v"
10328 version ".zip"))
10329 (sha256
10330 (base32
10331 "0b9qdxi4521gsrjvbhgky8g7kry9b5nx3byzaxkgxz7p4k8bn1mn"))))
10332 (build-system gnu-build-system)
10333 (arguments
10334 `(#:tests? #f ; no tests
10335 #:phases
10336 (modify-phases %standard-phases
10337 ;; The archive contains plain files.
10338 (replace 'unpack
10339 (lambda* (#:key source #:allow-other-keys)
10340 (zero? (system* "unzip" source))))
10341 (delete 'configure)
10342 (delete 'build)
10343 (add-after 'unpack 'hardcode-tool-references
10344 (lambda* (#:key inputs #:allow-other-keys)
10345 (substitute* "trim_galore"
10346 (("\\$path_to_cutadapt = 'cutadapt'")
10347 (string-append "$path_to_cutadapt = '"
10348 (assoc-ref inputs "cutadapt")
10349 "/bin/cutadapt'"))
10350 (("\\| gzip")
10351 (string-append "| "
10352 (assoc-ref inputs "gzip")
10353 "/bin/gzip"))
10354 (("\"gunzip")
10355 (string-append "\""
10356 (assoc-ref inputs "gzip")
10357 "/bin/gunzip")))
10358 #t))
10359 (replace 'install
10360 (lambda* (#:key outputs #:allow-other-keys)
10361 (let ((bin (string-append (assoc-ref outputs "out")
10362 "/bin")))
10363 (mkdir-p bin)
10364 (install-file "trim_galore" bin)
10365 #t))))))
10366 (inputs
10367 `(("gzip" ,gzip)
10368 ("perl" ,perl)
10369 ("cutadapt" ,cutadapt)))
10370 (native-inputs
10371 `(("unzip" ,unzip)))
10372 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/")
10373 (synopsis "Wrapper around Cutadapt and FastQC")
10374 (description "Trim Galore! is a wrapper script to automate quality and
10375adapter trimming as well as quality control, with some added functionality to
10376remove biased methylation positions for RRBS sequence files.")
10377 (license license:gpl3+)))
3420c905
RW		 10378
10379(define-public gess
10380 (package
10381 (name "gess")
10382 (version "1.0")
10383 (source (origin
10384 (method url-fetch)
10385 (uri (string-append "http://compbio.uthscsa.edu/"
10386 "GESS_Web/files/"
10387 "gess-" version ".src.tar.gz"))
10388 (sha256
10389 (base32
10390 "0hyk403kxscclzfs24pvdgiv0wm03kjcziqdrp5w46cb049gz0d7"))))
10391 (build-system gnu-build-system)
10392 (arguments
10393 `(#:tests? #f ; no tests
10394 #:phases
10395 (modify-phases %standard-phases
10396 (delete 'configure)
10397 (delete 'build)
10398 (replace 'install
10399 (lambda* (#:key inputs outputs #:allow-other-keys)
10400 (let* ((python (assoc-ref inputs "python"))
10401 (out (assoc-ref outputs "out"))
10402 (bin (string-append out "/bin/"))
10403 (target (string-append
10404 out "/lib/python2.7/site-packages/gess/")))
10405 (mkdir-p target)
10406 (copy-recursively "." target)
10407 ;; Make GESS.py executable
10408 (chmod (string-append target "GESS.py") #o555)
10409 ;; Add Python shebang to the top and make Matplotlib
10410 ;; usable.
