HCoop / jackhill / guix / guix.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
gnu: libsrtp: Fix building on mips.
[jackhill/guix/guix.git] / gnu / packages / bioinformatics.scm
CommitLineData
4e10a221 1;;; GNU Guix --- Functional package management for GNU
0047d26a 2;;; Copyright © 2014, 2015, 2016 Ricardo Wurmus <rekado@elephly.net>
9b9b7ffd 3;;; Copyright © 2015, 2016 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>
ddb83129 7;;; Copyright © 2016 Efraim Flashner <efraim@flashner.co.il>
318c0aee 8;;; Copyright © 2016 Marius Bakke <mbakke@fastmail.com>
4e10a221
RW		 9;;;
10;;; This file is part of GNU Guix.
11;;;
12;;; GNU Guix is free software; you can redistribute it and/or modify it
13;;; under the terms of the GNU General Public License as published by
14;;; the Free Software Foundation; either version 3 of the License, or (at
15;;; your option) any later version.
16;;;
17;;; GNU Guix is distributed in the hope that it will be useful, but
18;;; WITHOUT ANY WARRANTY; without even the implied warranty of
19;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20;;; GNU General Public License for more details.
21;;;
22;;; You should have received a copy of the GNU General Public License
23;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
24
25(define-module (gnu packages bioinformatics)
26 #:use-module ((guix licenses) #:prefix license:)
27 #:use-module (guix packages)
8e913213 28 #:use-module (guix utils)
4e10a221 29 #:use-module (guix download)
2c16316e 30 #:use-module (guix git-download)
ec946638 31 #:use-module (guix hg-download)
10b4a969 32 #:use-module (guix build-system ant)
4e10a221 33 #:use-module (guix build-system gnu)
d7678942 34 #:use-module (guix build-system cmake)
365c8153 35 #:use-module (guix build-system perl)
8622a072 36 #:use-module (guix build-system python)
a5002ae7 37 #:use-module (guix build-system r)
9c38b540 38 #:use-module (guix build-system ruby)
d3517eda 39 #:use-module (guix build-system trivial)
4e10a221 40 #:use-module (gnu packages)
a2950fa4 41 #:use-module (gnu packages autotools)
684bf7c7 42 #:use-module (gnu packages algebra)
d3517eda 43 #:use-module (gnu packages base)
318c0aee 44 #:use-module (gnu packages bash)
a0a71439 45 #:use-module (gnu packages bison)
e4e5a4d8 46 #:use-module (gnu packages boost)
4e10a221 47 #:use-module (gnu packages compression)
82c370de 48 #:use-module (gnu packages cpio)
1baee943 49 #:use-module (gnu packages curl)
99828fa7 50 #:use-module (gnu packages documentation)
d29150b5 51 #:use-module (gnu packages datastructures)
75dd2424 52 #:use-module (gnu packages file)
99268755 53 #:use-module (gnu packages flex)
02f35bb5 54 #:use-module (gnu packages gawk)
2409f37f 55 #:use-module (gnu packages gcc)
66e40e00 56 #:use-module (gnu packages gd)
97b9da68 57 #:use-module (gnu packages gtk)
b16728b0 58 #:use-module (gnu packages glib)
db7a3444 59 #:use-module (gnu packages groff)
50937297 60 #:use-module (gnu packages guile)
89984be4 61 #:use-module (gnu packages haskell)
66e40e00 62 #:use-module (gnu packages image)
97b9da68 63 #:use-module (gnu packages imagemagick)
15a3c3d4 64 #:use-module (gnu packages java)
51c64999 65 #:use-module (gnu packages linux)
ec946638 66 #:use-module (gnu packages logging)
36742f43 67 #:use-module (gnu packages machine-learning)
db7a3444 68 #:use-module (gnu packages man)
c833ab55 69 #:use-module (gnu packages maths)
6c2b26e2 70 #:use-module (gnu packages mpi)
4e10a221 71 #:use-module (gnu packages ncurses)
81f3e0c1 72 #:use-module (gnu packages pcre)
ceb62d54 73 #:use-module (gnu packages parallel)
66e40e00 74 #:use-module (gnu packages pdf)
4e10a221
RW		 75 #:use-module (gnu packages perl)
76 #:use-module (gnu packages pkg-config)
bfe3c685 77 #:use-module (gnu packages popt)
e4e5a4d8 78 #:use-module (gnu packages protobuf)
346a829a 79 #:use-module (gnu packages python)
ec946638 80 #:use-module (gnu packages readline)
9c38b540 81 #:use-module (gnu packages ruby)
84be3b99 82 #:use-module (gnu packages serialization)
c833ab55 83 #:use-module (gnu packages statistics)
d7678942 84 #:use-module (gnu packages tbb)
97b9da68 85 #:use-module (gnu packages tex)
db7a3444 86 #:use-module (gnu packages texinfo)
2127cedb 87 #:use-module (gnu packages textutils)
43c565d2 88 #:use-module (gnu packages time)
a2950fa4 89 #:use-module (gnu packages tls)
ce7155d5 90 #:use-module (gnu packages vim)
365c8153 91 #:use-module (gnu packages web)
c833ab55 92 #:use-module (gnu packages xml)
66e40e00 93 #:use-module (gnu packages xorg)
f7283db3
RW		 94 #:use-module (gnu packages zip)
95 #:use-module (srfi srfi-1))
4e10a221 96
8dc797fa
BW		 97(define-public aragorn
98 (package
99 (name "aragorn")
e990c81d 100 (version "1.2.38")
8dc797fa
BW		 101 (source (origin
102 (method url-fetch)
103 (uri (string-append
104 "http://mbio-serv2.mbioekol.lu.se/ARAGORN/Downloads/aragorn"
105 version ".tgz"))
106 (sha256
107 (base32
e990c81d 108 "09i1rg716smlbnixfm7q1ml2mfpaa2fpn3hwjg625ysmfwwy712b"))))
8dc797fa
BW		 109 (build-system gnu-build-system)
110 (arguments
111 `(#:tests? #f ; there are no tests
112 #:phases
113 (modify-phases %standard-phases
114 (delete 'configure)
115 (replace 'build
116 (lambda _
117 (zero? (system* "gcc"
118 "-O3"
119 "-ffast-math"
120 "-finline-functions"
121 "-o"
122 "aragorn"
123 (string-append "aragorn" ,version ".c")))))
124 (replace 'install
125 (lambda* (#:key outputs #:allow-other-keys)
126 (let* ((out (assoc-ref outputs "out"))
127 (bin (string-append out "/bin"))
128 (man (string-append out "/share/man/man1")))
129 (mkdir-p bin)
130 (copy-file "aragorn"
131 (string-append bin "/aragorn"))
132 (mkdir-p man)
133 (copy-file "aragorn.1"
134 (string-append man "/aragorn.1")))
135 #t)))))
136 (home-page "http://mbio-serv2.mbioekol.lu.se/ARAGORN")
137 (synopsis "Detect tRNA, mtRNA and tmRNA genes in nucleotide sequences")
138 (description
139 "Aragorn identifies transfer RNA, mitochondrial RNA and
140transfer-messenger RNA from nucleotide sequences, based on homology to known
141tRNA consensus sequences and RNA structure. It also outputs the secondary
142structure of the predicted RNA.")
143 (license license:gpl2)))
144
a12ba6e8
BW		 145(define-public bamm
146 (package
147 (name "bamm")
4b6da268 148 (version "1.7.3")
a12ba6e8
BW		 149 (source (origin
150 (method url-fetch)
151 ;; BamM is not available on pypi.
152 (uri (string-append
4b6da268 153 "https://github.com/Ecogenomics/BamM/archive/"
a12ba6e8
BW		 154 version ".tar.gz"))
155 (file-name (string-append name "-" version ".tar.gz"))
156 (sha256
157 (base32
4b6da268 158 "1f35yxp4pc8aadsvbpg6r4kg2jh4fkjci0iby4iyljm6980sac0s"))
a12ba6e8
BW		 159 (modules '((guix build utils)))
160 (snippet
161 `(begin
162 ;; Delete bundled htslib.
163 (delete-file-recursively "c/htslib-1.3.1")
164 #t))))
165 (build-system python-build-system)
166 (arguments
167 `(#:python ,python-2 ; BamM is Python 2 only.
168 ;; Do not use bundled libhts. Do use the bundled libcfu because it has
169 ;; been modified from its original form.
170 #:configure-flags
171 (let ((htslib (assoc-ref %build-inputs "htslib")))
172 (list "--with-libhts-lib" (string-append htslib "/lib")
173 "--with-libhts-inc" (string-append htslib "/include/htslib")))
174 #:phases
175 (modify-phases %standard-phases
176 (add-after 'unpack 'autogen
177 (lambda _
178 (with-directory-excursion "c"
179 (let ((sh (which "sh")))
180 ;; Use autogen so that 'configure' works.
181 (substitute* "autogen.sh" (("/bin/sh") sh))
182 (setenv "CONFIG_SHELL" sh)
183 (substitute* "configure" (("/bin/sh") sh))
184 (zero? (system* "./autogen.sh"))))))
185 (delete 'build)
186 ;; Run tests after installation so compilation only happens once.
187 (delete 'check)
188 (add-after 'install 'wrap-executable
189 (lambda* (#:key outputs #:allow-other-keys)
190 (let* ((out (assoc-ref outputs "out"))
191 (path (getenv "PATH")))
192 (wrap-program (string-append out "/bin/bamm")
193 `("PATH" ":" prefix (,path))))
194 #t))
195 (add-after 'wrap-executable 'post-install-check
196 (lambda* (#:key inputs outputs #:allow-other-keys)
197 (setenv "PATH"
198 (string-append (assoc-ref outputs "out")
199 "/bin:"
200 (getenv "PATH")))
201 (setenv "PYTHONPATH"
202 (string-append
203 (assoc-ref outputs "out")
204 "/lib/python"
205 (string-take (string-take-right
206 (assoc-ref inputs "python") 5) 3)
207 "/site-packages:"
208 (getenv "PYTHONPATH")))
209 ;; There are 2 errors printed, but they are safe to ignore:
210 ;; 1) [E::hts_open_format] fail to open file ...
211 ;; 2) samtools view: failed to open ...
212 (zero? (system* "nosetests")))))))
213 (native-inputs
214 `(("autoconf" ,autoconf)
215 ("automake" ,automake)
216 ("libtool" ,libtool)
217 ("zlib" ,zlib)
218 ("python-nose" ,python2-nose)
f3b98f4f 219 ("python-pysam" ,python2-pysam)))
a12ba6e8
BW		 220 (inputs
221 `(("htslib" ,htslib)
222 ("samtools" ,samtools)
223 ("bwa" ,bwa)
224 ("grep" ,grep)
225 ("sed" ,sed)
226 ("coreutils" ,coreutils)))
227 (propagated-inputs
228 `(("python-numpy" ,python2-numpy)))
229 (home-page "http://ecogenomics.github.io/BamM/")
230 (synopsis "Metagenomics-focused BAM file manipulator")
231 (description
232 "BamM is a C library, wrapped in python, to efficiently generate and
233parse BAM files, specifically for the analysis of metagenomic data. For
234instance, it implements several methods to assess contig-wise read coverage.")
235 (license license:lgpl3+)))
236
9794180d
RW		 237(define-public bamtools
238 (package
239 (name "bamtools")
240 (version "2.3.0")
241 (source (origin
242 (method url-fetch)
243 (uri (string-append
244 "https://github.com/pezmaster31/bamtools/archive/v"
245 version ".tar.gz"))
246 (file-name (string-append name "-" version ".tar.gz"))
247 (sha256
248 (base32
249 "1brry29bw2xr2l9pqn240rkqwayg85b8qq78zk2zs6nlspk4d018"))))
250 (build-system cmake-build-system)
4702cec2
RW		 251 (arguments
252 `(#:tests? #f ;no "check" target
253 #:phases
254 (modify-phases %standard-phases
255 (add-before
256 'configure 'set-ldflags
257 (lambda* (#:key outputs #:allow-other-keys)
258 (setenv "LDFLAGS"
259 (string-append
260 "-Wl,-rpath="
261 (assoc-ref outputs "out") "/lib/bamtools")))))))
9794180d
RW		 262 (inputs `(("zlib" ,zlib)))
263 (home-page "https://github.com/pezmaster31/bamtools")
264 (synopsis "C++ API and command-line toolkit for working with BAM data")
265 (description
266 "BamTools provides both a C++ API and a command-line toolkit for handling
267BAM files.")
268 (license license:expat)))
269
bdc7be59
MB		 270(define-public bcftools
271 (package
272 (name "bcftools")
273 (version "1.3.1")
274 (source (origin
275 (method url-fetch)
276 (uri (string-append
277 "https://github.com/samtools/bcftools/releases/download/"
278 version "/bcftools-" version ".tar.bz2"))
279 (sha256
280 (base32
281 "095ry68vmz9q5s1scjsa698dhgyvgw5aicz24c19iwfbai07mhqj"))
282 (modules '((guix build utils)))
283 (snippet
284 ;; Delete bundled htslib.
285 '(delete-file-recursively "htslib-1.3.1"))))
286 (build-system gnu-build-system)
287 (arguments
288 `(#:test-target "test"
289 #:make-flags
290 (list
291 "USE_GPL=1"
292 (string-append "prefix=" (assoc-ref %outputs "out"))
293 (string-append "HTSDIR=" (assoc-ref %build-inputs "htslib") "/include")
294 (string-append "HTSLIB=" (assoc-ref %build-inputs "htslib") "/lib/libhts.a")
295 (string-append "BGZIP=" (assoc-ref %build-inputs "htslib") "/bin/bgzip")
296 (string-append "TABIX=" (assoc-ref %build-inputs "htslib") "/bin/tabix"))
297 #:phases
298 (modify-phases %standard-phases
299 (add-after 'unpack 'patch-Makefile
300 (lambda _
301 (substitute* "Makefile"
302 ;; Do not attempt to build htslib.
303 (("^include \\$\\(HTSDIR\\)/htslib\\.mk") "")
304 ;; Link against GSL cblas.
305 (("-lcblas") "-lgslcblas"))
306 #t))
307 (delete 'configure)
308 (add-before 'check 'patch-tests
309 (lambda _
310 (substitute* "test/test.pl"
311 (("/bin/bash") (which "bash")))
312 #t)))))
313 (native-inputs
314 `(("htslib" ,htslib)
315 ("perl" ,perl)))
316 (inputs
317 `(("gsl" ,gsl)
318 ("zlib" ,zlib)))
319 (home-page "https://samtools.github.io/bcftools/")
320 (synopsis "Utilities for variant calling and manipulating VCFs and BCFs")
321 (description
322 "BCFtools is a set of utilities that manipulate variant calls in the
323Variant Call Format (VCF) and its binary counterpart BCF. All commands work
324transparently with both VCFs and BCFs, both uncompressed and BGZF-compressed.")
325 ;; The sources are dual MIT/GPL, but becomes GPL-only when USE_GPL=1.
326 (license (list license:gpl3+ license:expat))))
327
8dd4ff11
RW		 328(define-public bedops
329 (package
330 (name "bedops")
1bbc3b1d 331 (version "2.4.14")
8dd4ff11
RW		 332 (source (origin
333 (method url-fetch)
334 (uri (string-append "https://github.com/bedops/bedops/archive/v"
335 version ".tar.gz"))
f586c877 336 (file-name (string-append name "-" version ".tar.gz"))
8dd4ff11
RW		 337 (sha256
338 (base32
1bbc3b1d 339 "1kqbac547wyqma81cyky9n7mkgikjpsfd3nnmcm6hpqwanqgh10v"))))
8dd4ff11
RW		 340 (build-system gnu-build-system)
341 (arguments
342 '(#:tests? #f
343 #:make-flags (list (string-append "BINDIR=" %output "/bin"))
344 #:phases
345 (alist-cons-after
346 'unpack 'unpack-tarballs
347 (lambda _
348 ;; FIXME: Bedops includes tarballs of minimally patched upstream
349 ;; libraries jansson, zlib, and bzip2. We cannot just use stock
350 ;; libraries because at least one of the libraries (zlib) is
351 ;; patched to add a C++ function definition (deflateInit2cpp).
352 ;; Until the Bedops developers offer a way to link against system
353 ;; libraries we have to build the in-tree copies of these three
354 ;; libraries.
355
356 ;; See upstream discussion:
357 ;; https://github.com/bedops/bedops/issues/124
358
359 ;; Unpack the tarballs to benefit from shebang patching.
360 (with-directory-excursion "third-party"
361 (and (zero? (system* "tar" "xvf" "jansson-2.6.tar.bz2"))
362 (zero? (system* "tar" "xvf" "zlib-1.2.7.tar.bz2"))
363 (zero? (system* "tar" "xvf" "bzip2-1.0.6.tar.bz2"))))
364 ;; Disable unpacking of tarballs in Makefile.
365 (substitute* "system.mk/Makefile.linux"
366 (("^\tbzcat .*") "\t@echo \"not unpacking\"\n")
367 (("\\./configure") "CONFIG_SHELL=bash ./configure"))
368 (substitute* "third-party/zlib-1.2.7/Makefile.in"
369 (("^SHELL=.*$") "SHELL=bash\n")))
370 (alist-delete 'configure %standard-phases))))
371 (home-page "https://github.com/bedops/bedops")
372 (synopsis "Tools for high-performance genomic feature operations")
373 (description
374 "BEDOPS is a suite of tools to address common questions raised in genomic
375studies---mostly with regard to overlap and proximity relationships between
376data sets. It aims to be scalable and flexible, facilitating the efficient
377and accurate analysis and management of large-scale genomic data.
378
379BEDOPS provides tools that perform highly efficient and scalable Boolean and
380other set operations, statistical calculations, archiving, conversion and
381other management of genomic data of arbitrary scale. Tasks can be easily
382split by chromosome for distributing whole-genome analyses across a
383computational cluster.")
384 (license license:gpl2+)))
385
81de5647
RW		 386(define-public bedtools
387 (package
388 (name "bedtools")
d285657e 389 (version "2.26.0")
81de5647
RW		 390 (source (origin
391 (method url-fetch)
392 (uri (string-append "https://github.com/arq5x/bedtools2/archive/v"
393 version ".tar.gz"))
f586c877 394 (file-name (string-append name "-" version ".tar.gz"))
81de5647
RW		 395 (sha256
396 (base32
d285657e 397 "0xvri5hnp2iim1cx6mcd5d9f102p5ql41x69rd6106x1c17pinqm"))))
81de5647
RW		 398 (build-system gnu-build-system)
399 (native-inputs `(("python" ,python-2)))
400 (inputs `(("samtools" ,samtools)
401 ("zlib" ,zlib)))
402 (arguments
403 '(#:test-target "test"
404 #:phases
6573ac82 405 (modify-phases %standard-phases
6573ac82
BW		 406 (delete 'configure)
407 (replace 'install
408 (lambda* (#:key outputs #:allow-other-keys)
409 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
410 (for-each (lambda (file)
411 (install-file file bin))
412 (find-files "bin" ".*")))
413 #t)))))
81de5647
RW		 414 (home-page "https://github.com/arq5x/bedtools2")
415 (synopsis "Tools for genome analysis and arithmetic")
416 (description
417 "Collectively, the bedtools utilities are a swiss-army knife of tools for
418a wide-range of genomics analysis tasks. The most widely-used tools enable
419genome arithmetic: that is, set theory on the genome. For example, bedtools
420allows one to intersect, merge, count, complement, and shuffle genomic
421intervals from multiple files in widely-used genomic file formats such as BAM,
422BED, GFF/GTF, VCF.")
423 (license license:gpl2)))
424
9a8f309c
RW		 425;; Later releases of bedtools produce files with more columns than
426;; what Ribotaper expects.
427(define-public bedtools-2.18
428 (package (inherit bedtools)
429 (name "bedtools")
430 (version "2.18.0")
431 (source (origin
432 (method url-fetch)
433 (uri (string-append "https://github.com/arq5x/bedtools2/"
434 "archive/v" version ".tar.gz"))
435 (file-name (string-append name "-" version ".tar.gz"))
436 (sha256
437 (base32
438 "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))))
439
17dc32a4
RW		 440(define-public ribotaper
441 (package
442 (name "ribotaper")
443 (version "1.3.1")
444 (source (origin
445 (method url-fetch)
446 (uri (string-append "https://ohlerlab.mdc-berlin.de/"
447 "files/RiboTaper/RiboTaper_Version_"
448 version ".tar.gz"))
449 (sha256
450 (base32
451 "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
452 (build-system gnu-build-system)
453 (inputs
454 `(("bedtools" ,bedtools-2.18)
455 ("samtools" ,samtools-0.1)
456 ("r" ,r)
457 ("r-foreach" ,r-foreach)
458 ("r-xnomial" ,r-xnomial)
459 ("r-domc" ,r-domc)
460 ("r-multitaper" ,r-multitaper)
461 ("r-seqinr" ,r-seqinr)))
462 (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
463 (synopsis "Define translated ORFs using ribosome profiling data")
464 (description
465 "Ribotaper is a method for defining translated @dfn{open reading
466frames} (ORFs) using ribosome profiling (ribo-seq) data. This package
467provides the Ribotaper pipeline.")
468 (license license:gpl3+)))
469
a0a71439
RJ		 470(define-public bioawk
471 (package
472 (name "bioawk")
473 (version "1.0")
474 (source (origin
475 (method url-fetch)
476 (uri (string-append "https://github.com/lh3/bioawk/archive/v"
477 version ".tar.gz"))
478 (file-name (string-append name "-" version ".tar.gz"))
479 (sha256
480 (base32 "1daizxsk17ahi9n58fj8vpgwyhzrzh54bzqhanjanp88kgrz7gjw"))))
481 (build-system gnu-build-system)
482 (inputs
483 `(("zlib" ,zlib)))
484 (native-inputs
485 `(("bison" ,bison)))
486 (arguments
487 `(#:tests? #f ; There are no tests to run.
488 ;; Bison must generate files, before other targets can build.
489 #:parallel-build? #f
490 #:phases
491 (modify-phases %standard-phases
492 (delete 'configure) ; There is no configure phase.
493 (replace 'install
494 (lambda* (#:key outputs #:allow-other-keys)
495 (let* ((out (assoc-ref outputs "out"))
496 (bin (string-append out "/bin"))
497 (man (string-append out "/share/man/man1")))
498 (mkdir-p man)
499 (copy-file "awk.1" (string-append man "/bioawk.1"))
500 (install-file "bioawk" bin)))))))
501 (home-page "https://github.com/lh3/bioawk")
502 (synopsis "AWK with bioinformatics extensions")
503 (description "Bioawk is an extension to Brian Kernighan's awk, adding the
504support of several common biological data formats, including optionally gzip'ed
505BED, GFF, SAM, VCF, FASTA/Q and TAB-delimited formats with column names. It
506also adds a few built-in functions and a command line option to use TAB as the
507input/output delimiter. When the new functionality is not used, bioawk is
508intended to behave exactly the same as the original BWK awk.")
509 (license license:x11)))
510
a2fb1492
RW		 511(define-public python2-pybedtools
512 (package
513 (name "python2-pybedtools")
514 (version "0.6.9")
515 (source (origin
516 (method url-fetch)
517 (uri (string-append
518 "https://pypi.python.org/packages/source/p/pybedtools/pybedtools-"
519 version ".tar.gz"))
520 (sha256
521 (base32
522 "1ldzdxw1p4y3g2ignmggsdypvqkcwqwzhdha4rbgpih048z5p4an"))))
523 (build-system python-build-system)
524 (arguments `(#:python ,python-2)) ; no Python 3 support
525 (inputs
f2516de2 526 `(("python-matplotlib" ,python2-matplotlib)))
a2fb1492
RW		 527 (propagated-inputs
528 `(("bedtools" ,bedtools)
529 ("samtools" ,samtools)))
530 (native-inputs
f2516de2
HG		 531 `(("python-cython" ,python2-cython)
532 ("python-pyyaml" ,python2-pyyaml)
f3b98f4f 533 ("python-nose" ,python2-nose)))
a2fb1492
RW		 534 (home-page "https://pythonhosted.org/pybedtools/")
535 (synopsis "Python wrapper for BEDtools programs")
536 (description
537 "pybedtools is a Python wrapper for Aaron Quinlan's BEDtools programs,
538which are widely used for genomic interval manipulation or \"genome algebra\".
539pybedtools extends BEDTools by offering feature-level manipulations from with
540Python.")
541 (license license:gpl2+)))
542
9e12eba8
BW		 543(define-public python-biom-format
544 (package
545 (name "python-biom-format")
546 (version "2.1.5")
547 (source
548 (origin
549 (method url-fetch)
550 ;; Use GitHub as source because PyPI distribution does not contain
551 ;; test data: https://github.com/biocore/biom-format/issues/693
552 (uri (string-append "https://github.com/biocore/biom-format/archive/"
553 version ".tar.gz"))
554 (file-name (string-append name "-" version ".tar.gz"))
555 (sha256
556 (base32
557 "1n25w3p1rixbpac8iysmzcja6m4ip5r6sz19l8y6wlwi49hxn278"))))
558 (build-system python-build-system)
de96ea28 559 (propagated-inputs
9e12eba8
BW		 560 `(("python-numpy" ,python-numpy)
561 ("python-scipy" ,python-scipy)
562 ("python-future" ,python-future)
563 ("python-click" ,python-click)
564 ("python-h5py" ,python-h5py)))
565 (home-page "http://www.biom-format.org")
566 (synopsis "Biological Observation Matrix (BIOM) format utilities")
567 (description
568 "The BIOM file format is designed to be a general-use format for
569representing counts of observations e.g. operational taxonomic units, KEGG
570orthology groups or lipid types, in one or more biological samples
571e.g. microbiome samples, genomes, metagenomes.")
572 (license license:bsd-3)
573 (properties `((python2-variant . ,(delay python2-biom-format))))))
574
575(define-public python2-biom-format
576 (let ((base (package-with-python2 (strip-python2-variant python-biom-format))))
577 (package
578 (inherit base)
579 (arguments
580 `(#:phases
581 (modify-phases %standard-phases
582 ;; Do not require the unmaintained pyqi library.
583 (add-after 'unpack 'remove-pyqi
584 (lambda _
585 (substitute* "setup.py"
586 (("install_requires.append\\(\"pyqi\"\\)") "pass"))
587 #t)))
00e10c6e 588 ,@(package-arguments base))))))
9e12eba8 589
f7283db3
RW		 590(define-public bioperl-minimal
591 (let* ((inputs `(("perl-module-build" ,perl-module-build)
592 ("perl-data-stag" ,perl-data-stag)
593 ("perl-libwww" ,perl-libwww)
594 ("perl-uri" ,perl-uri)))
595 (transitive-inputs
596 (map (compose package-name cadr)
597 (delete-duplicates
598 (concatenate
599 (map (compose package-transitive-target-inputs cadr) inputs))))))
600 (package
601 (name "bioperl-minimal")
c70271ec 602 (version "1.7.0")
f7283db3
RW		 603 (source
604 (origin
605 (method url-fetch)
c70271ec
RW		 606 (uri (string-append "https://github.com/bioperl/bioperl-live/"
607 "archive/release-"
608 (string-map (lambda (c)
609 (if (char=? c #\.)