10411 (substitute* (string-append target "GESS.py")
10412 (("\"\"\"Description:" line)
10413 (string-append "#!" (which "python") "
10414import matplotlib
10415matplotlib.use('Agg')
10416" line)))
10417 ;; Make sure GESS has all modules in its path
10418 (wrap-program (string-append target "GESS.py")
10419 `("PYTHONPATH" ":" prefix (,target ,(getenv "PYTHONPATH"))))
10420 (mkdir-p bin)
10421 (symlink (string-append target "GESS.py")
10422 (string-append bin "GESS.py"))
10423 #t))))))
10424 (inputs
10425 `(("python" ,python-2)
10426 ("python2-pysam" ,python2-pysam)
10427 ("python2-scipy" ,python2-scipy)
10428 ("python2-numpy" ,python2-numpy)
10429 ("python2-networkx" ,python2-networkx)
10430 ("python2-biopython" ,python2-biopython)))
10431 (home-page "http://compbio.uthscsa.edu/GESS_Web/")
10432 (synopsis "Detect exon-skipping events from raw RNA-seq data")
10433 (description
10434 "GESS is an implementation of a novel computational method to detect de
10435novo exon-skipping events directly from raw RNA-seq data without the prior
10436knowledge of gene annotation information. GESS stands for the graph-based
10437exon-skipping scanner detection scheme.")
10438 (license license:bsd-3)))
282b0151
RW		 10439
10440(define-public phylip
10441 (package
10442 (name "phylip")
10443 (version "3.696")
10444 (source
10445 (origin
10446 (method url-fetch)
10447 (uri (string-append "http://evolution.gs.washington.edu/phylip/"
10448 "download/phylip-" version ".tar.gz"))
10449 (sha256
10450 (base32
10451 "01jar1rayhr2gba2pgbw49m56rc5z4p5wn3ds0m188hrlln4a2nd"))))
10452 (build-system gnu-build-system)
10453 (arguments
10454 `(#:tests? #f ; no check target
10455 #:make-flags (list "-f" "Makefile.unx" "install")
10456 #:parallel-build? #f ; not supported
10457 #:phases
10458 (modify-phases %standard-phases
10459 (add-after 'unpack 'enter-dir
10460 (lambda _ (chdir "src") #t))
10461 (delete 'configure)
10462 (replace 'install
10463 (lambda* (#:key inputs outputs #:allow-other-keys)
10464 (let ((target (string-append (assoc-ref outputs "out")
10465 "/bin")))
10466 (mkdir-p target)
10467 (for-each (lambda (file)
10468 (install-file file target))
10469 (find-files "../exe" ".*")))
10470 #t)))))
10471 (home-page "http://evolution.genetics.washington.edu/phylip/")
10472 (synopsis "Tools for inferring phylogenies")
10473 (description "PHYLIP (the PHYLogeny Inference Package) is a package of
10474programs for inferring phylogenies (evolutionary trees).")
10475 (license license:bsd-2)))
aa163424
RW		 10476
10477(define-public imp
10478 (package
10479 (name "imp")
10480 (version "2.6.2")
10481 (source
10482 (origin
10483 (method url-fetch)
10484 (uri (string-append "https://integrativemodeling.org/"
10485 version "/download/imp-" version ".tar.gz"))
10486 (sha256
10487 (base32
10488 "0lxqx7vh79d771svr611dkilp6sn30qrbw8zvscbrm37v38d2j6h"))))
10489 (build-system cmake-build-system)
10490 (arguments
10491 `(;; FIXME: Some tests fail because they produce warnings, others fail
10492 ;; because the PYTHONPATH does not include the modeller's directory.
10493 #:tests? #f
10494 ;; Do not place libraries in an architecture-specific directory.
10495 #:configure-flags
10496 (list "-DCMAKE_INSTALL_LIBDIR=lib")))
10497 (inputs
10498 `(("boost" ,boost)
10499 ("gsl" ,gsl)
10500 ("swig" ,swig)
10501 ("hdf5" ,hdf5)
10502 ("fftw" ,fftw)
10503 ("python" ,python-2)))
10504 (propagated-inputs
10505 `(("python2-numpy" ,python2-numpy)
10506 ("python2-scipy" ,python2-scipy)
10507 ("python2-pandas" ,python2-pandas)
10508 ("python2-scikit-learn" ,python2-scikit-learn)
10509 ("python2-networkx" ,python2-networkx)))
10510 (home-page "https://integrativemodeling.org")
10511 (synopsis "Integrative modeling platform")
10512 (description "IMP's broad goal is to contribute to a comprehensive
10513structural characterization of biomolecules ranging in size and complexity
10514from small peptides to large macromolecular assemblies, by integrating data
10515from diverse biochemical and biophysical experiments. IMP provides a C++ and
10516Python toolbox for solving complex modeling problems, and a number of
10517applications for tackling some common problems in a user-friendly way.")