610 #\- c)) version)
611 ".tar.gz"))
f7283db3
RW		 612 (sha256
613 (base32
c70271ec 614 "12phgpxwgkqflkwfb9dcqg7a31dpjlfhar8wcgv0aj5ln4akfz06"))))
f7283db3
RW		 615 (build-system perl-build-system)
616 (arguments
617 `(#:phases
618 (modify-phases %standard-phases
619 (add-after
620 'install 'wrap-programs
621 (lambda* (#:key outputs #:allow-other-keys)
622 ;; Make sure all executables in "bin" find the required Perl
623 ;; modules at runtime. As the PERL5LIB variable contains also
624 ;; the paths of native inputs, we pick the transitive target
625 ;; inputs from %build-inputs.
626 (let* ((out (assoc-ref outputs "out"))
627 (bin (string-append out "/bin/"))
628 (path (string-join
629 (cons (string-append out "/lib/perl5/site_perl")
630 (map (lambda (name)
631 (assoc-ref %build-inputs name))
632 ',transitive-inputs))
633 ":")))
634 (for-each (lambda (file)
635 (wrap-program file
636 `("PERL5LIB" ":" prefix (,path))))
637 (find-files bin "\\.pl$"))
638 #t))))))
639 (inputs inputs)
640 (native-inputs
641 `(("perl-test-most" ,perl-test-most)))
642 (home-page "http://search.cpan.org/dist/BioPerl")
643 (synopsis "Bioinformatics toolkit")
644 (description
645 "BioPerl is the product of a community effort to produce Perl code which
646is useful in biology. Examples include Sequence objects, Alignment objects
647and database searching objects. These objects not only do what they are
648advertised to do in the documentation, but they also interact - Alignment
649objects are made from the Sequence objects, Sequence objects have access to
650Annotation and SeqFeature objects and databases, Blast objects can be
651converted to Alignment objects, and so on. This means that the objects
652provide a coordinated and extensible framework to do computational biology.")
653 (license (package-license perl)))))
654
85c37e29
RW		 655(define-public python-biopython
656 (package
657 (name "python-biopython")
4ce60305 658 (version "1.68")
85c37e29
RW		 659 (source (origin
660 (method url-fetch)
e815c094
BW		 661 ;; use PyPi rather than biopython.org to ease updating
662 (uri (pypi-uri "biopython" version))
85c37e29
RW		 663 (sha256
664 (base32
4ce60305 665 "07qc7nz0k77y8hf8s18rscvibvm91zw0kkq7ylrhisf8vp8hkp6i"))))
85c37e29 666 (build-system python-build-system)
4ce60305
BW		 667 (arguments
668 `(#:phases
669 (modify-phases %standard-phases
670 (add-before 'check 'set-home
671 ;; Some tests require a home directory to be set.
672 (lambda _ (setenv "HOME" "/tmp") #t)))))
f22efa01 673 (propagated-inputs
85c37e29 674 `(("python-numpy" ,python-numpy)))
85c37e29
RW		 675 (home-page "http://biopython.org/")
676 (synopsis "Tools for biological computation in Python")
677 (description
678 "Biopython is a set of tools for biological computation including parsers
679for bioinformatics files into Python data structures; interfaces to common
680bioinformatics programs; a standard sequence class and tools for performing
681common operations on them; code to perform data classification; code for
682dealing with alignments; code making it easy to split up parallelizable tasks
683into separate processes; and more.")
5c31f4aa 684 (license (license:non-copyleft "http://www.biopython.org/DIST/LICENSE"))))
85c37e29
RW		 685
686(define-public python2-biopython
5c31f4aa 687 (package-with-python2 python-biopython))
85c37e29 688
4b1a1528
BW		 689;; An outdated version of biopython is required for seqmagick, see
690;; https://github.com/fhcrc/seqmagick/issues/59
691;; When that issue has been resolved this package should be removed.
692(define python2-biopython-1.66
693 (package
694 (inherit python2-biopython)
695 (version "1.66")
696 (source (origin
697 (method url-fetch)
698 (uri (pypi-uri "biopython" version))
699 (sha256
700 (base32
701 "1gdv92593klimg22icf5j9by7xiq86jnwzkpz4abaa05ylkdf6hp"))))))
702
985d8411
BW		 703(define-public bpp-core
704 ;; The last release was in 2014 and the recommended way to install from source
705 ;; is to clone the git repository, so we do this.
706 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
707 (let ((commit "7d8bced0d1a87291ea8dd7046b7fb5ff9c35c582"))
708 (package
709 (name "bpp-core")
710 (version (string-append "2.2.0-1." (string-take commit 7)))
711 (source (origin
712 (method git-fetch)
713 (uri (git-reference
714 (url "http://biopp.univ-montp2.fr/git/bpp-core")
715 (commit commit)))
716 (file-name (string-append name "-" version "-checkout"))
717 (sha256
718 (base32
719 "10djsq5vlnkilv436gnmh4irpk49v29pa69r6xiryg32xmvn909j"))))
720 (build-system cmake-build-system)
721 (arguments
722 `(#:parallel-build? #f))
723 (inputs
724 `(("gcc" ,gcc-5))) ; Compilation of bpp-phyl fails with GCC 4.9 so we
725 ; compile all of the bpp packages with GCC 5.
726 (home-page "http://biopp.univ-montp2.fr")
727 (synopsis "C++ libraries for Bioinformatics")
728 (description
729 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
730analysis, phylogenetics, molecular evolution and population genetics. It is
731Object Oriented and is designed to be both easy to use and computer efficient.
732Bio++ intends to help programmers to write computer expensive programs, by
733providing them a set of re-usable tools.")
734 (license license:cecill-c))))
735
8b5f4d57
BW		 736(define-public bpp-phyl
737 ;; The last release was in 2014 and the recommended way to install from source
738 ;; is to clone the git repository, so we do this.
739 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
740 (let ((commit "0c07167b629f68b569bf274d1ad0c4af83276ae2"))
741 (package
742 (name "bpp-phyl")
743 (version (string-append "2.2.0-1." (string-take commit 7)))
744 (source (origin
745 (method git-fetch)
746 (uri (git-reference
747 (url "http://biopp.univ-montp2.fr/git/bpp-phyl")
748 (commit commit)))
749 (file-name (string-append name "-" version "-checkout"))
750 (sha256
751 (base32
752 "1ssjgchzwj3iai26kyly7gwkdv8sk59nqhkb1wpap3sf5m6kyllh"))))
753 (build-system cmake-build-system)
754 (arguments
755 `(#:parallel-build? #f
756 ;; If out-of-source, test data is not copied into the build directory
757 ;; so the tests fail.
758 #:out-of-source? #f))
759 (inputs
760 `(("bpp-core" ,bpp-core)
761 ("bpp-seq" ,bpp-seq)
762 ;; GCC 4.8 fails due to an 'internal compiler error', so we use a more
763 ;; modern GCC.
764 ("gcc" ,gcc-5)))
765 (home-page "http://biopp.univ-montp2.fr")
766 (synopsis "Bio++ phylogenetic Library")
767 (description
768 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
769analysis, phylogenetics, molecular evolution and population genetics. This
770library provides phylogenetics-related modules.")
771 (license license:cecill-c))))
772
159a7016
BW		 773(define-public bpp-popgen
774 ;; The last release was in 2014 and the recommended way to install from source
775 ;; is to clone the git repository, so we do this.
776 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
777 (let ((commit "e472bac9b1a148803895d747cd6d0c5904f85d9f"))
778 (package
779 (name "bpp-popgen")
780 (version (string-append "2.2.0-1." (string-take commit 7)))
781 (source (origin
782 (method git-fetch)
783 (uri (git-reference
784 (url "http://biopp.univ-montp2.fr/git/bpp-popgen")
785 (commit commit)))
786 (file-name (string-append name "-" version "-checkout"))
787 (sha256
788 (base32
789 "0yn82dzn1n5629nzja68xfrhi655709rjanyryb36vzkmymy6dw5"))))
790 (build-system cmake-build-system)
791 (arguments
792 `(#:parallel-build? #f
793 #:tests? #f)) ; There are no tests.
794 (inputs
795 `(("bpp-core" ,bpp-core)
796 ("bpp-seq" ,bpp-seq)
797 ("gcc" ,gcc-5)))
798 (home-page "http://biopp.univ-montp2.fr")
799 (synopsis "Bio++ population genetics library")
800 (description
801 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
802analysis, phylogenetics, molecular evolution and population genetics. This
803library provides population genetics-related modules.")
804 (license license:cecill-c))))
805
70f1bc05
BW		 806(define-public bpp-seq
807 ;; The last release was in 2014 and the recommended way to install from source
808 ;; is to clone the git repository, so we do this.
809 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
810 (let ((commit "6cfa07965ce152e5598a89df2fa80a75973bfa33"))
811 (package
812 (name "bpp-seq")
813 (version (string-append "2.2.0-1." (string-take commit 7)))
814 (source (origin
815 (method git-fetch)
816 (uri (git-reference
817 (url "http://biopp.univ-montp2.fr/git/bpp-seq")
818 (commit commit)))
819 (file-name (string-append name "-" version "-checkout"))
820 (sha256
821 (base32
822 "1nys5jq7jqvdg40d91wsmj3q2yzy4276cp7sp44n67p468f27zf2"))))
823 (build-system cmake-build-system)
824 (arguments
825 `(#:parallel-build? #f
826 ;; If out-of-source, test data is not copied into the build directory
827 ;; so the tests fail.
828 #:out-of-source? #f))
829 (inputs
830 `(("bpp-core" ,bpp-core)
831 ("gcc" ,gcc-5))) ; Use GCC 5 as per 'bpp-core'.
832 (home-page "http://biopp.univ-montp2.fr")
833 (synopsis "Bio++ sequence library")
834 (description
835 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
836analysis, phylogenetics, molecular evolution and population genetics. This
837library provides sequence-related modules.")
838 (license license:cecill-c))))
839
db7a3444
BW		 840(define-public bppsuite
841 ;; The last release was in 2014 and the recommended way to install from source
842 ;; is to clone the git repository, so we do this.
843 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
844 (let ((commit "c516147f57aa50961121cd505bed52cd7603698b"))
845 (package
846 (name "bppsuite")
847 (version (string-append "2.2.0-1." (string-take commit 7)))
848 (source (origin
849 (method git-fetch)
850 (uri (git-reference
851 (url "http://biopp.univ-montp2.fr/git/bppsuite")
852 (commit commit)))
853 (file-name (string-append name "-" version "-checkout"))
854 (sha256
855 (base32
856 "1y87pxvw0jxjizhq2dr9g2r91md45k1p9ih2sl1yy1y3p934l2kb"))))
857 (build-system cmake-build-system)
858 (arguments
859 `(#:parallel-build? #f
860 #:tests? #f)) ; There are no tests.
861 (native-inputs
862 `(("groff" ,groff)
863 ("man-db" ,man-db)
864 ("texinfo" ,texinfo)))
865 (inputs
866 `(("bpp-core" ,bpp-core)
867 ("bpp-seq" ,bpp-seq)
868 ("bpp-phyl" ,bpp-phyl)
869 ("bpp-phyl" ,bpp-popgen)
870 ("gcc" ,gcc-5)))
871 (home-page "http://biopp.univ-montp2.fr")
872 (synopsis "Bioinformatics tools written with the Bio++ libraries")
873 (description
874 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
875analysis, phylogenetics, molecular evolution and population genetics. This
876package provides command line tools using the Bio++ library.")
877 (license license:cecill-c))))
878
82c370de
RW		 879(define-public blast+
880 (package
881 (name "blast+")
8dec2229 882 (version "2.4.0")
82c370de
RW		 883 (source (origin
884 (method url-fetch)
885 (uri (string-append
886 "ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/"
887 version "/ncbi-blast-" version "+-src.tar.gz"))
888 (sha256
889 (base32
8dec2229 890 "14n9jik6vhiwjd3m7bach4xj1pzfn0szbsbyfxybd9l9cc43b6mb"))
82c370de
RW		 891 (modules '((guix build utils)))
892 (snippet
893 '(begin
894 ;; Remove bundled bzip2 and zlib
895 (delete-file-recursively "c++/src/util/compress/bzip2")
896 (delete-file-recursively "c++/src/util/compress/zlib")
897 (substitute* "c++/src/util/compress/Makefile.in"
898 (("bzip2 zlib api") "api"))
899 ;; Remove useless msbuild directory
900 (delete-file-recursively
901 "c++/src/build-system/project_tree_builder/msbuild")
902 #t))))
903 (build-system gnu-build-system)
904 (arguments
905 `(;; There are three(!) tests for this massive library, and all fail with
906 ;; "unparsable timing stats".
907 ;; ERR [127] -- [util/regexp] test_pcre.sh (unparsable timing stats)
908 ;; ERR [127] -- [serial/datatool] datatool.sh (unparsable timing stats)
909 ;; ERR [127] -- [serial/datatool] datatool_xml.sh (unparsable timing stats)
910 #:tests? #f
911 #:out-of-source? #t
912 #:parallel-build? #f ; not supported
913 #:phases
914 (modify-phases %standard-phases
915 (add-before
916 'configure 'set-HOME
917 ;; $HOME needs to be set at some point during the configure phase
918 (lambda _ (setenv "HOME" "/tmp") #t))
919 (add-after
920 'unpack 'enter-dir
921 (lambda _ (chdir "c++") #t))
922 (add-after
923 'enter-dir 'fix-build-system
924 (lambda _
925 (define (which* cmd)
926 (cond ((string=? cmd "date")
927 ;; make call to "date" deterministic
928 "date -d @0")
929 ((which cmd)
930 => identity)
931 (else
932 (format (current-error-port)
933 "WARNING: Unable to find absolute path for ~s~%"
934 cmd)
935 #f)))
936
937 ;; Rewrite hardcoded paths to various tools
938 (substitute* (append '("src/build-system/configure.ac"
939 "src/build-system/configure"
940 "scripts/common/impl/if_diff.sh"
941 "scripts/common/impl/run_with_lock.sh"
942 "src/build-system/Makefile.configurables.real"
943 "src/build-system/Makefile.in.top"
944 "src/build-system/Makefile.meta.gmake=no"
945 "src/build-system/Makefile.meta.in"
946 "src/build-system/Makefile.meta_l"
947 "src/build-system/Makefile.meta_p"
948 "src/build-system/Makefile.meta_r"
949 "src/build-system/Makefile.mk.in"
950 "src/build-system/Makefile.requirements"
951 "src/build-system/Makefile.rules_with_autodep.in")
952 (find-files "scripts/common/check" "\\.sh$"))
953 (("(/usr/bin/|/bin/)([a-z][-_.a-z]*)" all dir cmd)
954 (or (which* cmd) all)))
955
956 (substitute* (find-files "src/build-system" "^config.*")
957 (("LN_S=/bin/\\$LN_S") (string-append "LN_S=" (which "ln")))
958 (("^PATH=.*") ""))
959
960 ;; rewrite "/var/tmp" in check script
961 (substitute* "scripts/common/check/check_make_unix.sh"
962 (("/var/tmp") "/tmp"))
963
964 ;; do not reset PATH
965 (substitute* (find-files "scripts/common/impl/" "\\.sh$")
966 (("^ *PATH=.*") "")
967 (("action=/bin/") "action=")
968 (("export PATH") ":"))
969 #t))
970 (replace
971 'configure
972 (lambda* (#:key inputs outputs #:allow-other-keys)
973 (let ((out (assoc-ref outputs "out"))
974 (lib (string-append (assoc-ref outputs "lib") "/lib"))
975 (include (string-append (assoc-ref outputs "include")
976 "/include/ncbi-tools++")))
977 ;; The 'configure' script doesn't recognize things like
978 ;; '--enable-fast-install'.
979 (zero? (system* "./configure.orig"
980 (string-append "--with-build-root=" (getcwd) "/build")
981 (string-append "--prefix=" out)
982 (string-append "--libdir=" lib)
983 (string-append "--includedir=" include)
984 (string-append "--with-bz2="
985 (assoc-ref inputs "bzip2"))
986 (string-append "--with-z="
987 (assoc-ref inputs "zlib"))
988 ;; Each library is built twice by default, once
989 ;; with "-static" in its name, and again
990 ;; without.
991 "--without-static"
992 "--with-dll"))))))))
993 (outputs '("out" ; 19 MB
994 "lib" ; 203 MB
995 "include")) ; 32 MB
996 (inputs
997 `(("bzip2" ,bzip2)
998 ("zlib" ,zlib)))
999 (native-inputs
1000 `(("cpio" ,cpio)))
1001 (home-page "http://blast.ncbi.nlm.nih.gov")
1002 (synopsis "Basic local alignment search tool")
1003 (description
1004 "BLAST is a popular method of performing a DNA or protein sequence
1005similarity search, using heuristics to produce results quickly. It also
1006calculates an “expect value” that estimates how many matches would have
1007occurred at a given score by chance, which can aid a user in judging how much
1008confidence to have in an alignment.")
1009 ;; Most of the sources are in the public domain, with the following
1010 ;; exceptions:
1011 ;; * Expat:
1012 ;; * ./c++/include/util/bitset/
1013 ;; * ./c++/src/html/ncbi_menu*.js
1014 ;; * Boost license:
1015 ;; * ./c++/include/util/impl/floating_point_comparison.hpp
1016 ;; * LGPL 2+:
1017 ;; * ./c++/include/dbapi/driver/odbc/unix_odbc/
1018 ;; * ASL 2.0:
1019 ;; * ./c++/src/corelib/teamcity_*
1020 (license (list license:public-domain
1021 license:expat
1022 license:boost1.0
1023 license:lgpl2.0+
1024 license:asl2.0))))
1025
6c2b26e2
RW		 1026(define-public bless
1027 (package
1028 (name "bless")
1029 (version "1p02")
1030 (source (origin
1031 (method url-fetch)
1032 (uri (string-append "mirror://sourceforge/bless-ec/bless.v"
1033 version ".tgz"))
1034 (sha256
1035 (base32
4d75e03a
RW		 1036 "0rm0gw2s18dqwzzpl3c2x1z05ni2v0xz5dmfk3d33j6g4cgrlrdd"))
1037 (modules '((guix build utils)))
6c2b26e2
RW		 1038 (snippet
1039 `(begin
1040 ;; Remove bundled boost, pigz, zlib, and .git directory
953c1223
RW		 1041 ;; FIXME: also remove bundled sources for murmurhash3 and
1042 ;; kmc once packaged.
6c2b26e2
RW		 1043 (delete-file-recursively "boost")
1044 (delete-file-recursively "pigz")
953c1223 1045 (delete-file-recursively "google-sparsehash")
6c2b26e2
RW		 1046 (delete-file-recursively "zlib")
1047 (delete-file-recursively ".git")
1048 #t))))
1049 (build-system gnu-build-system)
1050 (arguments
1051 '(#:tests? #f ;no "check" target
1052 #:make-flags
1053 (list (string-append "ZLIB="
1054 (assoc-ref %build-inputs "zlib")
1055 "/lib/libz.a")
1056 (string-append "LDFLAGS="
1057 (string-join '("-lboost_filesystem"
1058 "-lboost_system"
1059 "-lboost_iostreams"
1060 "-lz"
1061 "-fopenmp"
1062 "-std=c++11"))))
1063 #:phases
1064 (modify-phases %standard-phases
1065 (add-after 'unpack 'do-not-build-bundled-pigz
1066 (lambda* (#:key inputs outputs #:allow-other-keys)
1067 (substitute* "Makefile"
1068 (("cd pigz/pigz-2.3.3; make") ""))
1069 #t))
1070 (add-after 'unpack 'patch-paths-to-executables
1071 (lambda* (#:key inputs outputs #:allow-other-keys)
1072 (substitute* "parse_args.cpp"
1073 (("kmc_binary = .*")
1074 (string-append "kmc_binary = \""
1075 (assoc-ref outputs "out")
1076 "/bin/kmc\";"))
1077 (("pigz_binary = .*")
1078 (string-append "pigz_binary = \""
1079 (assoc-ref inputs "pigz")
1080 "/bin/pigz\";")))
1081 #t))
1082 (replace 'install
1083 (lambda* (#:key outputs #:allow-other-keys)
1084 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
1085 (for-each (lambda (file)
1086 (install-file file bin))
1087 '("bless" "kmc/bin/kmc"))
1088 #t)))
1089 (delete 'configure))))
1090 (native-inputs
1091 `(("perl" ,perl)))
1092 (inputs
1093 `(("openmpi" ,openmpi)
1094 ("boost" ,boost)
953c1223 1095 ("sparsehash" ,sparsehash)
6c2b26e2
RW		 1096 ("pigz" ,pigz)
1097 ("zlib" ,zlib)))
9641a899 1098 (supported-systems '("x86_64-linux"))
4d75e03a 1099 (home-page "http://sourceforge.net/p/bless-ec/wiki/Home/")
6c2b26e2
RW		 1100 (synopsis "Bloom-filter-based error correction tool for NGS reads")
1101 (description
1102 "@dfn{Bloom-filter-based error correction solution for high-throughput
1103sequencing reads} (BLESS) uses a single minimum-sized bloom filter is a
1104correction tool for genomic reads produced by @dfn{Next-generation
1105sequencing} (NGS). BLESS produces accurate correction results with much less
1106memory compared with previous solutions and is also able to tolerate a higher
1107false-positive rate. BLESS can extend reads like DNA assemblers to correct
1108errors at the end of reads.")
1109 (license license:gpl3+)))
1110
2c7ee167
RW		 1111(define-public bowtie
1112 (package
1113 (name "bowtie")
2642231b 1114 (version "2.2.9")
2c7ee167
RW		 1115 (source (origin
1116 (method url-fetch)
1117 (uri (string-append "https://github.com/BenLangmead/bowtie2/archive/v"
1118 version ".tar.gz"))
f586c877 1119 (file-name (string-append name "-" version ".tar.gz"))
2c7ee167
RW		 1120 (sha256
1121 (base32
2642231b 1122 "1vp5db8i7is57iwjybcdg18f5ivyzlj5g1ix1nlvxainzivhz55g"))
2c7ee167
RW		 1123 (modules '((guix build utils)))
1124 (snippet
1125 '(substitute* "Makefile"
2c7ee167
RW		 1126 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
1127 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
0047d26a 1128 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\"")))))
2c7ee167
RW		 1129 (build-system gnu-build-system)
1130 (inputs `(("perl" ,perl)
1131 ("perl-clone" ,perl-clone)
1132 ("perl-test-deep" ,perl-test-deep)
1133 ("perl-test-simple" ,perl-test-simple)
0047d26a
RW		 1134 ("python" ,python-2)
1135 ("tbb" ,tbb)))
2c7ee167 1136 (arguments
0047d26a
RW		 1137 '(#:make-flags
1138 (list "allall"
1139 "WITH_TBB=1"
1140 (string-append "prefix=" (assoc-ref %outputs "out")))
2c7ee167
RW		 1141 #:phases
1142 (alist-delete
1143 'configure
1144 (alist-replace
0047d26a 1145 'check
2c7ee167 1146 (lambda* (#:key outputs #:allow-other-keys)
0047d26a
RW		 1147 (system* "perl"
1148 "scripts/test/simple_tests.pl"
1149 "--bowtie2=./bowtie2"
1150 "--bowtie2-build=./bowtie2-build"))
1151 %standard-phases))))
2c7ee167
RW		 1152 (home-page "http://bowtie-bio.sourceforge.net/bowtie2/index.shtml")
1153 (synopsis "Fast and sensitive nucleotide sequence read aligner")
1154 (description
1155 "Bowtie 2 is a fast and memory-efficient tool for aligning sequencing
1156reads to long reference sequences. It is particularly good at aligning reads
1157of about 50 up to 100s or 1,000s of characters, and particularly good at
1158aligning to relatively long (e.g. mammalian) genomes. Bowtie 2 indexes the
1159genome with an FM Index to keep its memory footprint small: for the human
1160genome, its memory footprint is typically around 3.2 GB. Bowtie 2 supports
1161gapped, local, and paired-end alignment modes.")
241e1221 1162 (supported-systems '("x86_64-linux"))
2c7ee167
RW		 1163 (license license:gpl3+)))
1164
94ce537e
RW		 1165(define-public tophat
1166 (package
1167 (name "tophat")
1168 (version "2.1.0")
1169 (source (origin
1170 (method url-fetch)
1171 (uri (string-append
1172 "http://ccb.jhu.edu/software/tophat/downloads/tophat-"
1173 version ".tar.gz"))
1174 (sha256
1175 (base32
1176 "168zlzykq622zbgkh90a90f1bdgsxkscq2zxzbj8brq80hbjpyp7"))
fc1adab1 1177 (patches (search-patches "tophat-build-with-later-seqan.patch"))
94ce537e
RW		 1178 (modules '((guix build utils)))
1179 (snippet
1180 '(begin
1181 ;; Remove bundled SeqAn and samtools
1182 (delete-file-recursively "src/SeqAn-1.3")
1183 (delete-file-recursively "src/samtools-0.1.18")
1184 #t))))
1185 (build-system gnu-build-system)
1186 (arguments
1187 '(#:parallel-build? #f ; not supported
1188 #:phases
1189 (modify-phases %standard-phases
1190 (add-after 'unpack 'use-system-samtools
1191 (lambda* (#:key inputs #:allow-other-keys)
1192 (substitute* "src/Makefile.in"
1193 (("(noinst_LIBRARIES = )\\$\\(SAMLIB\\)" _ prefix) prefix)
1194 (("\\$\\(SAMPROG\\): \\$\\(SAMLIB\\)") "")
1195 (("SAMPROG = samtools_0\\.1\\.18") "")
1196 (("\\$\\(samtools_0_1_18_SOURCES\\)") "")
1197 (("am__EXEEXT_1 = samtools_0\\.1\\.18\\$\\(EXEEXT\\)") ""))
1198 (substitute* '("src/common.cpp"
1199 "src/tophat.py")
1200 (("samtools_0.1.18") (which "samtools")))
1201 (substitute* '("src/common.h"
1202 "src/bam2fastx.cpp")
1203 (("#include \"bam.h\"") "#include <samtools/bam.h>")
1204 (("#include \"sam.h\"") "#include <samtools/sam.h>"))
1205 (substitute* '("src/bwt_map.h"
1206 "src/map2gtf.h"
1207 "src/align_status.h")
1208 (("#include <bam.h>") "#include <samtools/bam.h>")
1209 (("#include <sam.h>") "#include <samtools/sam.h>"))
1210 #t)))))
1211 (inputs
1212 `(("boost" ,boost)
1213 ("bowtie" ,bowtie)
1214 ("samtools" ,samtools-0.1)
1215 ("ncurses" ,ncurses)
1216 ("python" ,python-2)
1217 ("perl" ,perl)
1218 ("zlib" ,zlib)
1219 ("seqan" ,seqan)))
1220 (home-page "http://ccb.jhu.edu/software/tophat/index.shtml")
1221 (synopsis "Spliced read mapper for RNA-Seq data")
1222 (description
1223 "TopHat is a fast splice junction mapper for nucleotide sequence
1224reads produced by the RNA-Seq method. It aligns RNA-Seq reads to
1225mammalian-sized genomes using the ultra high-throughput short read
1226aligner Bowtie, and then analyzes the mapping results to identify
1227splice junctions between exons.")