10518 ;; IMP is largely available under the GNU Lesser GPL; see the file
10519 ;; COPYING.LGPL for the full text of this license. Some IMP modules are
10520 ;; available under the GNU GPL (see the file COPYING.GPL).
10521 (license (list license:lgpl2.1+
10522 license:gpl3+))))
8befd094
RW		 10523
10524(define-public tadbit
10525 (package
10526 (name "tadbit")
10527 (version "0.2")
10528 (source (origin
10529 (method url-fetch)
10530 (uri (string-append "https://github.com/3DGenomes/TADbit/"
10531 "archive/v" version ".tar.gz"))
10532 (file-name (string-append name "-" version ".tar.gz"))
10533 (sha256
10534 (base32
10535 "1cnfqrl4685zar4nnw94j94nhvl2h29jm448nadqi1h05z6fdk4f"))))
10536 (build-system python-build-system)
10537 (arguments
10538 `(;; Tests are included and must be run after installation, but
10539 ;; they are incomplete and thus cannot be run.
10540 #:tests? #f
10541 #:python ,python-2
10542 #:phases
10543 (modify-phases %standard-phases
10544 (add-after 'unpack 'fix-problems-with-setup.py
10545 (lambda* (#:key outputs #:allow-other-keys)
10546 ;; setup.py opens these files for writing
10547 (chmod "_pytadbit/_version.py" #o664)
10548 (chmod "README.rst" #o664)
10549
10550 ;; Don't attempt to install the bash completions to
10551 ;; the home directory.
10552 (rename-file "extras/.bash_completion"
10553 "extras/tadbit")
10554 (substitute* "setup.py"
10555 (("\\(path.expanduser\\('~'\\)")
10556 (string-append "(\""
10557 (assoc-ref outputs "out")
10558 "/etc/bash_completion.d\""))
10559 (("extras/\\.bash_completion")
10560 "extras/tadbit"))
10561 #t)))))
10562 (inputs
10563 ;; TODO: add Chimera for visualization
10564 `(("imp" ,imp)
10565 ("mcl" ,mcl)
10566 ("python2-scipy" ,python2-scipy)
10567 ("python2-numpy" ,python2-numpy)
10568 ("python2-matplotlib" ,python2-matplotlib)
10569 ("python2-pysam" ,python2-pysam)))
10570 (home-page "http://3dgenomes.github.io/TADbit/")
10571 (synopsis "Analyze, model, and explore 3C-based data")
10572 (description
10573 "TADbit is a complete Python library to deal with all steps to analyze,
10574model, and explore 3C-based data. With TADbit the user can map FASTQ files to
10575obtain raw interaction binned matrices (Hi-C like matrices), normalize and
a147fadd 10576correct interaction matrices, identify and compare the so-called
8befd094
RW		 10577@dfn{Topologically Associating Domains} (TADs), build 3D models from the
10578interaction matrices, and finally, extract structural properties from the
10579models. TADbit is complemented by TADkit for visualizing 3D models.")
10580 (license license:gpl3+)))
94820951
RW		 10581
10582(define-public kentutils
10583 (package
10584 (name "kentutils")
10585 ;; 302.1.0 is out, but the only difference is the inclusion of
10586 ;; pre-built binaries.