1228 ;; TopHat is released under the Boost Software License, Version 1.0
1229 ;; See https://github.com/infphilo/tophat/issues/11#issuecomment-121589893
1230 (license license:boost1.0)))
1231
9a8336d8
RW		 1232(define-public bwa
1233 (package
1234 (name "bwa")
1235 (version "0.7.12")
1236 (source (origin
1237 (method url-fetch)
1238 (uri (string-append "mirror://sourceforge/bio-bwa/bwa-"
1239 version ".tar.bz2"))
1240 (sha256
1241 (base32
1242 "1330dpqncv0px3pbhjzz1gwgg39kkcv2r9qp2xs0sixf8z8wl7bh"))))
1243 (build-system gnu-build-system)
1244 (arguments
1245 '(#:tests? #f ;no "check" target
1246 #:phases
1247 (alist-replace
1248 'install
1249 (lambda* (#:key outputs #:allow-other-keys)
1250 (let ((bin (string-append
1251 (assoc-ref outputs "out") "/bin"))
1252 (doc (string-append
1253 (assoc-ref outputs "out") "/share/doc/bwa"))
1254 (man (string-append
1255 (assoc-ref outputs "out") "/share/man/man1")))
96c46210
LC		 1256 (install-file "bwa" bin)
1257 (install-file "README.md" doc)
1258 (install-file "bwa.1" man)))
9a8336d8
RW		 1259 ;; no "configure" script
1260 (alist-delete 'configure %standard-phases))))
1261 (inputs `(("zlib" ,zlib)))
db94f8c7
RW		 1262 ;; Non-portable SSE instructions are used so building fails on platforms
1263 ;; other than x86_64.
1264 (supported-systems '("x86_64-linux"))
9a8336d8
RW		 1265 (home-page "http://bio-bwa.sourceforge.net/")
1266 (synopsis "Burrows-Wheeler sequence aligner")
1267 (description
1268 "BWA is a software package for mapping low-divergent sequences against a
1269large reference genome, such as the human genome. It consists of three
1270algorithms: BWA-backtrack, BWA-SW and BWA-MEM. The first algorithm is
1271designed for Illumina sequence reads up to 100bp, while the rest two for
1272longer sequences ranged from 70bp to 1Mbp. BWA-MEM and BWA-SW share similar
1273features such as long-read support and split alignment, but BWA-MEM, which is
1274the latest, is generally recommended for high-quality queries as it is faster
1275and more accurate. BWA-MEM also has better performance than BWA-backtrack for
127670-100bp Illumina reads.")
1277 (license license:gpl3+)))
1278
d29150b5
RW		 1279(define-public bwa-pssm
1280 (package (inherit bwa)
1281 (name "bwa-pssm")
1282 (version "0.5.11")
1283 (source (origin
1284 (method url-fetch)
1285 (uri (string-append "https://github.com/pkerpedjiev/bwa-pssm/"
1286 "archive/" version ".tar.gz"))
1287 (file-name (string-append name "-" version ".tar.gz"))
1288 (sha256
1289 (base32
1290 "02p7mpbs4mlxmn84g2x4ghak638vbj4lqix2ipx5g84pz9bhdavg"))))
1291 (build-system gnu-build-system)
1292 (inputs
1293 `(("gdsl" ,gdsl)
1294 ("zlib" ,zlib)
1295 ("perl" ,perl)))
1296 (home-page "http://bwa-pssm.binf.ku.dk/")
1297 (synopsis "Burrows-Wheeler transform-based probabilistic short read mapper")
1298 (description
1299 "BWA-PSSM is a probabilistic short genomic sequence read aligner based on
1300the use of @dfn{position specific scoring matrices} (PSSM). Like many of the
1301existing aligners it is fast and sensitive. Unlike most other aligners,
1302however, it is also adaptible in the sense that one can direct the alignment
1303based on known biases within the data set. It is coded as a modification of
1304the original BWA alignment program and shares the genome index structure as
1305well as many of the command line options.")
1306 (license license:gpl3+)))
1307
ad641d53
RW		 1308(define-public python2-bx-python
1309 (package
1310 (name "python2-bx-python")
1311 (version "0.7.2")
1312 (source (origin
1313 (method url-fetch)
1314 (uri (string-append
1315 "https://pypi.python.org/packages/source/b/bx-python/bx-python-"
1316 version ".tar.gz"))
1317 (sha256
1318 (base32
1319 "0ld49idhc5zjdvbhvjq1a2qmpjj7h5v58rqr25dzmfq7g34b50xh"))
1320 (modules '((guix build utils)))
1321 (snippet
1322 '(substitute* "setup.py"
1323 ;; remove dependency on outdated "distribute" module
1324 (("^from distribute_setup import use_setuptools") "")
1325 (("^use_setuptools\\(\\)") "")))))
1326 (build-system python-build-system)
1327 (arguments
1328 `(#:tests? #f ;tests fail because test data are not included
1329 #:python ,python-2))
1330 (inputs
1331 `(("python-numpy" ,python2-numpy)
1332 ("zlib" ,zlib)))
1333 (native-inputs
f3b98f4f 1334 `(("python-nose" ,python2-nose)))
ad641d53
RW		 1335 (home-page "http://bitbucket.org/james_taylor/bx-python/")
1336 (synopsis "Tools for manipulating biological data")
1337 (description
1338 "bx-python provides tools for manipulating biological data, particularly
1339multiple sequence alignments.")
1340 (license license:expat)))
1341
55a9a8c2
RW		 1342(define-public python-pysam
1343 (package
1344 (name "python-pysam")
fd49eb21 1345 (version "0.9.1.4")
d454640c
RW		 1346 (source (origin
1347 (method url-fetch)
f536dce5
MB		 1348 ;; Test data is missing on PyPi.
1349 (uri (string-append
1350 "https://github.com/pysam-developers/pysam/archive/v"
1351 version ".tar.gz"))
1352 (file-name (string-append name "-" version ".tar.gz"))
d454640c
RW		 1353 (sha256
1354 (base32
f536dce5 1355 "0y41ssbg6nvn2jgcbnrvkzblpjcwszaiv1rgyd8dwzjkrbfsgsmc"))
dff26b23
MB		 1356 (modules '((guix build utils)))
1357 (snippet
1358 ;; Drop bundled htslib. TODO: Also remove samtools and bcftools.
1359 '(delete-file-recursively "htslib"))))
55a9a8c2
RW		 1360 (build-system python-build-system)
1361 (arguments
f536dce5 1362 `(#:phases
397d463a
MB		 1363 (modify-phases %standard-phases
1364 (add-before 'build 'set-flags
dff26b23
MB		 1365 (lambda* (#:key inputs #:allow-other-keys)
1366 (setenv "HTSLIB_MODE" "external")
1367 (setenv "HTSLIB_LIBRARY_DIR"
1368 (string-append (assoc-ref inputs "htslib") "/lib"))
1369 (setenv "HTSLIB_INCLUDE_DIR"
1370 (string-append (assoc-ref inputs "htslib") "/include"))
397d463a
MB		 1371 (setenv "LDFLAGS" "-lncurses")
1372 (setenv "CFLAGS" "-D_CURSES_LIB=1")
f536dce5
MB		 1373 #t))
1374 (delete 'check)
1375 (add-after 'install 'check
1376 (lambda* (#:key inputs outputs #:allow-other-keys)
1377 (setenv "PYTHONPATH"
1378 (string-append
1379 (getenv "PYTHONPATH")
1380 ":" (assoc-ref outputs "out")
1381 "/lib/python"
1382 (string-take (string-take-right
1383 (assoc-ref inputs "python") 5) 3)
1384 "/site-packages"))
1385 ;; Step out of source dir so python does not import from CWD.
1386 (chdir "tests")
1387 (setenv "HOME" "/tmp")
1388 (and (zero? (system* "make" "-C" "pysam_data"))
1389 (zero? (system* "make" "-C" "cbcf_data"))
1390 (zero? (system* "nosetests" "-v"))))))))
dff26b23
MB		 1391 (propagated-inputs
1392 `(("htslib" ,htslib))) ; Included from installed header files.
55a9a8c2 1393 (inputs
649e9b3b 1394 `(("ncurses" ,ncurses)
55a9a8c2 1395 ("zlib" ,zlib)))
649e9b3b
RW		 1396 (native-inputs
1397 `(("python-cython" ,python-cython)
f536dce5
MB		 1398 ;; Dependencies below are are for tests only.
1399 ("samtools" ,samtools)
1400 ("bcftools" ,bcftools)
1401 ("python-nose" ,python-nose)))
55a9a8c2
RW		 1402 (home-page "https://github.com/pysam-developers/pysam")
1403 (synopsis "Python bindings to the SAMtools C API")
1404 (description
1405 "Pysam is a Python module for reading and manipulating files in the
1406SAM/BAM format. Pysam is a lightweight wrapper of the SAMtools C API. It
1407also includes an interface for tabix.")
1408 (license license:expat)))
1409
1410(define-public python2-pysam
1411 (package-with-python2 python-pysam))
1412
4db9433a
RW		 1413(define-public python-twobitreader
1414 (package
1415 (name "python-twobitreader")
044ac8d2 1416 (version "3.1.4")
4db9433a
RW		 1417 (source (origin
1418 (method url-fetch)
1419 (uri (pypi-uri "twobitreader" version))
1420 (sha256
1421 (base32
044ac8d2 1422 "1q8wnj2kga9nz1lwc4w7qv52smfm536hp6mc8w6s53lhyj0mpi22"))))
4db9433a 1423 (build-system python-build-system)
900fb8d0
LF		 1424 (arguments
1425 '(;; Tests are not distributed in the PyPi release.
1426 ;; TODO Try building from the Git repo or asking the upstream maintainer
1427 ;; to distribute the tests on PyPi.
1428 #:tests? #f))
4db9433a
RW		 1429 (native-inputs
1430 `(("python-sphinx" ,python-sphinx)))
1431 (home-page "https://github.com/benjschiller/twobitreader")
1432 (synopsis "Python library for reading .2bit files")
1433 (description
1434 "twobitreader is a Python library for reading .2bit files as used by the
1435UCSC genome browser.")
1436 (license license:artistic2.0)))
1437
1438(define-public python2-twobitreader
5c31f4aa 1439 (package-with-python2 python-twobitreader))
4db9433a 1440
f94bf198
RW		 1441(define-public python-plastid
1442 (package
1443 (name "python-plastid")
99caa6f7 1444 (version "0.4.6")
f94bf198
RW		 1445 (source (origin
1446 (method url-fetch)
1447 (uri (pypi-uri "plastid" version))
1448 (sha256
1449 (base32
99caa6f7 1450 "1sqkz5d3b9kf688mp7k771c87ins42j7j0whmkb49cb3fsg8s8lj"))))
f94bf198
RW		 1451 (build-system python-build-system)
1452 (arguments
1453 ;; Some test files are not included.
1454 `(#:tests? #f))
1455 (propagated-inputs
1456 `(("python-numpy" ,python-numpy)
1457 ("python-scipy" ,python-scipy)
1458 ("python-pandas" ,python-pandas)
1459 ("python-pysam" ,python-pysam)
1460 ("python-matplotlib" ,python-matplotlib)
1461 ("python-biopython" ,python-biopython)
99caa6f7
BW		 1462 ("python-twobitreader" ,python-twobitreader)
1463 ("python-termcolor" ,python-termcolor)))
f94bf198
RW		 1464 (native-inputs
1465 `(("python-cython" ,python-cython)
1466 ("python-nose" ,python-nose)))
1467 (home-page "https://github.com/joshuagryphon/plastid")
1468 (synopsis "Python library for genomic analysis")
1469 (description
1470 "plastid is a Python library for genomic analysis – in particular,
1471high-throughput sequencing data – with an emphasis on simplicity.")
1472 (license license:bsd-3)))
1473
1474(define-public python2-plastid
5c31f4aa 1475 (package-with-python2 python-plastid))
f94bf198 1476
6c1305f9
RW		 1477(define-public cd-hit
1478 (package
1479 (name "cd-hit")
1480 (version "4.6.5")
1481 (source (origin
1482 (method url-fetch)
1483 (uri (string-append "https://github.com/weizhongli/cdhit"
1484 "/releases/download/V" version
1485 "/cd-hit-v" version "-2016-0304.tar.gz"))
1486 (sha256
1487 (base32
1488 "15db0hq38yyifwqx9b6l34z14jcq576dmjavhj8a426c18lvnhp3"))))
1489 (build-system gnu-build-system)
1490 (arguments
1491 `(#:tests? #f ; there are no tests
1492 #:make-flags
1493 ;; Executables are copied directly to the PREFIX.
1494 (list (string-append "PREFIX=" (assoc-ref %outputs "out") "/bin"))
1495 #:phases
1496 (modify-phases %standard-phases
1497 ;; No "configure" script
1498 (delete 'configure)
1499 ;; Remove sources of non-determinism
1500 (add-after 'unpack 'be-timeless
1501 (lambda _
1502 (substitute* "cdhit-utility.c++"
1503 ((" \\(built on \" __DATE__ \"\\)") ""))
1504 (substitute* "cdhit-common.c++"
1505 (("__DATE__") "\"0\"")
1506 (("\", %s, \" __TIME__ \"\\\\n\", date") ""))
1507 #t))
1508 ;; The "install" target does not create the target directory
1509 (add-before 'install 'create-target-dir
1510 (lambda* (#:key outputs #:allow-other-keys)
1511 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
1512 #t)))))
1513 (inputs
1514 `(("perl" ,perl)))
1515 (home-page "http://weizhongli-lab.org/cd-hit/")
1516 (synopsis "Cluster and compare protein or nucleotide sequences")
1517 (description
1518 "CD-HIT is a program for clustering and comparing protein or nucleotide
1519sequences. CD-HIT is designed to be fast and handle extremely large
1520databases.")
1521 ;; The manual says: "It can be copied under the GNU General Public License
1522 ;; version 2 (GPLv2)."
1523 (license license:gpl2)))
1524
810cff85
RW		 1525(define-public clipper
1526 (package
1527 (name "clipper")
433530a5 1528 (version "1.1")
810cff85
RW		 1529 (source (origin
1530 (method url-fetch)
1531 (uri (string-append
1532 "https://github.com/YeoLab/clipper/archive/"
1533 version ".tar.gz"))
9ab5ea44 1534 (file-name (string-append name "-" version ".tar.gz"))
810cff85
RW		 1535 (sha256
1536 (base32
433530a5 1537 "0pflmsvhbf8izbgwhbhj1i7349sw1f55qpqj8ljmapp16hb0p0qi"))
810cff85
RW		 1538 (modules '((guix build utils)))
1539 (snippet
433530a5
RW		 1540 '(begin
1541 ;; remove unnecessary setup dependency
1542 (substitute* "setup.py"
1543 (("setup_requires = .*") ""))
1544 (for-each delete-file
1545 '("clipper/src/peaks.so"
1546 "clipper/src/readsToWiggle.so"))
1547 (delete-file-recursively "dist/")
1548 #t))))
810cff85
RW		 1549 (build-system python-build-system)
1550 (arguments `(#:python ,python-2)) ; only Python 2 is supported
1551 (inputs
1552 `(("htseq" ,htseq)
1553 ("python-pybedtools" ,python2-pybedtools)
1554 ("python-cython" ,python2-cython)
1555 ("python-scikit-learn" ,python2-scikit-learn)
1556 ("python-matplotlib" ,python2-matplotlib)
433530a5 1557 ("python-pandas" ,python2-pandas)
810cff85
RW		 1558 ("python-pysam" ,python2-pysam)
1559 ("python-numpy" ,python2-numpy)
1560 ("python-scipy" ,python2-scipy)))
1561 (native-inputs
f3b98f4f
HG		 1562 `(("python-mock" ,python2-mock) ; for tests
1563 ("python-pytz" ,python2-pytz))) ; for tests
810cff85
RW		 1564 (home-page "https://github.com/YeoLab/clipper")
1565 (synopsis "CLIP peak enrichment recognition")
1566 (description
1567 "CLIPper is a tool to define peaks in CLIP-seq datasets.")
1568 (license license:gpl2)))
1569
6a35566d
RS		 1570(define-public codingquarry
1571 (package
1572 (name "codingquarry")
1573 (version "2.0")
1574 (source (origin
1575 (method url-fetch)
1576 (uri (string-append
1577 "mirror://sourceforge/codingquarry/CodingQuarry_v"
1578 version ".tar.gz"))
1579 (sha256
1580 (base32
1581 "0115hkjflsnfzn36xppwf9h9avfxlavr43djqmshkkzbgjzsz60i"))))
1582 (build-system gnu-build-system)
1583 (arguments
1584 '(#:tests? #f ; no "check" target
1585 #:phases
1586 (modify-phases %standard-phases
1587 (delete 'configure)
1588 (replace 'install
1589 (lambda* (#:key outputs #:allow-other-keys)
1590 (let* ((out (assoc-ref outputs "out"))
1591 (bin (string-append out "/bin"))
1592 (doc (string-append out "/share/doc/codingquarry")))
1593 (install-file "INSTRUCTIONS.pdf" doc)
1594 (copy-recursively "QuarryFiles"
1595 (string-append out "/QuarryFiles"))
1596 (install-file "CodingQuarry" bin)
1597 (install-file "CufflinksGTF_to_CodingQuarryGFF3.py" bin)))))))
1598 (inputs `(("openmpi" ,openmpi)))
1599 (native-search-paths
1600 (list (search-path-specification
1601 (variable "QUARRY_PATH")
1602 (files '("QuarryFiles")))))
1603 (native-inputs `(("python" ,python-2))) ; Only Python 2 is supported
1604 (synopsis "Fungal gene predictor")
1605 (description "CodingQuarry is a highly accurate, self-training GHMM fungal
1606gene predictor designed to work with assembled, aligned RNA-seq transcripts.")
1607 (home-page "https://sourceforge.net/projects/codingquarry/")
1608 (license license:gpl3+)))
1609
36742f43
RW		 1610(define-public couger
1611 (package
1612 (name "couger")
1613 (version "1.8.2")
1614 (source (origin
1615 (method url-fetch)
1616 (uri (string-append
1617 "http://couger.oit.duke.edu/static/assets/COUGER"
1618 version ".zip"))
1619 (sha256
1620 (base32
1621 "04p2b14nmhzxw5h72mpzdhalv21bx4w9b87z0wpw0xzxpysyncmq"))))
1622 (build-system gnu-build-system)
1623 (arguments
1624 `(#:tests? #f
1625 #:phases
1626 (modify-phases %standard-phases
1627 (delete 'configure)
1628 (delete 'build)
1629 (replace
1630 'install
1631 (lambda* (#:key outputs #:allow-other-keys)
1632 (let ((out (assoc-ref outputs "out")))
1633 (copy-recursively "src" (string-append out "/src"))
1634 (mkdir (string-append out "/bin"))
1635 ;; Add "src" directory to module lookup path.
1636 (substitute* "couger"
1637 (("from argparse")
1638 (string-append "import sys\nsys.path.append(\""
1639 out "\")\nfrom argparse")))
1640 (copy-file "couger" (string-append out "/bin/couger")))
1641 #t))
1642 (add-after
1643 'install 'wrap-program
1644 (lambda* (#:key inputs outputs #:allow-other-keys)
1645 ;; Make sure 'couger' runs with the correct PYTHONPATH.
1646 (let* ((out (assoc-ref outputs "out"))
1647 (path (getenv "PYTHONPATH")))
1648 (wrap-program (string-append out "/bin/couger")
1649 `("PYTHONPATH" ":" prefix (,path))))
1650 #t)))))
1651 (inputs
1652 `(("python" ,python-2)
1653 ("python2-pillow" ,python2-pillow)
1654 ("python2-numpy" ,python2-numpy)
1655 ("python2-scipy" ,python2-scipy)
1656 ("python2-matplotlib" ,python2-matplotlib)))
1657 (propagated-inputs
1658 `(("r" ,r)
1659 ("libsvm" ,libsvm)
1660 ("randomjungle" ,randomjungle)))
1661 (native-inputs
1662 `(("unzip" ,unzip)))
1663 (home-page "http://couger.oit.duke.edu")
1664 (synopsis "Identify co-factors in sets of genomic regions")
1665 (description
1666 "COUGER can be applied to any two sets of genomic regions bound by
1667paralogous TFs (e.g., regions derived from ChIP-seq experiments) to identify
1668putative co-factors that provide specificity to each TF. The framework
1669determines the genomic targets uniquely-bound by each TF, and identifies a
1670small set of co-factors that best explain the in vivo binding differences
1671between the two TFs.
1672
1673COUGER uses classification algorithms (support vector machines and random
1674forests) with features that reflect the DNA binding specificities of putative
1675co-factors. The features are generated either from high-throughput TF-DNA
1676binding data (from protein binding microarray experiments), or from large
1677collections of DNA motifs.")
1678 (license license:gpl3+)))
1679
bfe3c685
RW		 1680(define-public clustal-omega
1681 (package
1682 (name "clustal-omega")
1683 (version "1.2.1")
1684 (source (origin
1685 (method url-fetch)
1686 (uri (string-append
1687 "http://www.clustal.org/omega/clustal-omega-"
1688 version ".tar.gz"))
1689 (sha256
1690 (base32
1691 "02ibkx0m0iwz8nscg998bh41gg251y56cgh86bvyrii5m8kjgwqf"))))
1692 (build-system gnu-build-system)
1693 (inputs
1694 `(("argtable" ,argtable)))
1695 (home-page "http://www.clustal.org/omega/")
1696 (synopsis "Multiple sequence aligner for protein and DNA/RNA")
1697 (description
1698 "Clustal-Omega is a general purpose multiple sequence alignment (MSA)
1699program for protein and DNA/RNA. It produces high quality MSAs and is capable
1700of handling data-sets of hundreds of thousands of sequences in reasonable
1701time.")
1702 (license license:gpl2+)))
1703
191c7101
RW		 1704(define-public crossmap
1705 (package
1706 (name "crossmap")
61d5fd03 1707 (version "0.2.1")
191c7101
RW		 1708 (source (origin
1709 (method url-fetch)
1710 (uri (string-append "mirror://sourceforge/crossmap/CrossMap-"
1711 version ".tar.gz"))
1712 (sha256
1713 (base32
61d5fd03
RW		 1714 "07y179f63d7qnzdvkqcziwk9bs3k4zhp81q392fp1hwszjdvy22f"))
1715 ;; This patch has been sent upstream already and is available
1716 ;; for download from Sourceforge, but it has not been merged.
fc1adab1 1717 (patches (search-patches "crossmap-allow-system-pysam.patch"))
191c7101
RW		 1718 (modules '((guix build utils)))
1719 ;; remove bundled copy of pysam
1720 (snippet
1721 '(delete-file-recursively "lib/pysam"))))
1722 (build-system python-build-system)
1723 (arguments
1724 `(#:python ,python-2
1725 #:phases
1726 (alist-cons-after
1727 'unpack 'set-env
1728 (lambda _ (setenv "CROSSMAP_USE_SYSTEM_PYSAM" "1"))
1729 %standard-phases)))
1730 (inputs
1731 `(("python-numpy" ,python2-numpy)
1732 ("python-pysam" ,python2-pysam)
1733 ("zlib" ,zlib)))
1734 (native-inputs
1735 `(("python-cython" ,python2-cython)
f3b98f4f 1736 ("python-nose" ,python2-nose)))
191c7101
RW		 1737 (home-page "http://crossmap.sourceforge.net/")
1738 (synopsis "Convert genome coordinates between assemblies")
1739 (description
1740 "CrossMap is a program for conversion of genome coordinates or annotation
1741files between different genome assemblies. It supports most commonly used
1742file formats including SAM/BAM, Wiggle/BigWig, BED, GFF/GTF, VCF.")
1743 (license license:gpl2+)))
1744
3a40a92c
RW		 1745(define-public cufflinks
1746 (package
1747 (name "cufflinks")
1748 (version "2.2.1")
1749 (source (origin
1750 (method url-fetch)
1751 (uri (string-append "http://cole-trapnell-lab.github.io/"
1752 "cufflinks/assets/downloads/cufflinks-"
1753 version ".tar.gz"))
1754 (sha256
1755 (base32
1756 "1bnm10p8m7zq4qiipjhjqb24csiqdm1pwc8c795z253r2xk6ncg8"))))
1757 (build-system gnu-build-system)
1758 (arguments
1759 `(#:make-flags
1760 (list
1761 ;; The includes for "eigen" are located in a subdirectory.
1762 (string-append "EIGEN_CPPFLAGS="
1763 "-I" (assoc-ref %build-inputs "eigen")
1764 "/include/eigen3/")
1765 ;; Cufflinks must be linked with various boost libraries.
1766 (string-append "LDFLAGS="
1767 (string-join '("-lboost_system"
1768 "-lboost_serialization"
1769 "-lboost_thread"))))
1770 #:phases
1771 (modify-phases %standard-phases
1772 (add-after 'unpack 'fix-search-for-bam
1773 (lambda _
1774 (substitute* '("ax_bam.m4"
1775 "configure"
1776 "src/hits.h")
1777 (("<bam/sam\\.h>") "<samtools/sam.h>")
1778 (("<bam/bam\\.h>") "<samtools/bam.h>")
1779 (("<bam/version\\.hpp>") "<samtools/version.h>"))
1780 #t)))
1781 #:configure-flags
1782 (list (string-append "--with-bam="
1783 (assoc-ref %build-inputs "samtools")))))
1784 (inputs
1785 `(("eigen" ,eigen)
1786 ("samtools" ,samtools-0.1)
1787 ("htslib" ,htslib)
1788 ("boost" ,boost)
1789 ("python" ,python-2)
1790 ("zlib" ,zlib)))
1791 (home-page "http://cole-trapnell-lab.github.io/cufflinks/")
1792 (synopsis "Transcriptome assembly and RNA-Seq expression analysis")
1793 (description
1794 "Cufflinks assembles RNA transcripts, estimates their abundances,
1795and tests for differential expression and regulation in RNA-Seq
1796samples. It accepts aligned RNA-Seq reads and assembles the
1797alignments into a parsimonious set of transcripts. Cufflinks then
1798estimates the relative abundances of these transcripts based on how
1799many reads support each one, taking into account biases in library
1800preparation protocols.")
1801 (license license:boost1.0)))
1802
8e913213
RW		 1803(define-public cutadapt
1804 (package
1805 (name "cutadapt")
1806 (version "1.8")
1807 (source (origin
1808 (method url-fetch)
1809 (uri (string-append
1810 "https://github.com/marcelm/cutadapt/archive/v"
1811 version ".tar.gz"))
1812 (file-name (string-append name "-" version ".tar.gz"))
1813 (sha256
1814 (base32
1815 "161bp87y6gd6r5bmvjpn2b1k942i3fizfpa139f0jn6jv1wcp5h5"))))
1816 (build-system python-build-system)
1817 (arguments
1818 ;; tests must be run after install
1819 `(#:phases (alist-cons-after
1820 'install 'check
1821 (lambda* (#:key inputs outputs #:allow-other-keys)
1822 (setenv "PYTHONPATH"
1823 (string-append
1824 (getenv "PYTHONPATH")
1825 ":" (assoc-ref outputs "out")
1826 "/lib/python"
1827 (string-take (string-take-right
1828 (assoc-ref inputs "python") 5) 3)
1829 "/site-packages"))
1830 (zero? (system* "nosetests" "-P" "tests")))
1831 (alist-delete 'check %standard-phases))))
1832 (native-inputs
1833 `(("python-cython" ,python-cython)
f3b98f4f 1834 ("python-nose" ,python-nose)))
8e913213
RW		 1835 (home-page "https://code.google.com/p/cutadapt/")
1836 (synopsis "Remove adapter sequences from nucleotide sequencing reads")
1837 (description
1838 "Cutadapt finds and removes adapter sequences, primers, poly-A tails and
1839other types of unwanted sequence from high-throughput sequencing reads.")
1840 (license license:expat)))
1841
1baee943
RW		 1842(define-public libbigwig
1843 (package
1844 (name "libbigwig")
1845 (version "0.1.4")
1846 (source (origin
1847 (method url-fetch)
1848 (uri (string-append "https://github.com/dpryan79/libBigWig/"
1849 "archive/" version ".tar.gz"))
1850 (file-name (string-append name "-" version ".tar.gz"))
1851 (sha256
1852 (base32
1853 "098rjh35pi4a9q83n8wiwvyzykjqj6l8q189p1xgfw4ghywdlvw1"))))
1854 (build-system gnu-build-system)
1855 (arguments
1856 `(#:test-target "test"
1857 #:make-flags
1858 (list "CC=gcc"
1859 (string-append "prefix=" (assoc-ref %outputs "out")))
1860 #:phases
1861 (modify-phases %standard-phases
1862 (delete 'configure)
1863 (add-before 'check 'disable-curl-test
1864 (lambda _
1865 (substitute* "Makefile"
1866 (("./test/testRemote.*") ""))
1867 #t))
1868 ;; This has been fixed with the upstream commit 4ff6959cd8a0, but
1869 ;; there has not yet been a release containing this change.
1870 (add-before 'install 'create-target-dirs
1871 (lambda* (#:key outputs #:allow-other-keys)
1872 (let ((out (assoc-ref outputs "out")))
1873 (mkdir-p (string-append out "/lib"))
1874 (mkdir-p (string-append out "/include"))
1875 #t))))))
1876 (inputs
1877 `(("zlib" ,zlib)
1878 ("curl" ,curl)))
1879 (native-inputs
1880 `(("doxygen" ,doxygen)))
1881 (home-page "https://github.com/dpryan79/libBigWig")
1882 (synopsis "C library for handling bigWig files")
1883 (description
1884 "This package provides a C library for parsing local and remote BigWig
1885files.")
1886 (license license:expat)))
1887
69e0e03c
RW		 1888(define-public python-pybigwig
1889 (package
1890 (name "python-pybigwig")
1891 (version "0.2.5")
1892 (source (origin
1893 (method url-fetch)
1894 (uri (pypi-uri "pyBigWig" version))
1895 (sha256
1896 (base32
1897 "0yrpdxg3y0sny25x4w22lv1k47jzccqjmg7j4bp0hywklvp0hg7d"))
1898 (modules '((guix build utils)))
1899 (snippet
1900 '(begin
1901 ;; Delete bundled libBigWig sources
1902 (delete-file-recursively "libBigWig")))))
1903 (build-system python-build-system)
1904 (arguments
1905 `(#:phases
1906 (modify-phases %standard-phases
1907 (add-after 'unpack 'link-with-libBigWig
1908 (lambda* (#:key inputs #:allow-other-keys)
1909 (substitute* "setup.py"
1910 (("libs=\\[") "libs=[\"BigWig\", "))
1911 #t)))))
1912 (inputs
1913 `(("libbigwig" ,libbigwig)
1914 ("zlib" ,zlib)
1915 ("curl" ,curl)))
1916 (home-page "https://github.com/dpryan79/pyBigWig")
1917 (synopsis "Access bigWig files in Python using libBigWig")
1918 (description
1919 "This package provides Python bindings to the libBigWig library for
1920accessing bigWig files.")
1921 (license license:expat)))
1922
1923(define-public python2-pybigwig
5c31f4aa 1924 (package-with-python2 python-pybigwig))
69e0e03c 1925
ec2a67de
BW		 1926(define-public python-dendropy
1927 (package
1928 (name "python-dendropy")
1929 (version "4.1.0")
1930 (source
1931 (origin
1932 (method url-fetch)
1933 (uri (pypi-uri "DendroPy" version))
1934 (sha256
1935 (base32
1936 "1jfz7gp18wph311w1yygbvjanb3n5mdqal439bb6myw41dwb5m63"))
1937 ;; There are two known test failures that will be fixed in the next
1938 ;; release after 4.1.0.
1939 ;; https://github.com/jeetsukumaran/DendroPy/issues/48
1940 (patches (search-patches
1941 "python-dendropy-exclude-failing-tests.patch"))))
1942 (build-system python-build-system)
1943 (home-page "http://packages.python.org/DendroPy/")
1944 (synopsis "Library for phylogenetics and phylogenetic computing")
1945 (description
1946 "DendroPy is a library for phylogenetics and phylogenetic computing: reading,
1947writing, simulation, processing and manipulation of phylogenetic
1948trees (phylogenies) and characters.")
1949 (license license:bsd-3)
1950 (properties `((python2-variant . ,(delay python2-dendropy))))))
1951
1952(define-public python2-dendropy
1953 (let ((base (package-with-python2 (strip-python2-variant python-dendropy))))
1954 (package
1955 (inherit base)
9602e3cc
BW		 1956 ;; Do not use same source as 'python-dendropy' because the patched
1957 ;; failing tests do not occur on Python 2.
1958 (source
1959 (origin
1960 (method url-fetch)
1961 (uri (pypi-uri "DendroPy" (package-version base)))
1962 (sha256
1963 (base32
1964 "1jfz7gp18wph311w1yygbvjanb3n5mdqal439bb6myw41dwb5m63"))))
1965 (arguments
1966 `(#:python ,python-2
1967 #:phases
1968 (modify-phases %standard-phases
1969 (replace 'check
1970 ;; There is currently a test failure that only happens on some
1971 ;; systems, and only using "setup.py test"
1972 (lambda _ (zero? (system* "nosetests")))))))
f3b98f4f 1973 (native-inputs `(("python2-nose" ,python2-nose)
ec2a67de
BW		 1974 ,@(package-native-inputs base))))))
1975
1976
1921b1de
RW		 1977(define-public deeptools
1978 (package
1979 (name "deeptools")
3acb8c85 1980 (version "2.1.1")
1921b1de
RW		 1981 (source (origin
1982 (method url-fetch)
3acb8c85
RW		 1983 (uri (string-append "https://github.com/fidelram/deepTools/"
1984 "archive/" version ".tar.gz"))
1921b1de
RW		 1985 (file-name (string-append name "-" version ".tar.gz"))
1986 (sha256
1987 (base32
3acb8c85 1988 "1nmfin0zjdby3vay3r4flvz94dr6qjhj41ax4yz3vx13j6wz8izd"))))
1921b1de
RW		 1989 (build-system python-build-system)
1990 (arguments
1991 `(#:python ,python-2))
14bda1ff 1992 (inputs
1921b1de
RW		 1993 `(("python-scipy" ,python2-scipy)
1994 ("python-numpy" ,python2-numpy)
3acb8c85 1995 ("python-numpydoc" ,python2-numpydoc)
1921b1de
RW		 1996 ("python-matplotlib" ,python2-matplotlib)
1997 ("python-bx-python" ,python2-bx-python)
3acb8c85
RW		 1998 ("python-pysam" ,python2-pysam)
1999 ("python-pybigwig" ,python2-pybigwig)))
1921b1de 2000 (native-inputs
f3b98f4f
HG		 2001 `(("python-mock" ,python2-mock) ;for tests
2002 ("python-pytz" ,python2-pytz))) ;for tests
1921b1de
RW		 2003 (home-page "https://github.com/fidelram/deepTools")
2004 (synopsis "Tools for normalizing and visualizing deep-sequencing data")
2005 (description
2006 "DeepTools addresses the challenge of handling the large amounts of data
2007that are now routinely generated from DNA sequencing centers. To do so,
2008deepTools contains useful modules to process the mapped reads data to create
2009coverage files in standard bedGraph and bigWig file formats. By doing so,
2010deepTools allows the creation of normalized coverage files or the comparison
2011between two files (for example, treatment and control). Finally, using such
2012normalized and standardized files, multiple visualizations can be created to
2013identify enrichments with functional annotations of the genome.")
2014 (license license:gpl3+)))
2015
684bf7c7
BW		 2016(define-public diamond
2017 (package
2018 (name "diamond")
8e5f8c98 2019 (version "0.8.27")
684bf7c7
BW		 2020 (source (origin
2021 (method url-fetch)
2022 (uri (string-append
2023 "https://github.com/bbuchfink/diamond/archive/v"
2024 version ".tar.gz"))
2025 (file-name (string-append name "-" version ".tar.gz"))
2026 (sha256
2027 (base32
8e5f8c98 2028 "0g0zdyfnri9v7nfbh8f7zqs4af1xydqkiw8m0cx4jc2ql4chpf6a"))))
122395f9 2029 (build-system cmake-build-system)
684bf7c7 2030 (arguments
7c544991
BW		 2031 '(#:tests? #f ; no "check" target
2032 #:phases
2033 (modify-phases %standard-phases
2034 (add-after 'unpack 'remove-native-compilation
2035 (lambda _
2036 (substitute* "CMakeLists.txt" (("-march=native") ""))
2037 #t)))))
684bf7c7 2038 (inputs
122395f9 2039 `(("zlib" ,zlib)))
684bf7c7
BW		 2040 (home-page "https://github.com/bbuchfink/diamond")
2041 (synopsis "Accelerated BLAST compatible local sequence aligner")
2042 (description
2043 "DIAMOND is a BLAST-compatible local aligner for mapping protein and
2044translated DNA query sequences against a protein reference database (BLASTP
2045and BLASTX alignment mode). The speedup over BLAST is up to 20,000 on short
2046reads at a typical sensitivity of 90-99% relative to BLAST depending on the
2047data and settings.")
d9c44e9c
BW		 2048 ;; diamond fails to build on other platforms
2049 ;; https://github.com/bbuchfink/diamond/issues/18
2050 (supported-systems '("x86_64-linux"))
684bf7c7
BW		 2051 (license (license:non-copyleft "file://src/COPYING"
2052 "See src/COPYING in the distribution."))))
2053
97b9da68
RW		 2054(define-public discrover
2055 (package
2056 (name "discrover")
2057 (version "1.6.0")
2058 (source
2059 (origin
2060 (method url-fetch)
2061 (uri (string-append "https://github.com/maaskola/discrover/archive/"
2062 version ".tar.gz"))
2063 (file-name (string-append name "-" version ".tar.gz"))
2064 (sha256
2065 (base32
2066 "0rah9ja4m0rl5mldd6vag9rwrivw1zrqxssfq8qx64m7961fp68k"))))
2067 (build-system cmake-build-system)
2068 (arguments `(#:tests? #f)) ; there are no tests
2069 (inputs
2070 `(("boost" ,boost)
2071 ("cairo" ,cairo)))
2072 (native-inputs
2073 `(("texlive" ,texlive)
2074 ("imagemagick" ,imagemagick)))
2075 (home-page "http://dorina.mdc-berlin.de/public/rajewsky/discrover/")
2076 (synopsis "Discover discriminative nucleotide sequence motifs")
2077 (description "Discrover is a motif discovery method to find binding sites
2078of nucleic acid binding proteins.")
2079 (license license:gpl3+)))
2080
6619f9c7
RW		 2081(define-public eigensoft
2082 (let ((revision "1")
2083 (commit "b14d1e202e21e532536ff8004f0419cd5e259dc7"))
2084 (package
2085 (name "eigensoft")
2086 (version (string-append "6.1.2-"
2087 revision "."
2088 (string-take commit 9)))
2089 (source
2090 (origin
2091 (method git-fetch)
2092 (uri (git-reference
2093 (url "https://github.com/DReichLab/EIG.git")
2094 (commit commit)))
2095 (file-name (string-append "eigensoft-" commit "-checkout"))
2096 (sha256
2097 (base32
2098 "0f5m6k2j5c16xc3xbywcs989xyc26ncy1zfzp9j9n55n9r4xcaiq"))
2099 (modules '((guix build utils)))
2100 ;; Remove pre-built binaries.
2101 (snippet '(begin
2102 (delete-file-recursively "bin")
2103 (mkdir "bin")
2104 #t))))
2105 (build-system gnu-build-system)
2106 (arguments
2107 `(#:tests? #f ; There are no tests.
2108 #:make-flags '("CC=gcc")
2109 #:phases
2110 (modify-phases %standard-phases
2111 ;; There is no configure phase, but the Makefile is in a
2112 ;; sub-directory.
2113 (replace 'configure
2114 (lambda _
2115 (chdir "src")
2116 ;; The link flags are incomplete.
2117 (substitute* "Makefile"
2118 (("-lgsl") "-lgsl -lm -llapack -llapacke -lpthread"))
2119 #t))
2120 ;; The provided install target only copies executables to
2121 ;; the "bin" directory in the build root.
2122 (add-after 'install 'actually-install
2123 (lambda* (#:key outputs #:allow-other-keys)
2124 (let* ((out (assoc-ref outputs "out"))
2125 (bin (string-append out "/bin")))
6619f9c7
RW		 2126 (for-each (lambda (file)
2127 (install-file file bin))
2128 (find-files "../bin" ".*"))
2129 #t))))))
2130 (inputs
2131 `(("gsl" ,gsl)
2132 ("lapack" ,lapack)
6619f9c7
RW		 2133 ("openblas" ,openblas)
2134 ("perl" ,perl)
2135 ("gfortran" ,gfortran "lib")))
2136 (home-page "https://github.com/DReichLab/EIG")
2137 (synopsis "Tools for population genetics")
2138 (description "The EIGENSOFT package provides tools for population
2139genetics and stratification correction. EIGENSOFT implements methods commonly
2140used in population genetics analyses such as PCA, computation of Tracy-Widom
2141statistics, and finding related individuals in structured populations. It
2142comes with a built-in plotting script and supports multiple file formats and
2143quantitative phenotypes.")
2144 ;; The license of the eigensoft tools is Expat, but since it's
2145 ;; linking with the GNU Scientific Library (GSL) the effective
2146 ;; license is the GPL.
2147 (license license:gpl3+))))
2148
365c8153
RW		 2149(define-public edirect
2150 (package
2151 (name "edirect")
83b84fa8 2152 (version "4.10")
365c8153
RW		 2153 (source (origin
2154 (method url-fetch)
83b84fa8
RW		 2155 (uri (string-append "ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/"
2156 "versions/2016-05-03/edirect.tar.gz"))
365c8153
RW		 2157 (sha256
2158 (base32
83b84fa8 2159 "15zsprak5yh8c1yrz4r1knmb5s8qcmdid4xdhkh3lqcv64l60hli"))))
365c8153
RW		 2160 (build-system perl-build-system)
2161 (arguments
2162 `(#:tests? #f ;no "check" target
2163 #:phases
2164 (modify-phases %standard-phases
2165 (delete 'configure)
2166 (delete 'build)
2167 (replace 'install
2168 (lambda* (#:key outputs #:allow-other-keys)
2169 (let ((target (string-append (assoc-ref outputs "out")
2170 "/bin")))
2171 (mkdir-p target)
2172 (copy-file "edirect.pl"
2173 (string-append target "/edirect.pl"))
2174 #t)))
2175 (add-after
2176 'install 'wrap-program
2177 (lambda* (#:key inputs outputs #:allow-other-keys)
2178 ;; Make sure 'edirect.pl' finds all perl inputs at runtime.
2179 (let* ((out (assoc-ref outputs "out"))
2180 (path (getenv "PERL5LIB")))
2181 (wrap-program (string-append out "/bin/edirect.pl")
2182 `("PERL5LIB" ":" prefix (,path)))))))))
2183 (inputs
2184 `(("perl-html-parser" ,perl-html-parser)
2185 ("perl-encode-locale" ,perl-encode-locale)
2186 ("perl-file-listing" ,perl-file-listing)
2187 ("perl-html-tagset" ,perl-html-tagset)
2188 ("perl-html-tree" ,perl-html-tree)
2189 ("perl-http-cookies" ,perl-http-cookies)
2190 ("perl-http-date" ,perl-http-date)
2191 ("perl-http-message" ,perl-http-message)
2192 ("perl-http-negotiate" ,perl-http-negotiate)
2193 ("perl-lwp-mediatypes" ,perl-lwp-mediatypes)
2194 ("perl-lwp-protocol-https" ,perl-lwp-protocol-https)
2195 ("perl-net-http" ,perl-net-http)
2196 ("perl-uri" ,perl-uri)
2197 ("perl-www-robotrules" ,perl-www-robotrules)
2198 ("perl" ,perl)))
3d51ec91 2199 (home-page "http://www.ncbi.nlm.nih.gov/books/NBK179288/")
365c8153
RW		 2200 (synopsis "Tools for accessing the NCBI's set of databases")
2201 (description
2202 "Entrez Direct (EDirect) is a method for accessing the National Center
2203for Biotechnology Information's (NCBI) set of interconnected
2204databases (publication, sequence, structure, gene, variation, expression,
2205etc.) from a terminal. Functions take search terms from command-line
2206arguments. Individual operations are combined to build multi-step queries.
2207Record retrieval and formatting normally complete the process.
2208
2209EDirect also provides an argument-driven function that simplifies the
2210extraction of data from document summaries or other results that are returned
2211in structured XML format. This can eliminate the need for writing custom
2212software to answer ad hoc questions.")
2213 (license license:public-domain)))
2214
b16728b0
BW		 2215(define-public exonerate
2216 (package
2217 (name "exonerate")
2218 (version "2.4.0")
2219 (source
2220 (origin
2221 (method url-fetch)
2222 (uri
2223 (string-append
2224 "http://ftp.ebi.ac.uk/pub/software/vertebrategenomics/exonerate/"
2225 "exonerate-" version ".tar.gz"))
2226 (sha256
2227 (base32
2228 "0hj0m9xygiqsdxvbg79wq579kbrx1mdrabi2bzqz2zn9qwfjcjgq"))))
2229 (build-system gnu-build-system)
2230 (arguments
2231 `(#:parallel-build? #f)) ; Building in parallel fails on some machines.
2232 (native-inputs
2233 `(("pkg-config" ,pkg-config)))
2234 (inputs
2235 `(("glib" ,glib)))
2236 (home-page
2237 "https://www.ebi.ac.uk/about/vertebrate-genomics/software/exonerate")
2238 (synopsis "Generic tool for biological sequence alignment")
2239 (description
2240 "Exonerate is a generic tool for pairwise sequence comparison. It allows
2241the alignment of sequences using a many alignment models, either exhaustive
2242dynamic programming or a variety of heuristics.")
2243 (license license:gpl3)))
2244
e4e5a4d8
RW		 2245(define-public express
2246 (package
2247 (name "express")
2248 (version "1.5.1")
2249 (source (origin
2250 (method url-fetch)
2251 (uri
2252 (string-append
2253 "http://bio.math.berkeley.edu/eXpress/downloads/express-"
2254 version "/express-" version "-src.tgz"))
2255 (sha256
2256 (base32
2257 "03rczxd0gjp2l1jxcmjfmf5j94j77zqyxa6x063zsc585nj40n0c"))))
2258 (build-system cmake-build-system)
2259 (arguments
2260 `(#:tests? #f ;no "check" target
2261 #:phases
2262 (alist-cons-after
2263 'unpack 'use-shared-boost-libs-and-set-bamtools-paths
2264 (lambda* (#:key inputs #:allow-other-keys)
2265 (substitute* "CMakeLists.txt"
2266 (("set\\(Boost_USE_STATIC_LIBS ON\\)")
2267 "set(Boost_USE_STATIC_LIBS OFF)")
2268 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/bamtools/include")
2269 (string-append (assoc-ref inputs "bamtools") "/include/bamtools")))
2270 (substitute* "src/CMakeLists.txt"
2271 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/\\.\\./bamtools/lib")
2272 (string-append (assoc-ref inputs "bamtools") "/lib/bamtools")))
2273 #t)
2274 %standard-phases)))
2275 (inputs
2276 `(("boost" ,boost)
2277 ("bamtools" ,bamtools)
2278 ("protobuf" ,protobuf)
2279 ("zlib" ,zlib)))
2280 (home-page "http://bio.math.berkeley.edu/eXpress")
2281 (synopsis "Streaming quantification for high-throughput genomic sequencing")
2282 (description
2283 "eXpress is a streaming tool for quantifying the abundances of a set of
2284target sequences from sampled subsequences. Example applications include
2285transcript-level RNA-Seq quantification, allele-specific/haplotype expression
2286analysis (from RNA-Seq), transcription factor binding quantification in
2287ChIP-Seq, and analysis of metagenomic data.")
2288 (license license:artistic2.0)))
2289
f3674b1c
BW		 2290(define-public express-beta-diversity
2291 (package
2292 (name "express-beta-diversity")
2293 (version "1.0.7")
2294 (source (origin
2295 (method url-fetch)
2296 (uri
2297 (string-append
2298 "https://github.com/dparks1134/ExpressBetaDiversity/archive/v"
2299 version ".tar.gz"))
2300 (file-name (string-append name "-" version ".tar.gz"))
2301 (sha256
2302 (base32
2303 "1djvdlmqvjf6h0zq7w36y8cl5cli6rgj86x65znl48agnwmzxfxr"))))
2304 (build-system gnu-build-system)
2305 (arguments
2306 `(#:phases
2307 (modify-phases %standard-phases
2308 (delete 'configure)
2309 (add-before 'build 'enter-source (lambda _ (chdir "source") #t))
2310 (replace 'check
2311 (lambda _ (zero? (system* "../bin/ExpressBetaDiversity"
2312 "-u"))))
2313 (add-after 'check 'exit-source (lambda _ (chdir "..") #t))
2314 (replace 'install
2315 (lambda* (#:key outputs #:allow-other-keys)
2316 (let ((bin (string-append (assoc-ref outputs "out")
2317 "/bin")))
2318 (mkdir-p bin)
2319 (copy-file "scripts/convertToEBD.py"
2320 (string-append bin "/convertToEBD.py"))
2321 (copy-file "bin/ExpressBetaDiversity"
2322 (string-append bin "/ExpressBetaDiversity"))
2323 #t))))))
2324 (inputs
2325 `(("python" ,python-2)))
2326 (home-page "http://kiwi.cs.dal.ca/Software/ExpressBetaDiversity")
2327 (synopsis "Taxon- and phylogenetic-based beta diversity measures")
2328 (description
2329 "Express Beta Diversity (EBD) calculates ecological beta diversity
2330(dissimilarity) measures between biological communities. EBD implements a
2331variety of diversity measures including those that make use of phylogenetic
2332similarity of community members.")
2333 (license license:gpl3+)))
2334
12b04cbe
BW		 2335(define-public fasttree
2336 (package
2337 (name "fasttree")
e03a5153 2338 (version "2.1.9")
12b04cbe
BW		 2339 (source (origin
2340 (method url-fetch)
2341 (uri (string-append
2342 "http://www.microbesonline.org/fasttree/FastTree-"
2343 version ".c"))
2344 (sha256
2345 (base32
e03a5153 2346 "0ljvvw8i1als1wbfzvrf15c3ii2vw9db20a259g6pzg34xyyb97k"))))
12b04cbe
BW		 2347 (build-system gnu-build-system)
2348 (arguments
2349 `(#:tests? #f ; no "check" target
2350 #:phases
2351 (modify-phases %standard-phases
2352 (delete 'unpack)
2353 (delete 'configure)
2354 (replace 'build
e03a5153
BW		 2355 (lambda* (#:key source #:allow-other-keys)
2356 (and (zero? (system* "gcc"
2357 "-O3"
2358 "-finline-functions"
2359 "-funroll-loops"
2360 "-Wall"
2361 "-o"
2362 "FastTree"
2363 source
2364 "-lm"))
2365 (zero? (system* "gcc"
2366 "-DOPENMP"
2367 "-fopenmp"
2368 "-O3"
2369 "-finline-functions"
2370 "-funroll-loops"
2371 "-Wall"
2372 "-o"
2373 "FastTreeMP"
2374 source
2375 "-lm")))))
12b04cbe 2376 (replace 'install
e03a5153
BW		 2377 (lambda* (#:key outputs #:allow-other-keys)
2378 (let ((bin (string-append (assoc-ref outputs "out")
2379 "/bin")))
2380 (mkdir-p bin)
2381 (copy-file "FastTree"
2382 (string-append bin "/FastTree"))
2383 (copy-file "FastTreeMP"
2384 (string-append bin "/FastTreeMP"))
2385 #t))))))
12b04cbe
BW		 2386 (home-page "http://www.microbesonline.org/fasttree")
2387 (synopsis "Infers approximately-maximum-likelihood phylogenetic trees")
2388 (description
2389 "FastTree can handle alignments with up to a million of sequences in a
2390reasonable amount of time and memory. For large alignments, FastTree is
2391100-1,000 times faster than PhyML 3.0 or RAxML 7.")
2392 (license license:gpl2+)))
2393
2127cedb
RW		 2394(define-public fastx-toolkit
2395 (package
2396 (name "fastx-toolkit")
2397 (version "0.0.14")
2398 (source (origin
2399 (method url-fetch)
2400 (uri
2401 (string-append
2402 "https://github.com/agordon/fastx_toolkit/releases/download/"
2403 version "/fastx_toolkit-" version ".tar.bz2"))
2404 (sha256
2405 (base32
2406 "01jqzw386873sr0pjp1wr4rn8fsga2vxs1qfmicvx1pjr72007wy"))))
2407 (build-system gnu-build-system)
2408 (inputs
2409 `(("libgtextutils" ,libgtextutils)))
2410 (native-inputs
2411 `(("pkg-config" ,pkg-config)))
2412 (home-page "http://hannonlab.cshl.edu/fastx_toolkit/")
2413 (synopsis "Tools for FASTA/FASTQ file preprocessing")
2414 (description
2415 "The FASTX-Toolkit is a collection of command line tools for Short-Reads
2416FASTA/FASTQ files preprocessing.