10587 (version "302.0.0")
10588 (source
10589 (origin
10590 (method url-fetch)
10591 (uri (string-append "https://github.com/ENCODE-DCC/kentUtils/"
10592 "archive/v" version ".tar.gz"))
10593 (file-name (string-append name "-" version ".tar.gz"))
10594 (sha256
10595 (base32
10596 "134aja3k1cj32kbk1nnw0q9gxjb2krr15q6sga8qldzvc0585rmm"))
10597 (modules '((guix build utils)
10598 (srfi srfi-26)
10599 (ice-9 ftw)))
10600 (snippet
10601 '(begin
10602 ;; Only the contents of the specified directories are free
10603 ;; for all uses, so we remove the rest. "hg/autoSql" and
10604 ;; "hg/autoXml" are nominally free, but they depend on a
10605 ;; library that is built from the sources in "hg/lib",
10606 ;; which is nonfree.
10607 (let ((free (list "." ".."
10608 "utils" "lib" "inc" "tagStorm"
10609 "parasol" "htslib"))
10610 (directory? (lambda (file)
10611 (eq? 'directory (stat:type (stat file))))))
10612 (for-each (lambda (file)
10613 (and (directory? file)
10614 (delete-file-recursively file)))
10615 (map (cut string-append "src/" <>)
10616 (scandir "src"
10617 (lambda (file)
10618 (not (member file free)))))))
10619 ;; Only make the utils target, not the userApps target,
10620 ;; because that requires libraries we won't build.
10621 (substitute* "Makefile"
10622 ((" userApps") " utils"))
10623 ;; Only build libraries that are free.
10624 (substitute* "src/makefile"
10625 (("DIRS =.*") "DIRS =\n")
10626 (("cd jkOwnLib.*") "")
10627 ((" hgLib") "")
10628 (("cd hg.*") ""))
10629 (substitute* "src/utils/makefile"
10630 ;; These tools depend on "jkhgap.a", which is part of the
10631 ;; nonfree "src/hg/lib" directory.
10632 (("raSqlQuery") "")
10633 (("pslLiftSubrangeBlat") "")
10634
10635 ;; Do not build UCSC tools, which may require nonfree
10636 ;; components.
10637 (("ALL_APPS =.*") "ALL_APPS = $(UTILS_APPLIST)\n"))
10638 #t))))
10639 (build-system gnu-build-system)
10640 (arguments
10641 `( ;; There is no global test target and the test target for
10642 ;; individual tools depends on input files that are not
10643 ;; included.
10644 #:tests? #f
10645 #:phases
10646 (modify-phases %standard-phases
10647 (add-after 'unpack 'fix-paths
10648 (lambda _
10649 (substitute* "Makefile"
10650 (("/bin/echo") (which "echo")))
10651 #t))
10652 (add-after 'unpack 'prepare-samtabix
10653 (lambda* (#:key inputs #:allow-other-keys)
10654 (copy-recursively (assoc-ref inputs "samtabix")
10655 "samtabix")
10656 #t))
10657 (delete 'configure)
10658 (replace 'install
10659 (lambda* (#:key outputs #:allow-other-keys)
10660 (let ((bin (string-append (assoc-ref outputs "out")
10661 "/bin")))
10662 (copy-recursively "bin" bin))
10663 #t)))))
10664 (native-inputs
10665 `(("samtabix"
10666 ,(origin
10667 (method git-fetch)
10668 (uri (git-reference
10669 (url "http://genome-source.cse.ucsc.edu/samtabix.git")
10670 (commit "10fd107909c1ac4d679299908be4262a012965ba")))
10671 (sha256
10672 (base32
10673 "0c1nj64l42v395sa84n7az43xiap4i6f9n9dfz4058aqiwkhkmma"))))))
10674 (inputs
10675 `(("zlib" ,zlib)
10676 ("tcsh" ,tcsh)
10677 ("perl" ,perl)
10678 ("libpng" ,libpng)
10679 ("mysql" ,mysql)
10680 ("openssl" ,openssl)))
10681 (home-page "http://genome.cse.ucsc.edu/index.html")
10682 (synopsis "Assorted bioinformatics utilities")
10683 (description "This package provides the kentUtils, a selection of
10684bioinformatics utilities used in combination with the UCSC genome
10685browser.")
10686 ;; Only a subset of the sources are released under a non-copyleft
10687 ;; free software license. All other sources are removed in a
10688 ;; snippet. See this bug report for an explanation of how the
10689 ;; license statements apply:
10690 ;; https://github.com/ENCODE-DCC/kentUtils/issues/12
10691 (license (license:non-copyleft
10692 "http://genome.ucsc.edu/license/"
10693 "The contents of this package are free for all uses."))))
7ceb0a83
RW		 10694
10695(define-public f-seq
10696 (let ((commit "6ccded34cff38cf432deed8503648b4a66953f9b")
10697 (revision "1"))
10698 (package
10699 (name "f-seq")
c6643f2d 10700 (version (string-append "1.1-" revision "." (string-take commit 7)))
7ceb0a83
RW		 10701 (source (origin
10702 (method git-fetch)
10703 (uri (git-reference
10704 (url "https://github.com/aboyle/F-seq.git")
10705 (commit commit)))
10706 (file-name (string-append name "-" version))
10707 (sha256
10708 (base32
10709 "1nk33k0yajg2id4g59bc4szr58r2q6pdq42vgcw054m8ip9wv26h"))
10710 (modules '((guix build utils)))
10711 ;; Remove bundled Java library archives.
10712 (snippet
10713 '(begin
10714 (for-each delete-file (find-files "lib" ".*"))
10715 #t))))
10716 (build-system ant-build-system)
10717 (arguments
10718 `(#:tests? #f ; no tests included
10719 #:phases
10720 (modify-phases %standard-phases
10721 (replace 'install
c0770fee 10722 (lambda* (#:key inputs outputs #:allow-other-keys)
7ceb0a83
RW		 10723 (let* ((target (assoc-ref outputs "out"))
10724 (doc (string-append target "/share/doc/f-seq/")))
10725 (mkdir-p target)
10726 (mkdir-p doc)
10727 (substitute* "bin/linux/fseq"
c0770fee
RW		 10728 (("java") (which "java"))
10729 (("\\$REALDIR/../lib/commons-cli-1.1.jar")
10730 (string-append (assoc-ref inputs "java-commons-cli")
10731 "/share/java/commons-cli.jar"))
10732 (("REALDIR=.*")
10733 (string-append "REALDIR=" target "/bin\n")))
7ceb0a83
RW		 10734 (install-file "README.txt" doc)
10735 (install-file "bin/linux/fseq" (string-append target "/bin"))
10736 (install-file "build~/fseq.jar" (string-append target "/lib"))
10737 (copy-recursively "lib" (string-append target "/lib"))
10738 #t))))))
10739 (inputs
10740 `(("perl" ,perl)
10741 ("java-commons-cli" ,java-commons-cli)))
10742 (home-page "http://fureylab.web.unc.edu/software/fseq/")
10743 (synopsis "Feature density estimator for high-throughput sequence tags")
10744 (description
10745 "F-Seq is a software package that generates a continuous tag sequence
10746density estimation allowing identification of biologically meaningful sites
10747such as transcription factor binding sites (ChIP-seq) or regions of open
10748chromatin (DNase-seq). Output can be displayed directly in the UCSC Genome
10749Browser.")