2417
2418Next-Generation sequencing machines usually produce FASTA or FASTQ files,
2419containing multiple short-reads sequences. The main processing of such
2420FASTA/FASTQ files is mapping the sequences to reference genomes. However, it
2421is sometimes more productive to preprocess the files before mapping the
2422sequences to the genome---manipulating the sequences to produce better mapping
2423results. The FASTX-Toolkit tools perform some of these preprocessing tasks.")
2424 (license license:agpl3+)))
2425
d7678942
RW		 2426(define-public flexbar
2427 (package
2428 (name "flexbar")
2429 (version "2.5")
2430 (source (origin
2431 (method url-fetch)
2432 (uri
2433 (string-append "mirror://sourceforge/flexbar/"
2434 version "/flexbar_v" version "_src.tgz"))
2435 (sha256
2436 (base32
2437 "13jaykc3y1x8y5nn9j8ljnb79s5y51kyxz46hdmvvjj6qhyympmf"))))
2438 (build-system cmake-build-system)
2439 (arguments
4ca009c0 2440 `(#:configure-flags (list
d7678942
RW		 2441 (string-append "-DFLEXBAR_BINARY_DIR="
2442 (assoc-ref %outputs "out")
2443 "/bin/"))
2444 #:phases
4ca009c0
RW		 2445 (alist-replace
2446 'check
2447 (lambda* (#:key outputs #:allow-other-keys)
2448 (setenv "PATH" (string-append
2449 (assoc-ref outputs "out") "/bin:"
2450 (getenv "PATH")))
2451 (chdir "../flexbar_v2.5_src/test")
2452 (zero? (system* "bash" "flexbar_validate.sh")))
2453 (alist-delete 'install %standard-phases))))
d7678942
RW		 2454 (inputs
2455 `(("tbb" ,tbb)
2456 ("zlib" ,zlib)))
2457 (native-inputs
2458 `(("pkg-config" ,pkg-config)
2459 ("seqan" ,seqan)))
2460 (home-page "http://flexbar.sourceforge.net")
2461 (synopsis "Barcode and adapter removal tool for sequencing platforms")
2462 (description
2463 "Flexbar preprocesses high-throughput nucleotide sequencing data
2464efficiently. It demultiplexes barcoded runs and removes adapter sequences.
2465Moreover, trimming and filtering features are provided. Flexbar increases
2466read mapping rates and improves genome and transcriptome assemblies. It
2467supports next-generation sequencing data in fasta/q and csfasta/q format from
2468Illumina, Roche 454, and the SOLiD platform.")
2469 (license license:gpl3)))
2470
19f4554c
BW		 2471(define-public fraggenescan
2472 (package
2473 (name "fraggenescan")
2474 (version "1.20")
2475 (source
2476 (origin
2477 (method url-fetch)
2478 (uri
2479 (string-append "mirror://sourceforge/fraggenescan/"
2480 "FragGeneScan" version ".tar.gz"))
2481 (sha256
2482 (base32 "1zzigqmvqvjyqv4945kv6nc5ah2xxm1nxgrlsnbzav3f5c0n0pyj"))))
2483 (build-system gnu-build-system)
2484 (arguments
2485 `(#:phases
2486 (modify-phases %standard-phases
2487 (delete 'configure)
2488 (add-before 'build 'patch-paths
2489 (lambda* (#:key outputs #:allow-other-keys)
2490 (let* ((out (string-append (assoc-ref outputs "out")))
2491 (share (string-append out "/share/fraggenescan/")))
2492 (substitute* "run_FragGeneScan.pl"
2493 (("system\\(\"rm")
2494 (string-append "system(\"" (which "rm")))
2495 (("system\\(\"mv")
2496 (string-append "system(\"" (which "mv")))
2497 ;; This script and other programs expect the training files
2498 ;; to be in the non-standard location bin/train/XXX. Change
2499 ;; this to be share/fraggenescan/train/XXX instead.
2500 (("^\\$train.file = \\$dir.*")
2501 (string-append "$train_file = \""
2502 share
2503 "train/\".$FGS_train_file;")))
2504 (substitute* "run_hmm.c"
2505 (("^ strcat\\(train_dir, \\\"train/\\\"\\);")
2506 (string-append " strcpy(train_dir, \"" share "/train/\");")))
2507 (substitute* "post_process.pl"
2508 (("^my \\$dir = substr.*")
2509 (string-append "my $dir = \"" share "\";"))))
2510 #t))
2511 (replace 'build
2512 (lambda _ (and (zero? (system* "make" "clean"))
2513 (zero? (system* "make" "fgs")))))
2514 (replace 'install
2515 (lambda* (#:key outputs #:allow-other-keys)
2516 (let* ((out (string-append (assoc-ref outputs "out")))
2517 (bin (string-append out "/bin/"))
2518 (share (string-append out "/share/fraggenescan/train")))
2519 (install-file "run_FragGeneScan.pl" bin)
2520 (install-file "FragGeneScan" bin)
2521 (install-file "FGS_gff.py" bin)
2522 (install-file "post_process.pl" bin)
2523 (copy-recursively "train" share))))
2524 (delete 'check)
2525 (add-after 'install 'post-install-check
2526 ;; In lieu of 'make check', run one of the examples and check the
2527 ;; output files gets created.
2528 (lambda* (#:key outputs #:allow-other-keys)
2529 (let* ((out (string-append (assoc-ref outputs "out")))
2530 (bin (string-append out "/bin/")))
2531 (and (zero? (system* (string-append bin "run_FragGeneScan.pl")
2532 "-genome=./example/NC_000913.fna"
2533 "-out=./test2"
2534 "-complete=1"
2535 "-train=complete"))
2536 (file-exists? "test2.faa")
2537 (file-exists? "test2.ffn")
2538 (file-exists? "test2.gff")
2539 (file-exists? "test2.out"))))))))
2540 (inputs
2541 `(("perl" ,perl)
2542 ("python" ,python-2))) ;not compatible with python 3.
2543 (home-page "https://sourceforge.net/projects/fraggenescan/")
2544 (synopsis "Finds potentially fragmented genes in short reads")
2545 (description
2546 "FragGeneScan is a program for predicting bacterial and archaeal genes in
2547short and error-prone DNA sequencing reads. It can also be applied to predict
2548genes in incomplete assemblies or complete genomes.")
2549 ;; GPL3+ according to private correspondense with the authors.
2550 (license license:gpl3+)))
2551
81f3e0c1
BW		 2552(define-public fxtract
2553 (let ((util-commit "776ca85a18a47492af3794745efcb4a905113115"))
2554 (package
2555 (name "fxtract")
2556 (version "2.3")
2557 (source
2558 (origin
2559 (method url-fetch)
2560 (uri (string-append
2561 "https://github.com/ctSkennerton/fxtract/archive/"
2562 version ".tar.gz"))
2563 (file-name (string-append "ctstennerton-util-"
2564 (string-take util-commit 7)
2565 "-checkout"))
2566 (sha256
2567 (base32
2568 "0275cfdhis8517hm01is62062swmi06fxzifq7mr3knbbxjlaiwj"))))
2569 (build-system gnu-build-system)
2570 (arguments
2571 `(#:make-flags (list
2572 (string-append "PREFIX=" (assoc-ref %outputs "out"))
2573 "CC=gcc")
2574 #:test-target "fxtract_test"
2575 #:phases
2576 (modify-phases %standard-phases
2577 (delete 'configure)
2578 (add-before 'build 'copy-util
2579 (lambda* (#:key inputs #:allow-other-keys)
2580 (rmdir "util")
2581 (copy-recursively (assoc-ref inputs "ctskennerton-util") "util")
2582 #t))
2583 ;; Do not use make install as this requires additional dependencies.
2584 (replace 'install
2585 (lambda* (#:key outputs #:allow-other-keys)
2586 (let* ((out (assoc-ref outputs "out"))
2587 (bin (string-append out"/bin")))
2588 (install-file "fxtract" bin)
2589 #t))))))
2590 (inputs
2591 `(("pcre" ,pcre)
2592 ("zlib" ,zlib)))
2593 (native-inputs
2594 ;; ctskennerton-util is licensed under GPL2.
2595 `(("ctskennerton-util"
2596 ,(origin
2597 (method git-fetch)
2598 (uri (git-reference
2599 (url "https://github.com/ctSkennerton/util.git")
2600 (commit util-commit)))
2601 (file-name (string-append
2602 "ctstennerton-util-" util-commit "-checkout"))
2603 (sha256
2604 (base32
2605 "0cls1hd4vgj3f36fpzzg4xc77d6f3hpc60cbpfmn2gdr7ykzzad7"))))))
2606 (home-page "https://github.com/ctSkennerton/fxtract")
2607 (synopsis "Extract sequences from FASTA and FASTQ files")
2608 (description
2609 "Fxtract extracts sequences from a protein or nucleotide fastx (FASTA
2610or FASTQ) file given a subsequence. It uses a simple substring search for
2611basic tasks but can change to using POSIX regular expressions, PCRE, hash
2612lookups or multi-pattern searching as required. By default fxtract looks in
2613the sequence of each record but can also be told to look in the header,
2614comment or quality sections.")
afde1a26
BW		 2615 ;; 'util' requires SSE instructions.
2616 (supported-systems '("x86_64-linux"))
81f3e0c1
BW		 2617 (license license:expat))))
2618
5854f685
RW		 2619(define-public grit
2620 (package
2621 (name "grit")
2622 (version "2.0.2")
2623 (source (origin
2624 (method url-fetch)
2625 (uri (string-append
2626 "https://github.com/nboley/grit/archive/"
2627 version ".tar.gz"))
2628 (file-name (string-append name "-" version ".tar.gz"))
2629 (sha256
2630 (base32
2631 "157in84dj70wimbind3x7sy1whs3h57qfgcnj2s6lrd38fbrb7mj"))))
2632 (build-system python-build-system)
2633 (arguments
2634 `(#:python ,python-2
2635 #:phases
2636 (alist-cons-after
2637 'unpack 'generate-from-cython-sources
2638 (lambda* (#:key inputs outputs #:allow-other-keys)
2639 ;; Delete these C files to force fresh generation from pyx sources.
2640 (delete-file "grit/sparsify_support_fns.c")
2641 (delete-file "grit/call_peaks_support_fns.c")
2642 (substitute* "setup.py"
2643 (("Cython.Setup") "Cython.Build")
2644 ;; Add numpy include path to fix compilation
2645 (("pyx\", \\]")
2646 (string-append "pyx\", ], include_dirs = ['"
2647 (assoc-ref inputs "python-numpy")
2648 "/lib/python2.7/site-packages/numpy/core/include/"
2649 "']"))) #t)
2650 %standard-phases)))
2651 (inputs
2652 `(("python-scipy" ,python2-scipy)
2653 ("python-numpy" ,python2-numpy)
2654 ("python-pysam" ,python2-pysam)
2655 ("python-networkx" ,python2-networkx)))
2656 (native-inputs
f3b98f4f 2657 `(("python-cython" ,python2-cython)))
5854f685
RW		 2658 (home-page "http://grit-bio.org")
2659 (synopsis "Tool for integrative analysis of RNA-seq type assays")
2660 (description
2661 "GRIT is designed to use RNA-seq, TES, and TSS data to build and quantify
2662full length transcript models. When none of these data sources are available,
2663GRIT can be run by providing a candidate set of TES or TSS sites. In
2664addition, GRIT can merge in reference junctions and gene boundaries. GRIT can
2665also be run in quantification mode, where it uses a provided GTF file and just
2666estimates transcript expression.")
2667 (license license:gpl3+)))
2668
346a829a
RW		 2669(define-public hisat
2670 (package
2671 (name "hisat")
2672 (version "0.1.4")
2673 (source (origin
2674 (method url-fetch)
2675 (uri (string-append
2676 "http://ccb.jhu.edu/software/hisat/downloads/hisat-"
2677 version "-beta-source.zip"))
2678 (sha256
2679 (base32
2680 "1k381ydranqxp09yf2y7w1d0chz5d59vb6jchi89hbb0prq19lk5"))))
2681 (build-system gnu-build-system)
2682 (arguments
e58d01fa
RW		 2683 `(#:tests? #f ;no check target
2684 #:make-flags '("allall"
2685 ;; Disable unsupported `popcnt' instructions on
2686 ;; architectures other than x86_64
2687 ,@(if (string-prefix? "x86_64"
2688 (or (%current-target-system)
2689 (%current-system)))
2690 '()
2691 '("POPCNT_CAPABILITY=0")))
346a829a 2692 #:phases
da6dd842
LC		 2693 (alist-cons-after
2694 'unpack 'patch-sources
2695 (lambda _
2696 ;; XXX Cannot use snippet because zip files are not supported
2697 (substitute* "Makefile"
2698 (("^CC = .*$") "CC = gcc")
2699 (("^CPP = .*$") "CPP = g++")
2700 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
2701 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
2702 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\""))
2703 (substitute* '("hisat-build" "hisat-inspect")
2704 (("/usr/bin/env") (which "env"))))
2705 (alist-replace
2706 'install
2707 (lambda* (#:key outputs #:allow-other-keys)
84590149 2708 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
96c46210
LC		 2709 (for-each (lambda (file)
2710 (install-file file bin))
2711 (find-files
2712 "."
2713 "hisat(-(build|align|inspect)(-(s|l)(-debug)*)*)*$"))))
da6dd842 2714 (alist-delete 'configure %standard-phases)))))
346a829a
RW		 2715 (native-inputs
2716 `(("unzip" ,unzip)))
2717 (inputs
2718 `(("perl" ,perl)
2719 ("python" ,python)
2720 ("zlib" ,zlib)))
60af3d82
RW		 2721 ;; Non-portable SSE instructions are used so building fails on platforms
2722 ;; other than x86_64.
2723 (supported-systems '("x86_64-linux"))
346a829a
RW		 2724 (home-page "http://ccb.jhu.edu/software/hisat/index.shtml")
2725 (synopsis "Hierarchical indexing for spliced alignment of transcripts")
2726 (description
2727 "HISAT is a fast and sensitive spliced alignment program for mapping
2728RNA-seq reads. In addition to one global FM index that represents a whole
2729genome, HISAT uses a large set of small FM indexes that collectively cover the
2730whole genome. These small indexes (called local indexes) combined with
2731several alignment strategies enable effective alignment of RNA-seq reads, in
2732particular, reads spanning multiple exons.")
2733 (license license:gpl3+)))
2734
c684629f
BW		 2735(define-public hmmer
2736 (package
2737 (name "hmmer")
2738 (version "3.1b2")
2739 (source (origin
2740 (method url-fetch)
2741 (uri (string-append
a83e6046 2742 "http://eddylab.org/software/hmmer"
c684629f
BW		 2743 (version-prefix version 1) "/"
2744 version "/hmmer-" version ".tar.gz"))
2745 (sha256
2746 (base32
2747 "0djmgc0pfli0jilfx8hql1axhwhqxqb8rxg2r5rg07aw73sfs5nx"))))
2748 (build-system gnu-build-system)
b3546174 2749 (native-inputs `(("perl" ,perl)))
a83e6046 2750 (home-page "http://hmmer.org/")
c684629f
BW		 2751 (synopsis "Biosequence analysis using profile hidden Markov models")
2752 (description
2753 "HMMER is used for searching sequence databases for homologs of protein
2754sequences, and for making protein sequence alignments. It implements methods
2755using probabilistic models called profile hidden Markov models (profile
2756HMMs).")
2757 (license (list license:gpl3+
2758 ;; The bundled library 'easel' is distributed
2759 ;; under The Janelia Farm Software License.
2760 (license:non-copyleft
2761 "file://easel/LICENSE"
2762 "See easel/LICENSE in the distribution.")))))
2763
85652f59
RW		 2764(define-public htseq
2765 (package
2766 (name "htseq")
2767 (version "0.6.1")
2768 (source (origin
2769 (method url-fetch)
2770 (uri (string-append
2771 "https://pypi.python.org/packages/source/H/HTSeq/HTSeq-"
2772 version ".tar.gz"))
2773 (sha256
2774 (base32
2775 "1i85ppf2j2lj12m0x690qq5nn17xxk23pbbx2c83r8ayb5wngzwv"))))
2776 (build-system python-build-system)
2777 (arguments `(#:python ,python-2)) ; only Python 2 is supported
0536727e
RW		 2778 ;; Numpy needs to be propagated when htseq is used as a Python library.
2779 (propagated-inputs
2780 `(("python-numpy" ,python2-numpy)))
578b05d9
RW		 2781 (inputs
2782 `(("python-pysam" ,python2-pysam)))
85652f59
RW		 2783 (home-page "http://www-huber.embl.de/users/anders/HTSeq/")
2784 (synopsis "Analysing high-throughput sequencing data with Python")
2785 (description
2786 "HTSeq is a Python package that provides infrastructure to process data
2787from high-throughput sequencing assays.")
2788 (license license:gpl3+)))
2789
1ad15c16 2790(define-public java-htsjdk
15a3c3d4 2791 (package
1ad15c16 2792 (name "java-htsjdk")
15a3c3d4
RW		 2793 (version "1.129")
2794 (source (origin
2795 (method url-fetch)
2796 (uri (string-append
2797 "https://github.com/samtools/htsjdk/archive/"
2798 version ".tar.gz"))
2799 (file-name (string-append name "-" version ".tar.gz"))
2800 (sha256
2801 (base32
2802 "0asdk9b8jx2ij7yd6apg9qx03li8q7z3ml0qy2r2qczkra79y6fw"))
2803 (modules '((guix build utils)))
2804 ;; remove build dependency on git
2805 (snippet '(substitute* "build.xml"
2806 (("failifexecutionfails=\"true\"")
2807 "failifexecutionfails=\"false\"")))))
10b4a969 2808 (build-system ant-build-system)
15a3c3d4 2809 (arguments
10b4a969
RW		 2810 `(#:tests? #f ; test require Internet access
2811 #:make-flags
2812 (list (string-append "-Ddist=" (assoc-ref %outputs "out")
2813 "/share/java/htsjdk/"))
2814 #:build-target "all"
2815 #:phases
2816 (modify-phases %standard-phases
2817 ;; The build phase also installs the jars
2818 (delete 'install))))
15a3c3d4
RW		 2819 (home-page "http://samtools.github.io/htsjdk/")
2820 (synopsis "Java API for high-throughput sequencing data (HTS) formats")
2821 (description
2822 "HTSJDK is an implementation of a unified Java library for accessing
2823common file formats, such as SAM and VCF, used for high-throughput
2824sequencing (HTS) data. There are also an number of useful utilities for
2825manipulating HTS data.")
2826 (license license:expat)))
2827
e7c09730
RW		 2828(define-public htslib
2829 (package
2830 (name "htslib")
6c737963 2831 (version "1.3.1")
e7c09730
RW		 2832 (source (origin
2833 (method url-fetch)
2834 (uri (string-append
2835 "https://github.com/samtools/htslib/releases/download/"
2836 version "/htslib-" version ".tar.bz2"))
2837 (sha256
2838 (base32
6c737963 2839 "1rja282fwdc25ql6izkhdyh8ppw8x2fs0w0js78zgkmqjlikmma9"))))
e7c09730
RW		 2840 (build-system gnu-build-system)
2841 (arguments
2842 `(#:phases
2843 (modify-phases %standard-phases
2844 (add-after
2845 'unpack 'patch-tests
2846 (lambda _
2847 (substitute* "test/test.pl"
2848 (("/bin/bash") (which "bash")))
2849 #t)))))
2850 (inputs
2851 `(("zlib" ,zlib)))
2852 (native-inputs
2853 `(("perl" ,perl)))
2854 (home-page "http://www.htslib.org")
2855 (synopsis "C library for reading/writing high-throughput sequencing data")
2856 (description
2857 "HTSlib is a C library for reading/writing high-throughput sequencing
2858data. It also provides the bgzip, htsfile, and tabix utilities.")
2859 ;; Files under cram/ are released under the modified BSD license;
2860 ;; the rest is released under the Expat license
2861 (license (list license:expat license:bsd-3))))
2862
c4325f62
RW		 2863(define-public idr
2864 (package
2865 (name "idr")
2866 (version "2.0.0")
2867 (source (origin
2868 (method url-fetch)
2869 (uri (string-append
2870 "https://github.com/nboley/idr/archive/"
2871 version ".tar.gz"))
2872 (file-name (string-append name "-" version ".tar.gz"))
2873 (sha256
2874 (base32
2875 "1k3x44biak00aiv3hpm1yd6nn4hhp7n0qnbs3zh2q9sw7qr1qj5r"))))
2876 (build-system python-build-system)
2877 (arguments
2878 `(#:phases
2879 (modify-phases %standard-phases
2880 (add-after
2881 'install 'wrap-program
2882 (lambda* (#:key inputs outputs #:allow-other-keys)
2883 (let* ((out (assoc-ref outputs "out"))
2884 (python-version (string-take (string-take-right
2885 (assoc-ref inputs "python") 5) 3))
2886 (path (string-join
2887 (map (lambda (name)
2888 (string-append (assoc-ref inputs name)
2889 "/lib/python" python-version
2890 "/site-packages"))
2891 '("python-scipy"
2892 "python-numpy"
2893 "python-matplotlib"))
2894 ":")))
2895 (wrap-program (string-append out "/bin/idr")
2896 `("PYTHONPATH" ":" prefix (,path))))
2897 #t)))))
2898 (inputs
2899 `(("python-scipy" ,python-scipy)
2900 ("python-numpy" ,python-numpy)
2901 ("python-matplotlib" ,python-matplotlib)))
2902 (native-inputs
f3b98f4f 2903 `(("python-cython" ,python-cython)))
c4325f62
RW		 2904 (home-page "https://github.com/nboley/idr")
2905 (synopsis "Tool to measure the irreproducible discovery rate (IDR)")
2906 (description
2907 "The IDR (Irreproducible Discovery Rate) framework is a unified approach
2908to measure the reproducibility of findings identified from replicate
2909experiments and provide highly stable thresholds based on reproducibility.")
2910 (license license:gpl3+)))
2911
43c565d2
RW		 2912(define-public jellyfish
2913 (package
2914 (name "jellyfish")
2915 (version "2.2.4")
2916 (source (origin
2917 (method url-fetch)
2918 (uri (string-append "https://github.com/gmarcais/Jellyfish/"
2919 "releases/download/v" version
2920 "/jellyfish-" version ".tar.gz"))
2921 (sha256
2922 (base32
2923 "0a6xnynqy2ibfbfz86b9g2m2dgm7f1469pmymkpam333gi3p26nk"))))
2924 (build-system gnu-build-system)
2925 (outputs '("out" ;for library
2926 "ruby" ;for Ruby bindings
2927 "python")) ;for Python bindings
2928 (arguments
2929 `(#:configure-flags
2930 (list (string-append "--enable-ruby-binding="
2931 (assoc-ref %outputs "ruby"))
2932 (string-append "--enable-python-binding="
2933 (assoc-ref %outputs "python")))
2934 #:phases
2935 (modify-phases %standard-phases
2936 (add-before 'check 'set-SHELL-variable
2937 (lambda _
2938 ;; generator_manager.hpp either uses /bin/sh or $SHELL
2939 ;; to run tests.
2940 (setenv "SHELL" (which "bash"))
2941 #t)))))
2942 (native-inputs
2943 `(("bc" ,bc)
2944 ("time" ,time)
2945 ("ruby" ,ruby)
2946 ("python" ,python-2)))
2947 (synopsis "Tool for fast counting of k-mers in DNA")
2948 (description
2949 "Jellyfish is a tool for fast, memory-efficient counting of k-mers in
2950DNA. A k-mer is a substring of length k, and counting the occurrences of all
2951such substrings is a central step in many analyses of DNA sequence. Jellyfish
2952is a command-line program that reads FASTA and multi-FASTA files containing
2953DNA sequences. It outputs its k-mer counts in a binary format, which can be
2954translated into a human-readable text format using the @code{jellyfish dump}
2955command, or queried for specific k-mers with @code{jellyfish query}.")
2956 (home-page "http://www.genome.umd.edu/jellyfish.html")
6e8faf77
EF		 2957 ;; From their website: JELLYFISH runs on 64-bit Intel-compatible processors
2958 (supported-systems '("x86_64-linux"))
43c565d2
RW		 2959 ;; The combined work is published under the GPLv3 or later. Individual
2960 ;; files such as lib/jsoncpp.cpp are released under the Expat license.
2961 (license (list license:gpl3+ license:expat))))
2962
94ff3157
BW		 2963(define-public khmer
2964 (package
2965 (name "khmer")
2966 (version "2.0")
2967 (source
2968 (origin
2969 (method url-fetch)
2970 (uri (pypi-uri "khmer" version))
2971 (sha256
2972 (base32
2973 "0wb05shqh77v00256qlm68vbbx3kl76fyzihszbz5nhanl4ni33a"))
2974 (patches (search-patches "khmer-use-libraries.patch"))))
2975 (build-system python-build-system)
2976 (arguments
2977 `(#:phases
2978 (modify-phases %standard-phases
2979 (add-after 'unpack 'set-paths
2980 (lambda* (#:key inputs outputs #:allow-other-keys)
2981 ;; Delete bundled libraries.
2982 (delete-file-recursively "third-party/zlib")
2983 (delete-file-recursively "third-party/bzip2")
2984 ;; Replace bundled seqan.
2985 (let* ((seqan-all "third-party/seqan")
2986 (seqan-include (string-append
2987 seqan-all "/core/include")))
2988 (delete-file-recursively seqan-all)
2989 (copy-recursively (string-append (assoc-ref inputs "seqan")
2990 "/include/seqan")
2991 (string-append seqan-include "/seqan")))
2992 ;; We do not replace the bundled MurmurHash as the canonical
2993 ;; repository for this code 'SMHasher' is unsuitable for
2994 ;; providing a library. See
2995 ;; https://lists.gnu.org/archive/html/guix-devel/2016-06/msg00977.html
2996 #t))
2997 (add-after 'unpack 'set-cc
2998 (lambda _
2999 (setenv "CC" "gcc")
3000 #t))
3001 ;; It is simpler to test after installation.
3002 (delete 'check)
3003 (add-after 'install 'post-install-check
3004 (lambda* (#:key inputs outputs #:allow-other-keys)
3005 (let ((out (assoc-ref outputs "out")))
3006 (setenv "PATH"
3007 (string-append
3008 (getenv "PATH")
3009 ":"
3010 (assoc-ref outputs "out")
3011 "/bin"))
3012 (setenv "PYTHONPATH"
3013 (string-append
3014 (getenv "PYTHONPATH")
3015 ":"
3016 out
3017 "/lib/python"
3018 (string-take (string-take-right
3019 (assoc-ref inputs "python") 5) 3)
3020 "/site-packages"))
3021 (with-directory-excursion "build"
3022 (zero? (system* "nosetests" "khmer" "--attr"
3023 "!known_failing")))))))))
3024 (native-inputs
3025 `(("seqan" ,seqan)
3026 ("python-nose" ,python-nose)))
3027 (inputs
3028 `(("zlib" ,zlib)
3029 ("bzip2" ,bzip2)
3030 ("python-screed" ,python-screed)
3031 ("python-bz2file" ,python-bz2file)))
3032 (home-page "https://khmer.readthedocs.org/")
3033 (synopsis "K-mer counting, filtering and graph traversal library")
3034 (description "The khmer software is a set of command-line tools for
3035working with DNA shotgun sequencing data from genomes, transcriptomes,
3036metagenomes and single cells. Khmer can make de novo assemblies faster, and
3037sometimes better. Khmer can also identify and fix problems with shotgun
3038data.")
8157af2e
EF		 3039 ;; When building on i686, armhf and mips64el, we get the following error:
3040 ;; error: ['khmer', 'khmer.tests', 'oxli'] require 64-bit operating system
3041 (supported-systems '("x86_64-linux"))
94ff3157
BW		 3042 (license license:bsd-3)))
3043
d57e6d0f
RW		 3044(define-public macs
3045 (package
3046 (name "macs")
0d0bcaa0 3047 (version "2.1.0.20151222")
d57e6d0f
RW		 3048 (source (origin
3049 (method url-fetch)
43ec07f1 3050 (uri (pypi-uri "MACS2" version))
d57e6d0f
RW		 3051 (sha256
3052 (base32
0d0bcaa0 3053 "1r2hcz6irhcq7lwbafjks98jbn34hv05avgbdjnp6w6mlfjkf8x5"))))
d57e6d0f
RW		 3054 (build-system python-build-system)
3055 (arguments
3056 `(#:python ,python-2 ; only compatible with Python 2.7
3057 #:tests? #f)) ; no test target
3058 (inputs
3059 `(("python-numpy" ,python2-numpy)))
d57e6d0f
RW		 3060 (home-page "http://github.com/taoliu/MACS/")
3061 (synopsis "Model based analysis for ChIP-Seq data")
3062 (description
3063 "MACS is an implementation of a ChIP-Seq analysis algorithm for
3064identifying transcript factor binding sites named Model-based Analysis of
3065ChIP-Seq (MACS). MACS captures the influence of genome complexity to evaluate
3066the significance of enriched ChIP regions and it improves the spatial
3067resolution of binding sites through combining the information of both
3068sequencing tag position and orientation.")
3069 (license license:bsd-3)))
3070
41ddebdd
BW		 3071(define-public mafft
3072 (package
3073 (name "mafft")
71461f88 3074 (version "7.305")
41ddebdd
BW		 3075 (source (origin
3076 (method url-fetch)
3077 (uri (string-append
3078 "http://mafft.cbrc.jp/alignment/software/mafft-" version
3079 "-without-extensions-src.tgz"))
3080 (file-name (string-append name "-" version ".tgz"))
3081 (sha256
3082 (base32
71461f88 3083 "0ziim7g58n3z8gppsa713f5fxprl60ldj3xck186z0n9dpp06i8r"))))
41ddebdd
BW		 3084 (build-system gnu-build-system)
3085 (arguments
3086 `(#:tests? #f ; no automated tests, though there are tests in the read me
3087 #:make-flags (let ((out (assoc-ref %outputs "out")))
3088 (list (string-append "PREFIX=" out)
3089 (string-append "BINDIR="
3090 (string-append out "/bin"))))
3091 #:phases
3092 (modify-phases %standard-phases
3093 (add-after 'unpack 'enter-dir
101e8f71 3094 (lambda _ (chdir "core") #t))
41ddebdd 3095 (add-after 'enter-dir 'patch-makefile
101e8f71
BW		 3096 (lambda _
3097 ;; on advice from the MAFFT authors, there is no need to
3098 ;; distribute mafft-profile, mafft-distance, or
3099 ;; mafft-homologs.rb as they are too "specialised".
3100 (substitute* "Makefile"
3101 ;; remove mafft-homologs.rb from SCRIPTS
3102 (("^SCRIPTS = mafft mafft-homologs.rb")
3103 "SCRIPTS = mafft")
3104 ;; remove mafft-homologs from MANPAGES
3105 (("^MANPAGES = mafft.1 mafft-homologs.1")
3106 "MANPAGES = mafft.1")
3107 ;; remove mafft-distance from PROGS
3108 (("^PROGS = dvtditr dndfast7 dndblast sextet5 mafft-distance")
3109 "PROGS = dvtditr dndfast7 dndblast sextet5")
3110 ;; remove mafft-profile from PROGS
3111 (("splittbfast disttbfast tbfast mafft-profile 2cl mccaskillwrap")
3112 "splittbfast disttbfast tbfast f2cl mccaskillwrap")
3113 (("^rm -f mafft-profile mafft-profile.exe") "#")
3114 (("^rm -f mafft-distance mafft-distance.exe") ")#")
3115 ;; do not install MAN pages in libexec folder
3116 (("^\t\\$\\(INSTALL\\) -m 644 \\$\\(MANPAGES\\) \
41ddebdd 3117\\$\\(DESTDIR\\)\\$\\(LIBDIR\\)") "#"))
101e8f71 3118 #t))
02f35bb5
BW		 3119 (add-after 'enter-dir 'patch-paths
3120 (lambda* (#:key inputs #:allow-other-keys)
3121 (substitute* '("pairash.c"
3122 "mafft.tmpl")
3123 (("perl") (which "perl"))
3124 (("([\"`| ])awk" _ prefix)
3125 (string-append prefix (which "awk")))
3126 (("grep") (which "grep")))
3127 #t))
101e8f71
BW		 3128 (delete 'configure)
3129 (add-after 'install 'wrap-programs
3130 (lambda* (#:key outputs #:allow-other-keys)
3131 (let* ((out (assoc-ref outputs "out"))
3132 (bin (string-append out "/bin"))
3133 (path (string-append
3134 (assoc-ref %build-inputs "coreutils") "/bin:")))
3135 (for-each (lambda (file)
3136 (wrap-program file
3137 `("PATH" ":" prefix (,path))))
3138 (find-files bin)))
3139 #t)))))
41ddebdd 3140 (inputs
02f35bb5 3141 `(("perl" ,perl)
71461f88 3142 ("ruby" ,ruby)
02f35bb5 3143 ("gawk" ,gawk)
101e8f71
BW		 3144 ("grep" ,grep)
3145 ("coreutils" ,coreutils)))
41ddebdd
BW		 3146 (home-page "http://mafft.cbrc.jp/alignment/software/")
3147 (synopsis "Multiple sequence alignment program")
3148 (description
3149 "MAFFT offers a range of multiple alignment methods for nucleotide and
3150protein sequences. For instance, it offers L-INS-i (accurate; for alignment
3151of <~200 sequences) and FFT-NS-2 (fast; for alignment of <~30,000
3152sequences).")
3153 (license (license:non-copyleft
3154 "http://mafft.cbrc.jp/alignment/software/license.txt"
3155 "BSD-3 with different formatting"))))
8fd790eb 3156
84be3b99
MB		 3157(define-public mash
3158 (package
3159 (name "mash")
3160 (version "1.1.1")
3161 (source (origin
3162 (method url-fetch)
3163 (uri (string-append
3164 "https://github.com/marbl/mash/archive/v"
3165 version ".tar.gz"))
3166 (file-name (string-append name "-" version ".tar.gz"))
3167 (sha256
3168 (base32
3169 "08znbvqq5xknfhmpp3wcj574zvi4p7i8zifi67c9qw9a6ikp42fj"))
3170 (modules '((guix build utils)))
3171 (snippet
3172 ;; Delete bundled kseq.
3173 ;; TODO: Also delete bundled murmurhash and open bloom filter.
3174 '(delete-file "src/mash/kseq.h"))))
3175 (build-system gnu-build-system)
3176 (arguments
3177 `(#:tests? #f ; No tests.
3178 #:configure-flags
3179 (list
3180 (string-append "--with-capnp=" (assoc-ref %build-inputs "capnproto"))
3181 (string-append "--with-gsl=" (assoc-ref %build-inputs "gsl")))
3182 #:make-flags (list "CC=gcc")
3183 #:phases
3184 (modify-phases %standard-phases
3185 (add-after 'unpack 'fix-includes
3186 (lambda _
3187 (substitute* '("src/mash/Sketch.cpp" "src/mash/CommandFind.cpp")
3188 (("^#include \"kseq\\.h\"")
3189 "#include \"htslib/kseq.h\""))
3190 #t))
3191 (add-before 'configure 'autoconf
3192 (lambda _ (zero? (system* "autoconf")))))))
3193 (native-inputs
3194 `(("autoconf" ,autoconf)
3195 ;; Capnproto and htslib are statically embedded in the final
3196 ;; application. Therefore we also list their licenses, below.
3197 ("capnproto" ,capnproto)
3198 ("htslib" ,htslib)))
3199 (inputs
3200 `(("gsl" ,gsl)
3201 ("zlib" ,zlib)))
3202 (supported-systems '("x86_64-linux"))
3203 (home-page "https://mash.readthedocs.io")
3204 (synopsis "Fast genome and metagenome distance estimation using MinHash")
3205 (description "Mash is a fast sequence distance estimator that uses the
3206MinHash algorithm and is designed to work with genomes and metagenomes in the
3207form of assemblies or reads.")
3208 (license (list license:bsd-3 ; Mash
3209 license:expat ; HTSlib and capnproto
3210 license:public-domain ; MurmurHash 3
3211 license:cpl1.0)))) ; Open Bloom Filter
3212
8fd790eb 3213(define-public metabat
2c3eb4b8
BW		 3214 ;; We package from a git commit because compilation of the released version
3215 ;; fails.
3216 (let ((commit "cbdca756993e66ae57e50a27970595dda9cbde1b"))
3217 (package
3218 (name "metabat")
3219 (version (string-append "0.32.4-1." (string-take commit 8)))
3220 (source
3221 (origin
3222 (method git-fetch)
3223 (uri (git-reference
3224 (url "https://bitbucket.org/berkeleylab/metabat.git")
3225 (commit commit)))
3226 (file-name (string-append name "-" version))
3227 (sha256
3228 (base32
3229 "0byia8nsip6zvc4ha0qkxkxxyjf4x7jcvy48q2dvb0pzr989syzr"))
3230 (patches (search-patches "metabat-remove-compilation-date.patch"))))
8fd790eb
BW		 3231 (build-system gnu-build-system)
3232 (arguments
3233 `(#:phases
3234 (modify-phases %standard-phases
3235 (add-after 'unpack 'fix-includes
45469ebe
BW		 3236 (lambda _
3237 (substitute* "src/BamUtils.h"
3238 (("^#include \"bam/bam\\.h\"")
3239 "#include \"samtools/bam.h\"")
3240 (("^#include \"bam/sam\\.h\"")
3241 "#include \"samtools/sam.h\""))
3242 (substitute* "src/KseqReader.h"
3243 (("^#include \"bam/kseq\\.h\"")
3244 "#include \"htslib/kseq.h\""))
3245 #t))
8fd790eb 3246 (add-after 'unpack 'fix-scons
45469ebe
BW		 3247 (lambda* (#:key inputs #:allow-other-keys)
3248 (substitute* "SConstruct"
3249 (("^htslib_dir = 'samtools'")
3250 (string-append "hitslib_dir = '"
3251 (assoc-ref inputs "htslib")
3252 "'"))
3253 (("^samtools_dir = 'samtools'")
3254 (string-append "samtools_dir = '"
3255 (assoc-ref inputs "htslib")
3256 "'"))
3257 (("^findStaticOrShared\\('bam', hts_lib")
3258 (string-append "findStaticOrShared('bam', '"
3259 (assoc-ref inputs "samtools")
3260 "/lib'"))
3261 ;; Do not distribute README.
3262 (("^env\\.Install\\(idir_prefix, 'README\\.md'\\)") ""))
3263 #t))
8fd790eb
BW		 3264 (delete 'configure)
3265 (replace 'build
3266 (lambda* (#:key inputs outputs #:allow-other-keys)
3267 (mkdir (assoc-ref outputs "out"))
3268 (zero? (system* "scons"
3269 (string-append
3270 "PREFIX="
3271 (assoc-ref outputs "out"))
8fd790eb
BW		 3272 (string-append
3273 "BOOST_ROOT="
3274 (assoc-ref inputs "boost"))
3275 "install"))))
45469ebe 3276 ;; Check and install are carried out during build phase.
8fd790eb
BW		 3277 (delete 'check)
3278 (delete 'install))))
3279 (inputs
3280 `(("zlib" ,zlib)
3281 ("perl" ,perl)
3282 ("samtools" ,samtools)
3283 ("htslib" ,htslib)
3284 ("boost" ,boost)))
3285 (native-inputs
3286 `(("scons" ,scons)))
3287 (home-page "https://bitbucket.org/berkeleylab/metabat")
3288 (synopsis
3289 "Reconstruction of single genomes from complex microbial communities")
3290 (description
3291 "Grouping large genomic fragments assembled from shotgun metagenomic
3292sequences to deconvolute complex microbial communities, or metagenome binning,
3293enables the study of individual organisms and their interactions. MetaBAT is
3294an automated metagenome binning software, which integrates empirical
3295probabilistic distances of genome abundance and tetranucleotide frequency.")
3296 (license (license:non-copyleft "file://license.txt"
2c3eb4b8 3297 "See license.txt in the distribution.")))))
8fd790eb 3298
318c0aee
MB		 3299(define-public minced
3300 (package
3301 (name "minced")
3302 (version "0.2.0")
3303 (source (origin
3304 (method url-fetch)
3305 (uri (string-append
3306 "https://github.com/ctSkennerton/minced/archive/"
3307 version ".tar.gz"))
3308 (file-name (string-append name "-" version ".tar.gz"))
3309 (sha256
3310 (base32
3311 "0wxmlsapxfpxfd3ps9636h7i2xy6la8i42mwh0j2lsky63h63jp1"))))
3312 (build-system gnu-build-system)
3313 (arguments
3314 `(#:test-target "test"
3315 #:phases
3316 (modify-phases %standard-phases
3317 (delete 'configure)
3318 (add-before 'check 'fix-test
3319 (lambda _
3320 ;; Fix test for latest version.
3321 (substitute* "t/Aquifex_aeolicus_VF5.expected"
3322 (("minced:0.1.6") "minced:0.2.0"))
3323 #t))
3324 (replace 'install ; No install target.
3325 (lambda* (#:key inputs outputs #:allow-other-keys)
3326 (let* ((out (assoc-ref outputs "out"))
3327 (bin (string-append out "/bin"))
3328 (wrapper (string-append bin "/minced")))
3329 ;; Minced comes with a wrapper script that tries to figure out where
3330 ;; it is located before running the JAR. Since these paths are known
3331 ;; to us, we build our own wrapper to avoid coreutils dependency.
3332 (install-file "minced.jar" bin)
3333 (with-output-to-file wrapper
3334 (lambda _
3335 (display
3336 (string-append
3337 "#!" (assoc-ref inputs "bash") "/bin/sh\n\n"
3338 (assoc-ref inputs "jre") "/bin/java -jar "
3339 bin "/minced.jar \"$@\"\n"))))
3340 (chmod wrapper #o555)))))))
3341 (native-inputs
3342 `(("jdk" ,icedtea "jdk")))
3343 (inputs
3344 `(("bash" ,bash)
3345 ("jre" ,icedtea "out")))
3346 (home-page "https://github.com/ctSkennerton/minced")
3347 (synopsis "Mining CRISPRs in Environmental Datasets")
3348 (description
3349 "MinCED is a program to find Clustered Regularly Interspaced Short
3350Palindromic Repeats (CRISPRs) in DNA sequences. It can be used for
3351unassembled metagenomic reads, but is mainly designed for full genomes and
3352assembled metagenomic sequence.")
3353 (license license:gpl3+)))
3354
ddd82e0e
RW		 3355(define-public miso
3356 (package
3357 (name "miso")
3358 (version "0.5.3")
3359 (source (origin
3360 (method url-fetch)
3361 (uri (string-append
86517de6 3362 "https://pypi.python.org/packages/source/m/misopy/misopy-"
ddd82e0e
RW		 3363 version ".tar.gz"))
3364 (sha256
3365 (base32
3366 "0x446867az8ir0z8c1vjqffkp0ma37wm4sylixnkhgawllzx8v5w"))
3367 (modules '((guix build utils)))
ddd82e0e
RW		 3368 (snippet
3369 '(substitute* "setup.py"
0ec8b206
RW		 3370 ;; Use setuptools, or else the executables are not
3371 ;; installed.
3372 (("distutils.core") "setuptools")
3373 ;; use "gcc" instead of "cc" for compilation
ddd82e0e
RW		 3374 (("^defines")
3375 "cc.set_executables(
3376compiler='gcc',
3377compiler_so='gcc',
3378linker_exe='gcc',
3379linker_so='gcc -shared'); defines")))))
3380 (build-system python-build-system)
3381 (arguments
3382 `(#:python ,python-2 ; only Python 2 is supported
3383 #:tests? #f)) ; no "test" target
3384 (inputs
3385 `(("samtools" ,samtools)
3386 ("python-numpy" ,python2-numpy)
3387 ("python-pysam" ,python2-pysam)
3388 ("python-scipy" ,python2-scipy)
3389 ("python-matplotlib" ,python2-matplotlib)))
3390 (native-inputs
f3b98f4f
HG		 3391 `(("python-mock" ,python2-mock) ;for tests
3392 ("python-pytz" ,python2-pytz))) ;for tests
ddd82e0e
RW		 3393 (home-page "http://genes.mit.edu/burgelab/miso/index.html")
3394 (synopsis "Mixture of Isoforms model for RNA-Seq isoform quantitation")
3395 (description
3396 "MISO (Mixture-of-Isoforms) is a probabilistic framework that quantitates
3397the expression level of alternatively spliced genes from RNA-Seq data, and
3398identifies differentially regulated isoforms or exons across samples. By
3399modeling the generative process by which reads are produced from isoforms in
3400RNA-Seq, the MISO model uses Bayesian inference to compute the probability
3401that a read originated from a particular isoform.")
3402 (license license:gpl2)))
3403
324efb88
BW		 3404(define-public muscle
3405 (package
3406 (name "muscle")
3407 (version "3.8.1551")
3408 (source (origin
3409 (method url-fetch/tarbomb)
3410 (file-name (string-append name "-" version))
3411 (uri (string-append
3412 "http://www.drive5.com/muscle/muscle_src_"
3413 version ".tar.gz"))
3414 (sha256
3415 (base32
3416 "0bj8kj7sdizy3987zx6w7axihk40fk8rn76mpbqqjcnd64i5a367"))))
3417 (build-system gnu-build-system)
3418 (arguments
3419 `(#:make-flags (list "LDLIBS = -lm")
3420 #:phases
3421 (modify-phases %standard-phases
3422 (delete 'configure)
3423 (replace 'check
3424 ;; There are no tests, so just test if it runs.
3425 (lambda _ (zero? (system* "./muscle" "-version"))))
3426 (replace 'install
3427 (lambda* (#:key outputs #:allow-other-keys)
3428 (let* ((out (assoc-ref outputs "out"))
3429 (bin (string-append out "/bin")))
3430 (install-file "muscle" bin)))))))
3431 (home-page "http://www.drive5.com/muscle")
3432 (synopsis "Multiple sequence alignment program")
3433 (description
3434 "MUSCLE aims to be a fast and accurate multiple sequence alignment
3435program for nucleotide and protein sequences.")
3436 ;; License information found in 'muscle -h' and usage.cpp.
3437 (license license:public-domain)))
3438
99268755
BW		 3439(define-public newick-utils
3440 ;; There are no recent releases so we package from git.
3441 (let ((commit "da121155a977197cab9fbb15953ca1b40b11eb87"))
3442 (package
3443 (name "newick-utils")
3444 (version (string-append "1.6-1." (string-take commit 8)))
3445 (source (origin
3446 (method git-fetch)
3447 (uri (git-reference
3448 (url "https://github.com/tjunier/newick_utils.git")
3449 (commit commit)))
3450 (file-name (string-append name "-" version "-checkout"))
3451 (sha256
3452 (base32
3453 "1hkw21rq1mwf7xp0rmbb2gqc0i6p11108m69i7mr7xcjl268pxnb"))))
3454 (build-system gnu-build-system)
3455 (arguments
3456 `(#:phases
3457 (modify-phases %standard-phases
3458 (add-after 'unpack 'autoconf
3459 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
3460 (inputs
3461 ;; XXX: TODO: Enable Lua and Guile bindings.
3462 ;; https://github.com/tjunier/newick_utils/issues/13
3463 `(("libxml2" ,libxml2)
3464 ("flex" ,flex)
3465 ("bison" ,bison)))
3466 (native-inputs
3467 `(("autoconf" ,autoconf)
3468 ("automake" ,automake)
3469 ("libtool" ,libtool)))
3470 (synopsis "Programs for working with newick format phylogenetic trees")
3471 (description
3472 "Newick-utils is a suite of utilities for processing phylogenetic trees
3473in Newick format. Functions include re-rooting, extracting subtrees,
3474trimming, pruning, condensing, drawing (ASCII graphics or SVG).")
3475 (home-page "https://github.com/tjunier/newick_utils")
3476 (license license:bsd-3))))
3477
1e44cf8b
BW		 3478(define-public orfm
3479 (package
3480 (name "orfm")
2df3d147 3481 (version "0.6.0")
1e44cf8b
BW		 3482 (source (origin
3483 (method url-fetch)
3484 (uri (string-append
3485 "https://github.com/wwood/OrfM/releases/download/v"
3486 version "/orfm-" version ".tar.gz"))
3487 (sha256
3488 (base32
2df3d147 3489 "0kp5p0a8wh1shxlj9arv47qyhv23s1inbvmiv7k2i5mrwx2izh79"))))
1e44cf8b
BW		 3490 (build-system gnu-build-system)
3491 (inputs `(("zlib" ,zlib)))
6b6f7d6a
BW		 3492 (native-inputs
3493 `(("ruby-bio-commandeer" ,ruby-bio-commandeer)
3494 ("ruby-rspec" ,ruby-rspec)
3495 ("ruby" ,ruby)))
1e44cf8b
BW		 3496 (synopsis "Simple and not slow open reading frame (ORF) caller")
3497 (description
6b6f7d6a 3498 "An ORF caller finds stretches of DNA that, when translated, are not
1e44cf8b
BW		 3499interrupted by stop codons. OrfM finds and prints these ORFs.")
3500 (home-page "https://github.com/wwood/OrfM")
3501 (license license:lgpl3+)))
3502
19ee9201
RW		 3503(define-public python2-pbcore
3504 (package
3505 (name "python2-pbcore")
e301bfc8 3506 (version "1.2.10")
19ee9201
RW		 3507 (source (origin
3508 (method url-fetch)
ddb83129 3509 (uri (pypi-uri "pbcore" version))
19ee9201
RW		 3510 (sha256
3511 (base32
e301bfc8 3512 "1kjmv891d6qbpp4shhhvkl02ff4q5xlpnls2513sm2cjcrs52f1i"))))
19ee9201
RW		 3513 (build-system python-build-system)
3514 (arguments `(#:python ,python-2)) ; pbcore requires Python 2.7
de5bc890 3515 (propagated-inputs
19ee9201
RW		 3516 `(("python-cython" ,python2-cython)
3517 ("python-numpy" ,python2-numpy)
3518 ("python-pysam" ,python2-pysam)
3519 ("python-h5py" ,python2-h5py)))
3520 (native-inputs
de5bc890
HG		 3521 `(("python-nose" ,python2-nose)
3522 ("python-sphinx" ,python2-sphinx)
3523 ("python-pyxb" ,python2-pyxb)))
19ee9201
RW		 3524 (home-page "http://pacificbiosciences.github.io/pbcore/")
3525 (synopsis "Library for reading and writing PacBio data files")
3526 (description
3527 "The pbcore package provides Python APIs for interacting with PacBio data
3528files and writing bioinformatics applications.")
3529 (license license:bsd-3)))
3530
c61fe02c
RW		 3531(define-public python2-warpedlmm
3532 (package
3533 (name "python2-warpedlmm")
3534 (version "0.21")
3535 (source
3536 (origin
3537 (method url-fetch)
3538 (uri (string-append
3539 "https://pypi.python.org/packages/source/W/WarpedLMM/WarpedLMM-"
3540 version ".zip"))
3541 (sha256
3542 (base32
3543 "1agfz6zqa8nc6cw47yh0s3y14gkpa9wqazwcj7mwwj3ffnw39p3j"))))
3544 (build-system python-build-system)
3545 (arguments
54c85e12 3546 `(#:python ,python-2)) ; requires Python 2.7
c61fe02c
RW		 3547 (propagated-inputs
3548 `(("python-scipy" ,python2-scipy)
3549 ("python-numpy" ,python2-numpy)
3550 ("python-matplotlib" ,python2-matplotlib)
3551 ("python-fastlmm" ,python2-fastlmm)
3552 ("python-pandas" ,python2-pandas)
3553 ("python-pysnptools" ,python2-pysnptools)))
3554 (native-inputs
f3b98f4f 3555 `(("python-mock" ,python2-mock)
c61fe02c
RW		 3556 ("python-nose" ,python2-nose)
3557 ("unzip" ,unzip)))
3558 (home-page "https://github.com/PMBio/warpedLMM")
3559 (synopsis "Implementation of warped linear mixed models")
3560 (description
3561 "WarpedLMM is a Python implementation of the warped linear mixed model,
3562which automatically learns an optimal warping function (or transformation) for
3563the phenotype as it models the data.")
3564 (license license:asl2.0)))
3565
2c16316e 3566(define-public pbtranscript-tofu
698bd297 3567 (let ((commit "8f5467fe6a4472bcfb4226c8720993c8507adfe4"))
2c16316e
RW		 3568 (package
3569 (name "pbtranscript-tofu")
698bd297 3570 (version (string-append "2.2.3." (string-take commit 7)))
2c16316e
RW		 3571 (source (origin
3572 (method git-fetch)
3573 (uri (git-reference
3574 (url "https://github.com/PacificBiosciences/cDNA_primer.git")
3575 (commit commit)))
9a067efd 3576 (file-name (string-append name "-" version "-checkout"))
2c16316e
RW		 3577 (sha256
3578 (base32
9a067efd
RW		 3579 "1lgnpi35ihay42qx0b6yl3kkgra723i413j33kvs0kvs61h82w0f"))
3580 (modules '((guix build utils)))
3581 (snippet
3582 '(begin
3583 ;; remove bundled Cython sources
3584 (delete-file "pbtranscript-tofu/pbtranscript/Cython-0.20.1.tar.gz")
3585 #t))))
2c16316e
RW		 3586 (build-system python-build-system)
3587 (arguments
3588 `(#:python ,python-2
2c16316e 3589 #:phases
9a067efd
RW		 3590 (modify-phases %standard-phases
3591 (add-after 'unpack 'enter-directory
3592 (lambda _
3593 (chdir "pbtranscript-tofu/pbtranscript/")
3594 #t))
3595 ;; With setuptools version 18.0 and later this setup.py hack causes
3596 ;; a build error, so we disable it.