10750 (license license:gpl3+))))
96348048
RW		 10751
10752(define-public bismark
10753 (package
10754 (name "bismark")
10755 (version "0.16.3")
10756 (source
10757 (origin
10758 (method url-fetch)
10759 (uri (string-append "https://github.com/FelixKrueger/Bismark/"
10760 "archive/" version ".tar.gz"))
10761 (file-name (string-append name "-" version ".tar.gz"))
10762 (sha256
10763 (base32
10764 "1204i0pa02ll2jn5pnxypkclnskvv7a2nwh5nxhagmhxk9wfv9sq"))))
10765 (build-system perl-build-system)
10766 (arguments
10767 `(#:tests? #f ; there are no tests
10768 #:phases
10769 (modify-phases %standard-phases
10770 (delete 'configure)
10771 (delete 'build)
10772 (replace 'install
10773 (lambda* (#:key outputs #:allow-other-keys)
10774 (let ((bin (string-append (assoc-ref outputs "out")
10775 "/bin"))
10776 (docdir (string-append (assoc-ref outputs "out")
10777 "/share/doc/bismark"))
10778 (docs '("Bismark_User_Guide.pdf"
10779 "RELEASE_NOTES.txt"))
10780 (scripts '("bismark"
10781 "bismark_genome_preparation"
10782 "bismark_methylation_extractor"
10783 "bismark2bedGraph"
10784 "bismark2report"
10785 "coverage2cytosine"
10786 "deduplicate_bismark"
10787 "bismark_sitrep.tpl"
10788 "bam2nuc"
10789 "bismark2summary")))
10790 (mkdir-p docdir)
10791 (mkdir-p bin)
10792 (for-each (lambda (file) (install-file file bin))
10793 scripts)
10794 (for-each (lambda (file) (install-file file docdir))
10795 docs)
10796 #t))))))
10797 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/bismark/")
10798 (synopsis "Map bisulfite treated sequence reads and analyze methylation")
10799 (description "Bismark is a program to map bisulfite treated sequencing
10800reads to a genome of interest and perform methylation calls in a single step.
10801The output can be easily imported into a genome viewer, such as SeqMonk, and
10802enables a researcher to analyse the methylation levels of their samples
10803straight away. Its main features are:
10804
10805@itemize
10806@item Bisulfite mapping and methylation calling in one single step
10807@item Supports single-end and paired-end read alignments
10808@item Supports ungapped and gapped alignments
10809@item Alignment seed length, number of mismatches etc are adjustable
10810@item Output discriminates between cytosine methylation in CpG, CHG
10811 and CHH context
10812@end itemize\n")
10813 (license license:gpl3+)))
dc8ddbfb
RW		 10814
10815(define-public paml
10816 (package
10817 (name "paml")
10818 (version "4.9e")
10819 (source (origin
10820 (method url-fetch)
10821 (uri (string-append "http://abacus.gene.ucl.ac.uk/software/"
10822 "paml" version ".tgz"))
10823 (sha256
10824 (base32
10825 "13zf6h9fiqghwhch2h06x1zdr6s42plsnqahflp5g7myr3han3s6"))
10826 (modules '((guix build utils)))
10827 ;; Remove Windows binaries
10828 (snippet
10829 '(begin
10830 (for-each delete-file (find-files "." "\\.exe$"))
10831 #t))))
10832 (build-system gnu-build-system)
10833 (arguments
10834 `(#:tests? #f ; there are no tests
10835 #:make-flags '("CC=gcc")
10836 #:phases
10837 (modify-phases %standard-phases
10838 (replace 'configure
10839 (lambda _
10840 (substitute* "src/BFdriver.c"
10841 (("/bin/bash") (which "bash")))
10842 (chdir "src")
10843 #t))
10844 (replace 'install
10845 (lambda* (#:key outputs #:allow-other-keys)
10846 (let ((tools '("baseml" "basemlg" "codeml"
10847 "pamp" "evolver" "yn00" "chi2"))
10848 (bin (string-append (assoc-ref outputs "out") "/bin"))
10849 (docdir (string-append (assoc-ref outputs "out")
10850 "/share/doc/paml")))
10851 (mkdir-p bin)
10852 (for-each (lambda (file) (install-file file bin)) tools)
10853 (copy-recursively "../doc" docdir)
10854 #t))))))
10855 (home-page "http://abacus.gene.ucl.ac.uk/software/paml.html")
10856 (synopsis "Phylogentic analysis by maximum likelihood")
10857 (description "PAML (for Phylogentic Analysis by Maximum Likelihood)
10858contains a few programs for model fitting and phylogenetic tree reconstruction
10859using nucleotide or amino-acid sequence data.")
10860 ;; GPLv3 only
10861 (license license:gpl3)))
jackhill's copy of GNU Guix: https://guix.gnu.org/