3597 (add-after 'enter-directory 'patch-setuppy
3598 (lambda _
3599 (substitute* "setup.py"
3600 (("if 'setuptools.extension' in sys.modules:")
3601 "if False:"))
3602 #t)))))
2c16316e 3603 (inputs
9a067efd 3604 `(("python-numpy" ,python2-numpy)
2c16316e 3605 ("python-bx-python" ,python2-bx-python)
c5372108
RW		 3606 ("python-networkx" ,python2-networkx)
3607 ("python-scipy" ,python2-scipy)
9a067efd
RW		 3608 ("python-pbcore" ,python2-pbcore)
3609 ("python-h5py" ,python2-h5py)))
2c16316e 3610 (native-inputs
9a067efd 3611 `(("python-cython" ,python2-cython)
f3b98f4f 3612 ("python-nose" ,python2-nose)))
2c16316e
RW		 3613 (home-page "https://github.com/PacificBiosciences/cDNA_primer")
3614 (synopsis "Analyze transcriptome data generated with the Iso-Seq protocol")
3615 (description
3616 "pbtranscript-tofu contains scripts to analyze transcriptome data
3617generated using the PacBio Iso-Seq protocol.")
3618 (license license:bsd-3))))
3619
024130d2
BW		 3620(define-public prank
3621 (package
3622 (name "prank")
3623 (version "150803")
3624 (source (origin
3625 (method url-fetch)
3626 (uri (string-append
3627 "http://wasabiapp.org/download/prank/prank.source."
3628 version ".tgz"))
3629 (sha256
3630 (base32
3631 "0am4z94fs3w2n5xpfls9zda61vq7qqz4q2i7b9hlsxz5q4j3kfm4"))))
3632 (build-system gnu-build-system)
3633 (arguments
3634 `(#:phases
3635 (modify-phases %standard-phases
3636 (add-after 'unpack 'enter-src-dir
3637 (lambda _
3638 (chdir "src")
3639 #t))
62d00095
EF		 3640 (add-after 'unpack 'remove-m64-flag
3641 ;; Prank will build with the correct 'bit-ness' without this flag
3642 ;; and this allows building on 32-bit machines.
3643 (lambda _ (substitute* "src/Makefile"
3644 (("-m64") ""))
3645 #t))
024130d2
BW		 3646 (delete 'configure)
3647 (replace 'install
3648 (lambda* (#:key outputs #:allow-other-keys)
3649 (let* ((out (assoc-ref outputs "out"))
3650 (bin (string-append out "/bin"))
3651 (man (string-append out "/share/man/man1"))
3652 (path (string-append
3653 (assoc-ref %build-inputs "mafft") "/bin:"
3654 (assoc-ref %build-inputs "exonerate") "/bin:"
3655 (assoc-ref %build-inputs "bppsuite") "/bin")))
3656 (install-file "prank" bin)
3657 (wrap-program (string-append bin "/prank")
3658 `("PATH" ":" prefix (,path)))
3659 (install-file "prank.1" man))
3660 #t)))))
3661 (inputs
3662 `(("mafft" ,mafft)
3663 ("exonerate" ,exonerate)
3664 ("bppsuite" ,bppsuite)))
3665 (home-page "http://wasabiapp.org/software/prank/")
3666 (synopsis "Probabilistic multiple sequence alignment program")
3667 (description
3668 "PRANK is a probabilistic multiple sequence alignment program for DNA,
3669codon and amino-acid sequences. It is based on a novel algorithm that treats
3670insertions correctly and avoids over-estimation of the number of deletion
3671events. In addition, PRANK borrows ideas from maximum likelihood methods used
3672in phylogenetics and correctly takes into account the evolutionary distances
3673between sequences. Lastly, PRANK allows for defining a potential structure
3674for sequences to be aligned and then, simultaneously with the alignment,
3675predicts the locations of structural units in the sequences.")
3676 (license license:gpl2+)))
3677
31a9d653
BW		 3678(define-public proteinortho
3679 (package
3680 (name "proteinortho")
3681 (version "5.15")
3682 (source
3683 (origin
3684 (method url-fetch)
3685 (uri
3686 (string-append
3687 "http://www.bioinf.uni-leipzig.de/Software/proteinortho/proteinortho_v"
3688 version "_src.tar.gz"))
3689 (sha256
3690 (base32
3691 "05wacnnbx56avpcwhzlcf6b7s77swcpv3qnwz5sh1z54i51gg2ki"))))
3692 (build-system gnu-build-system)
3693 (arguments
3694 `(#:test-target "test"
3695 #:phases
3696 (modify-phases %standard-phases
3697 (replace 'configure
3698 ;; There is no configure script, so we modify the Makefile directly.
3699 (lambda* (#:key outputs #:allow-other-keys)
3700 (substitute* "Makefile"
3701 (("INSTALLDIR=.*")
3702 (string-append
3703 "INSTALLDIR=" (assoc-ref outputs "out") "/bin\n")))
3704 #t))
3705 (add-before 'install 'make-install-directory
3706 ;; The install directory is not created during 'make install'.
3707 (lambda* (#:key outputs #:allow-other-keys)
3708 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
3709 #t))
3710 (add-after 'install 'wrap-programs
3711 (lambda* (#:key inputs outputs #:allow-other-keys)
3712 (let* ((path (getenv "PATH"))
3713 (out (assoc-ref outputs "out"))
3714 (binary (string-append out "/bin/proteinortho5.pl")))
3715 (wrap-program binary `("PATH" ":" prefix (,path))))
3716 #t)))))
3717 (inputs
3718 `(("perl" ,perl)
3719 ("python" ,python-2)
3720 ("blast+" ,blast+)))
3721 (home-page "http://www.bioinf.uni-leipzig.de/Software/proteinortho")
3722 (synopsis "Detect orthologous genes across species")
3723 (description
3724 "Proteinortho is a tool to detect orthologous genes across different
3725species. For doing so, it compares similarities of given gene sequences and
3726clusters them to find significant groups. The algorithm was designed to handle
3727large-scale data and can be applied to hundreds of species at once.")
3728 (license license:gpl2+)))
3729
846e3409
RW		 3730(define-public pyicoteo
3731 (package
3732 (name "pyicoteo")
3733 (version "2.0.7")
3734 (source
3735 (origin
3736 (method url-fetch)
3737 (uri (string-append "https://bitbucket.org/regulatorygenomicsupf/"
3738 "pyicoteo/get/v" version ".tar.bz2"))
3739 (file-name (string-append name "-" version ".tar.bz2"))
3740 (sha256
3741 (base32
3742 "0d6087f29xp8wxwlj111c3sylli98n0l8ry58c51ixzq0zfm50wa"))))
3743 (build-system python-build-system)
3744 (arguments
3745 `(#:python ,python-2 ; does not work with Python 3
3746 #:tests? #f)) ; there are no tests
3747 (inputs
3748 `(("python2-matplotlib" ,python2-matplotlib)))
3749 (home-page "https://bitbucket.org/regulatorygenomicsupf/pyicoteo")
3750 (synopsis "Analyze high-throughput genetic sequencing data")
3751 (description
3752 "Pyicoteo is a suite of tools for the analysis of high-throughput genetic
3753sequencing data. It works with genomic coordinates. There are currently six
3754different command-line tools:
3755
3756@enumerate
3757@item pyicoregion: for generating exploratory regions automatically;
3758@item pyicoenrich: for differential enrichment between two conditions;
3759@item pyicoclip: for calling CLIP-Seq peaks without a control;
3760@item pyicos: for genomic coordinates manipulation;
3761@item pyicoller: for peak calling on punctuated ChIP-Seq;
3762@item pyicount: to count how many reads from N experiment files overlap in a
3763 region file;
3764@item pyicotrocol: to combine operations from pyicoteo.
3765@end enumerate\n")
3766 (license license:gpl3+)))
3767
af860475
BW		 3768(define-public prodigal
3769 (package
3770 (name "prodigal")
e70f7a23 3771 (version "2.6.3")
af860475
BW		 3772 (source (origin
3773 (method url-fetch)
3774 (uri (string-append
3775 "https://github.com/hyattpd/Prodigal/archive/v"
3776 version ".tar.gz"))
3777 (file-name (string-append name "-" version ".tar.gz"))
3778 (sha256
3779 (base32
e70f7a23 3780 "17srxkqd3jc77xk15pfbgg1a9xahqg7337w95mrsia7mpza4l2c9"))))
af860475
BW		 3781 (build-system gnu-build-system)
3782 (arguments
3783 `(#:tests? #f ;no check target
3784 #:make-flags (list (string-append "INSTALLDIR="
3785 (assoc-ref %outputs "out")
3786 "/bin"))
3787 #:phases
3788 (modify-phases %standard-phases
3789 (delete 'configure))))
3790 (home-page "http://prodigal.ornl.gov")
3791 (synopsis "Protein-coding gene prediction for Archaea and Bacteria")
3792 (description
3793 "Prodigal runs smoothly on finished genomes, draft genomes, and
3794metagenomes, providing gene predictions in GFF3, Genbank, or Sequin table
3795format. It runs quickly, in an unsupervised fashion, handles gaps, handles
3796partial genes, and identifies translation initiation sites.")
3797 (license license:gpl3+)))
608dd932 3798
ceb62d54
BW		 3799(define-public roary
3800 (package
3801 (name "roary")
5e0a0f42 3802 (version "3.7.0")
ceb62d54
BW		 3803 (source
3804 (origin
3805 (method url-fetch)
3806 (uri (string-append
3807 "mirror://cpan/authors/id/A/AJ/AJPAGE/Bio-Roary-"
3808 version ".tar.gz"))
3809 (sha256
3810 (base32
5e0a0f42 3811 "0x2hpb3nfsc6x2nq1788w0fhqfzc7cn2dp4xwyva9m3k6xlz0m43"))))
ceb62d54
BW		 3812 (build-system perl-build-system)
3813 (arguments
3814 `(#:phases
3815 (modify-phases %standard-phases
3816 (delete 'configure)
3817 (delete 'build)
3818 (replace 'check
3819 (lambda _
3820 ;; The tests are not run by default, so we run each test file
3821 ;; directly.
3822 (setenv "PATH" (string-append (getcwd) "/bin" ":"
3823 (getenv "PATH")))
3824 (setenv "PERL5LIB" (string-append (getcwd) "/lib" ":"
3825 (getenv "PERL5LIB")))
3826 (zero? (length (filter (lambda (file)
3827 (display file)(display "\n")
3828 (not (zero? (system* "perl" file))))
3829 (find-files "t" ".*\\.t$"))))))
3830 (replace 'install
3831 ;; There is no 'install' target in the Makefile.
3832 (lambda* (#:key outputs #:allow-other-keys)
3833 (let* ((out (assoc-ref outputs "out"))
3834 (bin (string-append out "/bin"))
3835 (perl (string-append out "/lib/perl5/site_perl"))
3836 (roary-plots "contrib/roary_plots"))
3837 (mkdir-p bin)
3838 (mkdir-p perl)
3839 (copy-recursively "bin" bin)
3840 (copy-recursively "lib" perl)
3841 #t)))
3842 (add-after 'install 'wrap-programs
3843 (lambda* (#:key inputs outputs #:allow-other-keys)
3844 (let* ((out (assoc-ref outputs "out"))
3845 (perl5lib (getenv "PERL5LIB"))
3846 (path (getenv "PATH")))
3847 (for-each (lambda (prog)
3848 (let ((binary (string-append out "/" prog)))
3849 (wrap-program binary
3850 `("PERL5LIB" ":" prefix
3851 (,(string-append perl5lib ":" out
3852 "/lib/perl5/site_perl"))))
3853 (wrap-program binary
3854 `("PATH" ":" prefix
3855 (,(string-append path ":" out "/bin"))))))
3856 (find-files "bin" ".*[^R]$"))
3857 (let ((file
3858 (string-append out "/bin/roary-create_pan_genome_plots.R"))
3859 (r-site-lib (getenv "R_LIBS_SITE"))
3860 (coreutils-path
3861 (string-append (assoc-ref inputs "coreutils") "/bin")))
3862 (wrap-program file
3863 `("R_LIBS_SITE" ":" prefix
3864 (,(string-append r-site-lib ":" out "/site-library/"))))
3865 (wrap-program file
3866 `("PATH" ":" prefix
3867 (,(string-append coreutils-path ":" out "/bin"))))))
3868 #t)))))
3869 (native-inputs
3870 `(("perl-env-path" ,perl-env-path)
3871 ("perl-test-files" ,perl-test-files)
3872 ("perl-test-most" ,perl-test-most)
3873 ("perl-test-output" ,perl-test-output)))
3874 (inputs
3875 `(("perl-array-utils" ,perl-array-utils)
3876 ("bioperl" ,bioperl-minimal)
3877 ("perl-exception-class" ,perl-exception-class)
3878 ("perl-file-find-rule" ,perl-file-find-rule)
3879 ("perl-file-grep" ,perl-file-grep)
3880 ("perl-file-slurper" ,perl-file-slurper)
3881 ("perl-file-which" ,perl-file-which)
3882 ("perl-graph" ,perl-graph)
3883 ("perl-graph-readwrite" ,perl-graph-readwrite)
3884 ("perl-log-log4perl" ,perl-log-log4perl)
3885 ("perl-moose" ,perl-moose)
3886 ("perl-perlio-utf8_strict" ,perl-perlio-utf8_strict)
3887 ("perl-text-csv" ,perl-text-csv)
3888 ("bedtools" ,bedtools)
3889 ("cd-hit" ,cd-hit)
3890 ("blast+" ,blast+)
3891 ("mcl" ,mcl)
3892 ("parallel" ,parallel)
3893 ("prank" ,prank)
3894 ("mafft" ,mafft)
3895 ("fasttree" ,fasttree)
3896 ("grep" ,grep)
3897 ("sed" ,sed)
3898 ("gawk" ,gawk)
3899 ("r" ,r)
3900 ("r-ggplot2" ,r-ggplot2)
3901 ("coreutils" ,coreutils)))
3902 (home-page "http://sanger-pathogens.github.io/Roary")
3903 (synopsis "High speed stand-alone pan genome pipeline")
3904 (description
3905 "Roary is a high speed stand alone pan genome pipeline, which takes
3906annotated assemblies in GFF3 format (produced by the Prokka program) and
3907calculates the pan genome. Using a standard desktop PC, it can analyse
3908datasets with thousands of samples, without compromising the quality of the
3909results. 128 samples can be analysed in under 1 hour using 1 GB of RAM and a
3910single processor. Roary is not intended for metagenomics or for comparing
3911extremely diverse sets of genomes.")
3912 (license license:gpl3)))
3913
608dd932
BW		 3914(define-public raxml
3915 (package
3916 (name "raxml")
3917 (version "8.2.9")
3918 (source
3919 (origin
3920 (method url-fetch)
3921 (uri
3922 (string-append
3923 "https://github.com/stamatak/standard-RAxML/archive/v"
3924 version ".tar.gz"))
3925 (file-name (string-append name "-" version ".tar.gz"))
3926 (sha256
3927 (base32
3928 "1pv8p2fy67y21a9y4cm7xpvxqjwz2v4201flfjshdq1p8j52rqf7"))))
3929 (build-system gnu-build-system)
3930 (arguments
3931 `(#:tests? #f ; There are no tests.
3932 ;; Use 'standard' Makefile rather than SSE or AVX ones.
3933 #:make-flags (list "-f" "Makefile.HYBRID.gcc")
3934 #:phases
3935 (modify-phases %standard-phases
3936 (delete 'configure)
3937 (replace 'install
3938 (lambda* (#:key outputs #:allow-other-keys)
3939 (let* ((out (assoc-ref outputs "out"))
3940 (bin (string-append out "/bin"))
3941 (executable "raxmlHPC-HYBRID"))
3942 (install-file executable bin)
3943 (symlink (string-append bin "/" executable) "raxml"))
3944 #t)))))
3945 (inputs
3946 `(("openmpi" ,openmpi)))
3947 (home-page "http://sco.h-its.org/exelixis/web/software/raxml/index.html")
3948 (synopsis "Randomized Axelerated Maximum Likelihood phylogenetic trees")
3949 (description
3950 "RAxML is a tool for phylogenetic analysis and post-analysis of large
3951phylogenies.")
3952 (license license:gpl2+)))
af860475 3953
66e3eff1
RW		 3954(define-public rsem
3955 (package
3956 (name "rsem")
3957 (version "1.2.20")
3958 (source
3959 (origin
3960 (method url-fetch)
3961 (uri
3962 (string-append "http://deweylab.biostat.wisc.edu/rsem/src/rsem-"
3963 version ".tar.gz"))
3964 (sha256
3965 (base32 "0nzdc0j0hjllhsd5f2xli95dafm3nawskigs140xzvjk67xh0r9q"))
fc1adab1 3966 (patches (search-patches "rsem-makefile.patch"))
66e3eff1
RW		 3967 (modules '((guix build utils)))
3968 (snippet
3969 '(begin
3970 ;; remove bundled copy of boost
3971 (delete-file-recursively "boost")
3972 #t))))
3973 (build-system gnu-build-system)
3974 (arguments
3975 `(#:tests? #f ;no "check" target
3976 #:phases
3977 (modify-phases %standard-phases
3978 ;; No "configure" script.
3979 ;; Do not build bundled samtools library.
3980 (replace 'configure
3981 (lambda _
3982 (substitute* "Makefile"
3983 (("^all : sam/libbam.a") "all : "))
3984 #t))
3985 (replace 'install
3986 (lambda* (#:key outputs #:allow-other-keys)
3987 (let* ((out (string-append (assoc-ref outputs "out")))
3988 (bin (string-append out "/bin/"))
3989 (perl (string-append out "/lib/perl5/site_perl")))
3990 (mkdir-p bin)
3991 (mkdir-p perl)
3992 (for-each (lambda (file)
3993 (copy-file file
3994 (string-append bin (basename file))))
3995 (find-files "." "rsem-.*"))
3996 (copy-file "rsem_perl_utils.pm"
3997 (string-append perl "/rsem_perl_utils.pm")))
3998 #t))
3999 (add-after
4000 'install 'wrap-program
4001 (lambda* (#:key outputs #:allow-other-keys)
4002 (let ((out (assoc-ref outputs "out")))
4003 (for-each (lambda (prog)
4004 (wrap-program (string-append out "/bin/" prog)
4005 `("PERL5LIB" ":" prefix
4006 (,(string-append out "/lib/perl5/site_perl")))))
4007 '("rsem-plot-transcript-wiggles"
4008 "rsem-calculate-expression"
4009 "rsem-generate-ngvector"
4010 "rsem-run-ebseq"
4011 "rsem-prepare-reference")))
4012 #t)))))
4013 (inputs
4014 `(("boost" ,boost)
4015 ("ncurses" ,ncurses)
4016 ("r" ,r)
4017 ("perl" ,perl)
4018 ("samtools" ,samtools-0.1)
4019 ("zlib" ,zlib)))
4020 (home-page "http://deweylab.biostat.wisc.edu/rsem/")
4021 (synopsis "Estimate gene expression levels from RNA-Seq data")
4022 (description
4023 "RSEM is a software package for estimating gene and isoform expression
4024levels from RNA-Seq data. The RSEM package provides a user-friendly
4025interface, supports threads for parallel computation of the EM algorithm,
4026single-end and paired-end read data, quality scores, variable-length reads and
4027RSPD estimation. In addition, it provides posterior mean and 95% credibility
4028interval estimates for expression levels. For visualization, it can generate
4029BAM and Wiggle files in both transcript-coordinate and genomic-coordinate.")
4030 (license license:gpl3+)))
4031
8622a072
RW		 4032(define-public rseqc
4033 (package
4034 (name "rseqc")
4035 (version "2.6.1")
4036 (source
4037 (origin
4038 (method url-fetch)
4039 (uri
4040 (string-append "mirror://sourceforge/rseqc/"
de67e922 4041 "RSeQC-" version ".tar.gz"))
8622a072 4042 (sha256
8214b7fb 4043 (base32 "15ly0254yi032qzkdplg00q144qfdsd986gh62829rl5bkxhj330"))
8622a072
RW		 4044 (modules '((guix build utils)))
4045 (snippet
4046 '(begin
4047 ;; remove bundled copy of pysam
4048 (delete-file-recursively "lib/pysam")
4049 (substitute* "setup.py"
4050 ;; remove dependency on outdated "distribute" module
4051 (("^from distribute_setup import use_setuptools") "")
4052 (("^use_setuptools\\(\\)") "")
4053 ;; do not use bundled copy of pysam
4054 (("^have_pysam = False") "have_pysam = True"))))))
4055 (build-system python-build-system)
4056 (arguments `(#:python ,python-2))
4057 (inputs
4058 `(("python-cython" ,python2-cython)
4059 ("python-pysam" ,python2-pysam)
4060 ("python-numpy" ,python2-numpy)
8622a072
RW		 4061 ("zlib" ,zlib)))
4062 (native-inputs
4063 `(("python-nose" ,python2-nose)))
4064 (home-page "http://rseqc.sourceforge.net/")
4065 (synopsis "RNA-seq quality control package")
4066 (description
4067 "RSeQC provides a number of modules that can comprehensively evaluate
4068high throughput sequence data, especially RNA-seq data. Some basic modules
4069inspect sequence quality, nucleotide composition bias, PCR bias and GC bias,
4070while RNA-seq specific modules evaluate sequencing saturation, mapped reads
4071distribution, coverage uniformity, strand specificity, etc.")
4072 (license license:gpl3+)))
4073
ec946638
RW		 4074(define-public seek
4075 ;; There are no release tarballs. According to the installation
4076 ;; instructions at http://seek.princeton.edu/installation.jsp, the latest
4077 ;; stable release is identified by this changeset ID.
4078 (let ((changeset "2329130")
4079 (revision "1"))
4080 (package
4081 (name "seek")
4082 (version (string-append "0-" revision "." changeset))
4083 (source (origin
4084 (method hg-fetch)
4085 (uri (hg-reference
4086 (url "https://bitbucket.org/libsleipnir/sleipnir")
4087 (changeset changeset)))
4088 (sha256
4089 (base32
4090 "0qrvilwh18dpbhkf92qvxbmay0j75ra3jg2wrhz67gf538zzphsx"))))
4091 (build-system gnu-build-system)
4092 (arguments
4093 `(#:modules ((srfi srfi-1)
4094 (guix build gnu-build-system)
4095 (guix build utils))
4096 #:phases
4097 (let ((dirs '("SeekMiner"
4098 "SeekEvaluator"
4099 "SeekPrep"
4100 "Distancer"
4101 "Data2DB"
4102 "PCL2Bin")))
4103 (modify-phases %standard-phases
4104 (add-before 'configure 'bootstrap
4105 (lambda _
4106 (zero? (system* "bash" "gen_auto"))))
4107 (add-after 'build 'build-additional-tools
4108 (lambda* (#:key make-flags #:allow-other-keys)
4109 (every (lambda (dir)
4110 (with-directory-excursion (string-append "tools/" dir)
4111 (zero? (apply system* "make" make-flags))))
4112 dirs)))
4113 (add-after 'install 'install-additional-tools
4114 (lambda* (#:key make-flags #:allow-other-keys)
4115 (fold (lambda (dir result)
4116 (with-directory-excursion (string-append "tools/" dir)
4117 (and result
4118 (zero? (apply system*
4119 `("make" ,@make-flags "install"))))))
4120 #t dirs)))))))
4121 (inputs
4122 `(("gsl" ,gsl)
4123 ("boost" ,boost)
4124 ("libsvm" ,libsvm)
4125 ("readline" ,readline)
4126 ("gengetopt" ,gengetopt)
4127 ("log4cpp" ,log4cpp)))
4128 (native-inputs
4129 `(("autoconf" ,autoconf)
4130 ("automake" ,automake)
4131 ("perl" ,perl)))
4132 (home-page "http://seek.princeton.edu")
4133 (synopsis "Gene co-expression search engine")
4134 (description
4135 "SEEK is a computational gene co-expression search engine. SEEK provides
4136biologists with a way to navigate the massive human expression compendium that
4137now contains thousands of expression datasets. SEEK returns a robust ranking
4138of co-expressed genes in the biological area of interest defined by the user's
4139query genes. It also prioritizes thousands of expression datasets according
4140to the user's query of interest.")
4141 (license license:cc-by3.0))))
4142
4e10a221
RW		 4143(define-public samtools
4144 (package
4145 (name "samtools")
79b555ed 4146 (version "1.3.1")
4e10a221
RW		 4147 (source
4148 (origin
4149 (method url-fetch)
4150 (uri
de67e922 4151 (string-append "mirror://sourceforge/samtools/samtools/"
4e10a221
RW		 4152 version "/samtools-" version ".tar.bz2"))
4153 (sha256
4154 (base32
79b555ed 4155 "0znnnxc467jbf1as2dpskrjhfh8mbll760j6w6rdkwlwbqsp8gbc"))))
4e10a221
RW		 4156 (build-system gnu-build-system)
4157 (arguments
c4473411 4158 `(#:modules ((ice-9 ftw)
5bdda30b
RW		 4159 (ice-9 regex)
4160 (guix build gnu-build-system)
4161 (guix build utils))
c4473411
RW		 4162 #:make-flags (list (string-append "prefix=" (assoc-ref %outputs "out")))
4163 #:configure-flags (list "--with-ncurses")
4e10a221
RW		 4164 #:phases
4165 (alist-cons-after
c4473411
RW		 4166 'unpack 'patch-tests
4167 (lambda _
4168 (substitute* "test/test.pl"
4169 ;; The test script calls out to /bin/bash
4170 (("/bin/bash") (which "bash")))
4171 #t)
41dd7126
RW		 4172 (alist-cons-after
4173 'install 'install-library
4174 (lambda* (#:key outputs #:allow-other-keys)
4175 (let ((lib (string-append (assoc-ref outputs "out") "/lib")))
96c46210 4176 (install-file "libbam.a" lib)))
5bdda30b
RW		 4177 (alist-cons-after
4178 'install 'install-headers
4179 (lambda* (#:key outputs #:allow-other-keys)
4180 (let ((include (string-append (assoc-ref outputs "out")
4181 "/include/samtools/")))
5bdda30b 4182 (for-each (lambda (file)
96c46210 4183 (install-file file include))
5bdda30b
RW		 4184 (scandir "." (lambda (name) (string-match "\\.h$" name))))
4185 #t))
c4473411 4186 %standard-phases)))))
4e10a221
RW		 4187 (native-inputs `(("pkg-config" ,pkg-config)))
4188 (inputs `(("ncurses" ,ncurses)
4189 ("perl" ,perl)
4190 ("python" ,python)
4191 ("zlib" ,zlib)))
4192 (home-page "http://samtools.sourceforge.net")
4193 (synopsis "Utilities to efficiently manipulate nucleotide sequence alignments")
4194 (description
4195 "Samtools implements various utilities for post-processing nucleotide
4196sequence alignments in the SAM, BAM, and CRAM formats, including indexing,
4197variant calling (in conjunction with bcftools), and a simple alignment
4198viewer.")
4199 (license license:expat)))
d3517eda 4200
0b84a0aa
RW		 4201(define-public samtools-0.1
4202 ;; This is the most recent version of the 0.1 line of samtools. The input
4203 ;; and output formats differ greatly from that used and produced by samtools
4204 ;; 1.x and is still used in many bioinformatics pipelines.
4205 (package (inherit samtools)
4206 (version "0.1.19")
4207 (source
4208 (origin
4209 (method url-fetch)
4210 (uri
de67e922 4211 (string-append "mirror://sourceforge/samtools/samtools/"
0b84a0aa
RW		 4212 version "/samtools-" version ".tar.bz2"))
4213 (sha256
4214 (base32 "1m33xsfwz0s8qi45lylagfllqg7fphf4dr0780rsvw75av9wk06h"))))
4215 (arguments
2309ed68
RW		 4216 `(#:tests? #f ;no "check" target
4217 ,@(substitute-keyword-arguments (package-arguments samtools)
4218 ((#:make-flags flags)
4219 `(cons "LIBCURSES=-lncurses" ,flags))
4220 ((#:phases phases)
4221 `(modify-phases ,phases
4222 (replace 'install
4223 (lambda* (#:key outputs #:allow-other-keys)
4224 (let ((bin (string-append
4225 (assoc-ref outputs "out") "/bin")))
4226 (mkdir-p bin)
4227 (copy-file "samtools"
4228 (string-append bin "/samtools")))))
4229 (delete 'patch-tests)
4230 (delete 'configure))))))))
0b84a0aa 4231
fe4c37c2 4232(define-public mosaik
698bd297 4233 (let ((commit "5c25216d3522d6a33e53875cd76a6d65001e4e67"))
fe4c37c2
RW		 4234 (package
4235 (name "mosaik")
4236 (version "2.2.30")
4237 (source (origin
4238 ;; There are no release tarballs nor tags.
4239 (method git-fetch)
4240 (uri (git-reference
4241 (url "https://github.com/wanpinglee/MOSAIK.git")
4242 (commit commit)))
4243 (file-name (string-append name "-" version))
4244 (sha256
4245 (base32
4246 "17gj3s07cm77r41z92awh0bim7w7q7fbn0sf5nkqmcm1vw052qgw"))))
4247 (build-system gnu-build-system)
4248 (arguments
4249 `(#:tests? #f ; no tests
4250 #:make-flags (list "CC=gcc")
4251 #:phases
4252 (modify-phases %standard-phases
4253 (replace 'configure
4254 (lambda _ (chdir "src") #t))
4255 (replace 'install
4256 (lambda* (#:key outputs #:allow-other-keys)
4257 (let ((bin (string-append (assoc-ref outputs "out")
4258 "/bin")))
4259 (mkdir-p bin)
4260 (copy-recursively "../bin" bin)
4261 #t))))))
4262 (inputs
4263 `(("perl" ,perl)
4264 ("zlib" ,zlib)))
029d9f77 4265 (supported-systems '("x86_64-linux"))
fe4c37c2
RW		 4266 (home-page "https://code.google.com/p/mosaik-aligner/")
4267 (synopsis "Map nucleotide sequence reads to reference genomes")
4268 (description
4269 "MOSAIK is a program for mapping second and third-generation sequencing
4270reads to a reference genome. MOSAIK can align reads generated by all the
4271major sequencing technologies, including Illumina, Applied Biosystems SOLiD,
4272Roche 454, Ion Torrent and Pacific BioSciences SMRT.")
4273 ;; MOSAIK is released under the GPLv2+ with the exception of third-party
4274 ;; code released into the public domain:
4275 ;; 1. fastlz by Ariya Hidayat - http://www.fastlz.org/
4276 ;; 2. MD5 implementation - RSA Data Security, RFC 1321
4277 (license (list license:gpl2+ license:public-domain)))))
4278
282c5087
RW		 4279(define-public ngs-sdk
4280 (package
4281 (name "ngs-sdk")
48b419eb 4282 (version "1.2.5")
282c5087
RW		 4283 (source
4284 (origin
4285 (method url-fetch)
4286 (uri
4287 (string-append "https://github.com/ncbi/ngs/archive/"
4288 version ".tar.gz"))
4289 (file-name (string-append name "-" version ".tar.gz"))
4290 (sha256
4291 (base32
48b419eb 4292 "04y1fsmdnb5y86m3gg6f5g9wcscr6r25n7m8mdlcxy0i2q6w6cia"))))
282c5087
RW		 4293 (build-system gnu-build-system)
4294 (arguments
4295 `(#:parallel-build? #f ; not supported
4296 #:tests? #f ; no "check" target
4297 #:phases
4298 (alist-replace
4299 'configure
4300 (lambda* (#:key outputs #:allow-other-keys)
4301 (let ((out (assoc-ref outputs "out")))
282c5087
RW		 4302 ;; The 'configure' script doesn't recognize things like
4303 ;; '--enable-fast-install'.
4304 (zero? (system* "./configure"
4305 (string-append "--build-prefix=" (getcwd) "/build")
4306 (string-append "--prefix=" out)))))
4307 (alist-cons-after
4308 'unpack 'enter-dir
4309 (lambda _ (chdir "ngs-sdk") #t)
4310 %standard-phases))))
4311 (native-inputs `(("perl" ,perl)))
a0dadf0c
AE		 4312 ;; According to the test
4313 ;; unless ($MARCH =~ /x86_64/i || $MARCH =~ /i?86/i)
4314 ;; in ngs-sdk/setup/konfigure.perl
ab29be81 4315 (supported-systems '("i686-linux" "x86_64-linux"))
282c5087
RW		 4316 (home-page "https://github.com/ncbi/ngs")
4317 (synopsis "API for accessing Next Generation Sequencing data")
4318 (description
4319 "NGS is a domain-specific API for accessing reads, alignments and pileups
4320produced from Next Generation Sequencing. The API itself is independent from
4321any particular back-end implementation, and supports use of multiple back-ends
4322simultaneously.")
4323 (license license:public-domain)))
4324
1ad15c16 4325(define-public java-ngs
2651a5e6 4326 (package (inherit ngs-sdk)
1ad15c16 4327 (name "java-ngs")
2651a5e6
RW		 4328 (arguments
4329 `(,@(substitute-keyword-arguments
4330 `(#:modules ((guix build gnu-build-system)
4331 (guix build utils)
4332 (srfi srfi-1)
4333 (srfi srfi-26))
4334 ,@(package-arguments ngs-sdk))
4335 ((#:phases phases)
614a8977
RW		 4336 `(modify-phases ,phases
4337 (replace 'enter-dir (lambda _ (chdir "ngs-java") #t)))))))
2651a5e6 4338 (inputs
d2540f80 4339 `(("jdk" ,icedtea "jdk")
2651a5e6
RW		 4340 ("ngs-sdk" ,ngs-sdk)))
4341 (synopsis "Java bindings for NGS SDK")))
4342
75dd2424
RW		 4343(define-public ncbi-vdb
4344 (package
4345 (name "ncbi-vdb")
40974c93 4346 (version "2.7.0")
75dd2424
RW		 4347 (source
4348 (origin
4349 (method url-fetch)
4350 (uri
4351 (string-append "https://github.com/ncbi/ncbi-vdb/archive/"
4352 version ".tar.gz"))
4353 (file-name (string-append name "-" version ".tar.gz"))
4354 (sha256
4355 (base32
40974c93 4356 "0x1cg1x8vy0yjlkp0snc1533zcjhxqzqsaiwqk598n7vvw37n8lf"))))
75dd2424
RW		 4357 (build-system gnu-build-system)
4358 (arguments
4359 `(#:parallel-build? #f ; not supported
4360 #:tests? #f ; no "check" target
4361 #:phases
4362 (alist-replace
4363 'configure
4364 (lambda* (#:key inputs outputs #:allow-other-keys)
4365 (let ((out (assoc-ref outputs "out")))
75dd2424
RW		 4366 ;; Override include path for libmagic
4367 (substitute* "setup/package.prl"
4368 (("name => 'magic', Include => '/usr/include'")
4369 (string-append "name=> 'magic', Include => '"
4370 (assoc-ref inputs "libmagic")
4371 "/include" "'")))
4372
4373 ;; Install kdf5 library (needed by sra-tools)
4374 (substitute* "build/Makefile.install"
4375 (("LIBRARIES_TO_INSTALL =")
4376 "LIBRARIES_TO_INSTALL = kdf5.$(VERSION_LIBX) kdf5.$(VERSION_SHLX)"))
4377
675d7ae2
RW		 4378 (substitute* "build/Makefile.env"
4379 (("CFLAGS =" prefix)
4380 (string-append prefix "-msse2 ")))
4381
75dd2424
RW		 4382 ;; The 'configure' script doesn't recognize things like
4383 ;; '--enable-fast-install'.
4384 (zero? (system*
4385 "./configure"
4386 (string-append "--build-prefix=" (getcwd) "/build")
4387 (string-append "--prefix=" (assoc-ref outputs "out"))
4388 (string-append "--debug")
4389 (string-append "--with-xml2-prefix="
4390 (assoc-ref inputs "libxml2"))
4391 (string-append "--with-ngs-sdk-prefix="
4392 (assoc-ref inputs "ngs-sdk"))
4393 (string-append "--with-ngs-java-prefix="
1ad15c16 4394 (assoc-ref inputs "java-ngs"))
75dd2424
RW		 4395 (string-append "--with-hdf5-prefix="
4396 (assoc-ref inputs "hdf5"))))))
4397 (alist-cons-after
4398 'install 'install-interfaces
132b4c8c
RW		 4399 (lambda* (#:key outputs #:allow-other-keys)
4400 ;; Install interface libraries. On i686 the interface libraries
4401 ;; are installed to "linux/gcc/i386", so we need to use the Linux
4402 ;; architecture name ("i386") instead of the target system prefix
4403 ;; ("i686").
75dd2424
RW		 4404 (mkdir (string-append (assoc-ref outputs "out") "/ilib"))
4405 (copy-recursively (string-append "build/ncbi-vdb/linux/gcc/"
132b4c8c
RW		 4406 ,(system->linux-architecture
4407 (or (%current-target-system)
4408 (%current-system)))
75dd2424
RW		 4409 "/rel/ilib")
4410 (string-append (assoc-ref outputs "out")
4411 "/ilib"))
4412 ;; Install interface headers
4413 (copy-recursively "interfaces"
4414 (string-append (assoc-ref outputs "out")
4415 "/include")))
4416 %standard-phases))))
4417 (inputs
4418 `(("libxml2" ,libxml2)
4419 ("ngs-sdk" ,ngs-sdk)
1ad15c16 4420 ("java-ngs" ,java-ngs)
75dd2424
RW		 4421 ("libmagic" ,file)
4422 ("hdf5" ,hdf5)))
4423 (native-inputs `(("perl" ,perl)))
675d7ae2
RW		 4424 ;; NCBI-VDB requires SSE capability.
4425 (supported-systems '("i686-linux" "x86_64-linux"))
75dd2424
RW		 4426 (home-page "https://github.com/ncbi/ncbi-vdb")
4427 (synopsis "Database engine for genetic information")
4428 (description
4429 "The NCBI-VDB library implements a highly compressed columnar data
4430warehousing engine that is most often used to store genetic information.
4431Databases are stored in a portable image within the file system, and can be
4432accessed/downloaded on demand across HTTP.")
4433 (license license:public-domain)))
4434
cc6ed477
RW		 4435(define-public plink
4436 (package
4437 (name "plink")
4438 (version "1.07")
4439 (source
4440 (origin
4441 (method url-fetch)
4442 (uri (string-append
4443 "http://pngu.mgh.harvard.edu/~purcell/plink/dist/plink-"
4444 version "-src.zip"))
4445 (sha256
4446 (base32 "0as8gxm4pjyc8dxmm1sl873rrd7wn5qs0l29nqfnl31x8i467xaa"))
0dbb7ac2
EF		 4447 (patches (search-patches "plink-1.07-unclobber-i.patch"
4448 "plink-endian-detection.patch"))))
cc6ed477
RW		 4449 (build-system gnu-build-system)
4450 (arguments
4451 '(#:tests? #f ;no "check" target
4452 #:make-flags (list (string-append "LIB_LAPACK="
4453 (assoc-ref %build-inputs "lapack")
4454 "/lib/liblapack.so")
4455 "WITH_LAPACK=1"
4456 "FORCE_DYNAMIC=1"
4457 ;; disable phoning home
4458 "WITH_WEBCHECK=")
4459 #:phases
4460 (modify-phases %standard-phases
4461 ;; no "configure" script
4462 (delete 'configure)
4463 (replace 'install
4464 (lambda* (#:key outputs #:allow-other-keys)
4465 (let ((bin (string-append (assoc-ref outputs "out")
4466 "/bin/")))
96c46210 4467 (install-file "plink" bin)
cc6ed477
RW		 4468 #t))))))
4469 (inputs
4470 `(("zlib" ,zlib)
4471 ("lapack" ,lapack)))
4472 (native-inputs
4473 `(("unzip" ,unzip)))
4474 (home-page "http://pngu.mgh.harvard.edu/~purcell/plink/")
4475 (synopsis "Whole genome association analysis toolset")
4476 (description
4477 "PLINK is a whole genome association analysis toolset, designed to
4478perform a range of basic, large-scale analyses in a computationally efficient
4479manner. The focus of PLINK is purely on analysis of genotype/phenotype data,
4480so there is no support for steps prior to this (e.g. study design and
4481planning, generating genotype or CNV calls from raw data). Through
4482integration with gPLINK and Haploview, there is some support for the
4483subsequent visualization, annotation and storage of results.")
4484 ;; Code is released under GPLv2, except for fisher.h, which is under
4485 ;; LGPLv2.1+
4486 (license (list license:gpl2 license:lgpl2.1+))))
4487
c6a24d6e
RW		 4488(define-public smithlab-cpp
4489 (let ((revision "1")
698bd297 4490 (commit "728a097bec88c6f4b8528b685932049e660eff2e"))
c6a24d6e
RW		 4491 (package
4492 (name "smithlab-cpp")
698bd297 4493 (version (string-append "0." revision "." (string-take commit 7)))
c6a24d6e
RW		 4494 (source (origin
4495 (method git-fetch)
4496 (uri (git-reference
4497 (url "https://github.com/smithlabcode/smithlab_cpp.git")
4498 (commit commit)))
4499 (file-name (string-append name "-" version "-checkout"))
4500 (sha256
4501 (base32
4502 "0d476lmj312xk77kr9fzrv7z1bv96yfyx0w7y62ycmnfbx32ll74"))))
4503 (build-system gnu-build-system)
4504 (arguments
4505 `(#:modules ((guix build gnu-build-system)
4506 (guix build utils)
4507 (srfi srfi-26))
4508 #:tests? #f ;no "check" target
4509 #:phases
4510 (modify-phases %standard-phases
4511 (add-after 'unpack 'use-samtools-headers
4512 (lambda _
4513 (substitute* '("SAM.cpp"
4514 "SAM.hpp")
4515 (("sam.h") "samtools/sam.h"))
4516 #t))
4517 (replace 'install
4518 (lambda* (#:key outputs #:allow-other-keys)
4519 (let* ((out (assoc-ref outputs "out"))
4520 (lib (string-append out "/lib"))
4521 (include (string-append out "/include/smithlab-cpp")))
4522 (mkdir-p lib)
4523 (mkdir-p include)
4524 (for-each (cut install-file <> lib)
4525 (find-files "." "\\.o$"))
4526 (for-each (cut install-file <> include)
4527 (find-files "." "\\.hpp$")))
4528 #t))
4529 (delete 'configure))))
4530 (inputs
4531 `(("samtools" ,samtools-0.1)
4532 ("zlib" ,zlib)))
4533 (home-page "https://github.com/smithlabcode/smithlab_cpp")
4534 (synopsis "C++ helper library for functions used in Smith lab projects")
4535 (description
4536 "Smithlab CPP is a C++ library that includes functions used in many of
4537the Smith lab bioinformatics projects, such as a wrapper around Samtools data
4538structures, classes for genomic regions, mapped sequencing reads, etc.")
4539 (license license:gpl3+))))
4540
56e373ef
RW		 4541(define-public preseq
4542 (package
4543 (name "preseq")
b49c5a58 4544 (version "2.0")
56e373ef
RW		 4545 (source (origin
4546 (method url-fetch)
b49c5a58
RW		 4547 (uri (string-append "https://github.com/smithlabcode/"
4548 "preseq/archive/v" version ".tar.gz"))
4549 (file-name (string-append name "-" version ".tar.gz"))
56e373ef 4550 (sha256
b49c5a58 4551 (base32 "08r684l50pnxjpvmhzjgqq56yv9rfw90k8vx0nsrnrzk8mf9hsdq"))
56e373ef
RW		 4552 (modules '((guix build utils)))
4553 (snippet
4554 ;; Remove bundled samtools.
b49c5a58 4555 '(delete-file-recursively "samtools"))))
56e373ef
RW		 4556 (build-system gnu-build-system)
4557 (arguments
4558 `(#:tests? #f ;no "check" target
4559 #:phases
4560 (modify-phases %standard-phases
56e373ef 4561 (delete 'configure))
b49c5a58
RW		 4562 #:make-flags
4563 (list (string-append "PREFIX="
4564 (assoc-ref %outputs "out"))
4565 (string-append "LIBBAM="
4566 (assoc-ref %build-inputs "samtools")
4567 "/lib/libbam.a")
4568 (string-append "SMITHLAB_CPP="
4569 (assoc-ref %build-inputs "smithlab-cpp")
4570 "/lib")
4571 "PROGS=preseq"
4572 "INCLUDEDIRS=$(SMITHLAB_CPP)/../include/smithlab-cpp $(SAMTOOLS_DIR)")))
56e373ef
RW		 4573 (inputs
4574 `(("gsl" ,gsl)
4575 ("samtools" ,samtools-0.1)
b49c5a58 4576 ("smithlab-cpp" ,smithlab-cpp)
56e373ef
RW		 4577 ("zlib" ,zlib)))
4578 (home-page "http://smithlabresearch.org/software/preseq/")
4579 (synopsis "Program for analyzing library complexity")
4580 (description
4581 "The preseq package is aimed at predicting and estimating the complexity
4582of a genomic sequencing library, equivalent to predicting and estimating the
4583number of redundant reads from a given sequencing depth and how many will be
4584expected from additional sequencing using an initial sequencing experiment.
4585The estimates can then be used to examine the utility of further sequencing,
4586optimize the sequencing depth, or to screen multiple libraries to avoid low
4587complexity samples.")
4588 (license license:gpl3+)))
4589
9ded1457
BW		 4590(define-public python-screed
4591 (package
4592 (name "python-screed")
4593 (version "0.9")
4594 (source
4595 (origin
4596 (method url-fetch)
4597 (uri (pypi-uri "screed" version))
4598 (sha256
4599 (base32
4600 "18czszp9fkx3j6jr7y5kp6dfialscgddk05mw1zkhh2zhn0jd8i0"))))
4601 (build-system python-build-system)
4602 (arguments
4603 `(#:phases
4604 (modify-phases %standard-phases
4605 (replace 'check
4606 (lambda _
4607 (setenv "PYTHONPATH"
4608 (string-append (getenv "PYTHONPATH") ":."))
4609 (zero? (system* "nosetests" "--attr" "!known_failing")))))))
4610 (native-inputs
4611 `(("python-nose" ,python-nose)))
4612 (inputs
4613 `(("python-bz2file" ,python-bz2file)))
4614 (home-page "http://github.com/dib-lab/screed/")
4615 (synopsis "Short read sequence database utilities")
4616 (description "Screed parses FASTA and FASTQ files and generates databases.
4617Values such as sequence name, sequence description, sequence quality and the
4618sequence itself can be retrieved from these databases.")
4619 (license license:bsd-3)))
4620
4621(define-public python2-screed
5c31f4aa 4622 (package-with-python2 python-screed))
9ded1457 4623
51c64999
RW		 4624(define-public sra-tools
4625 (package
4626 (name "sra-tools")
646a8433 4627 (version "2.7.0")
51c64999
RW		 4628 (source
4629 (origin
4630 (method url-fetch)
4631 (uri
4632 (string-append "https://github.com/ncbi/sra-tools/archive/"
4633 version ".tar.gz"))
4634 (file-name (string-append name "-" version ".tar.gz"))
4635 (sha256
4636 (base32
646a8433 4637 "13paw7bq6y47d2pl0ac5gpgcqp1xsy1g7v1fwysm3hr8lb2dck17"))))
51c64999
RW		 4638 (build-system gnu-build-system)
4639 (arguments
4640 `(#:parallel-build? #f ; not supported
4641 #:tests? #f ; no "check" target
2320e76b
RW		 4642 #:make-flags
4643 (list (string-append "VDB_LIBDIR="
4644 (assoc-ref %build-inputs "ncbi-vdb")
4645 ,(if (string-prefix? "x86_64"
4646 (or (%current-target-system)
4647 (%current-system)))
4648 "/lib64"
4649 "/lib32")))
51c64999
RW		 4650 #:phases
4651 (alist-replace
4652 'configure
4653 (lambda* (#:key inputs outputs #:allow-other-keys)
4654 ;; The build system expects a directory containing the sources and
4655 ;; raw build output of ncbi-vdb, including files that are not
4656 ;; installed. Since we are building against an installed version of
4657 ;; ncbi-vdb, the following modifications are needed.
4658 (substitute* "setup/konfigure.perl"
4659 ;; Make the configure script look for the "ilib" directory of
4660 ;; "ncbi-vdb" without first checking for the existence of a
4661 ;; matching library in its "lib" directory.
4662 (("^ my \\$f = File::Spec->catdir\\(\\$libdir, \\$lib\\);")
4663 "my $f = File::Spec->catdir($ilibdir, $ilib);")
4664 ;; Look for interface libraries in ncbi-vdb's "ilib" directory.
4665 (("my \\$ilibdir = File::Spec->catdir\\(\\$builddir, 'ilib'\\);")
4666 "my $ilibdir = File::Spec->catdir($dir, 'ilib');"))
4667
2320e76b
RW		 4668 ;; Dynamic linking
4669 (substitute* "tools/copycat/Makefile"
4670 (("smagic-static") "lmagic"))
4671
51c64999
RW		 4672 ;; The 'configure' script doesn't recognize things like
4673 ;; '--enable-fast-install'.
4674 (zero? (system*
4675 "./configure"
4676 (string-append "--build-prefix=" (getcwd) "/build")
4677 (string-append "--prefix=" (assoc-ref outputs "out"))
4678 (string-append "--debug")
4679 (string-append "--with-fuse-prefix="
4680 (assoc-ref inputs "fuse"))
4681 (string-append "--with-magic-prefix="
4682 (assoc-ref inputs "libmagic"))
4683 ;; TODO: building with libxml2 fails with linker errors
4684 ;; (string-append "--with-xml2-prefix="
4685 ;; (assoc-ref inputs "libxml2"))
4686 (string-append "--with-ncbi-vdb-sources="
4687 (assoc-ref inputs "ncbi-vdb"))
4688 (string-append "--with-ncbi-vdb-build="
4689 (assoc-ref inputs "ncbi-vdb"))
4690 (string-append "--with-ngs-sdk-prefix="
4691 (assoc-ref inputs "ngs-sdk"))
4692 (string-append "--with-hdf5-prefix="
4693 (assoc-ref inputs "hdf5")))))
4694 %standard-phases)))
4695 (native-inputs `(("perl" ,perl)))
4696 (inputs
4697 `(("ngs-sdk" ,ngs-sdk)
4698 ("ncbi-vdb" ,ncbi-vdb)
4699 ("libmagic" ,file)
4700 ("fuse" ,fuse)
4701 ("hdf5" ,hdf5)
4702 ("zlib" ,zlib)))
4703 (home-page "http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software")
4704 (synopsis "Tools and libraries for reading and writing sequencing data")
4705 (description
4706 "The SRA Toolkit from NCBI is a collection of tools and libraries for
4707reading of sequencing files from the Sequence Read Archive (SRA) database and
4708writing files into the .sra format.")
4709 (license license:public-domain)))
4710
d3517eda
RW		 4711(define-public seqan
4712 (package
4713 (name "seqan")
4714 (version "1.4.2")
4715 (source (origin
4716 (method url-fetch)
4717 (uri (string-append "http://packages.seqan.de/seqan-library/"
4718 "seqan-library-" version ".tar.bz2"))
4719 (sha256
4720 (base32
4721 "05s3wrrwn50f81aklfm65i4a749zag1vr8z03k21xm0pdxy47yvp"))))
4722 ;; The documentation is 7.8MB and the includes are 3.6MB heavy, so it
4723 ;; makes sense to split the outputs.
4724 (outputs '("out" "doc"))
4725 (build-system trivial-build-system)
4726 (arguments
4727 `(#:modules ((guix build utils))
4728 #:builder
4729 (begin
4730 (use-modules (guix build utils))
4731 (let ((tar (assoc-ref %build-inputs "tar"))
4732 (bzip (assoc-ref %build-inputs "bzip2"))
4733 (out (assoc-ref %outputs "out"))
4734 (doc (assoc-ref %outputs "doc")))
4735 (setenv "PATH" (string-append tar "/bin:" bzip "/bin"))
4736 (system* "tar" "xvf" (assoc-ref %build-inputs "source"))
4737 (chdir (string-append "seqan-library-" ,version))
4738 (copy-recursively "include" (string-append out "/include"))
4739 (copy-recursively "share" (string-append doc "/share"))))))
4740 (native-inputs
4741 `(("source" ,source)
4742 ("tar" ,tar)
4743 ("bzip2" ,bzip2)))
4744 (home-page "http://www.seqan.de")
4745 (synopsis "Library for nucleotide sequence analysis")
4746 (description
4747 "SeqAn is a C++ library of efficient algorithms and data structures for
4748the analysis of sequences with the focus on biological data. It contains
4749algorithms and data structures for string representation and their
4750manipulation, online and indexed string search, efficient I/O of
4751bioinformatics file formats, sequence alignment, and more.")
4752 (license license:bsd-3)))
ce7155d5 4753
d708b7a9
BW		 4754(define-public seqmagick
4755 (package
4756 (name "seqmagick")
4757 (version "0.6.1")
4758 (source
4759 (origin
4760 (method url-fetch)
4761 (uri (string-append
4762 "https://pypi.python.org/packages/source/s/seqmagick/seqmagick-"
4763 version ".tar.gz"))
4764 (sha256
4765 (base32
4766 "0cgn477n74gsl4qdaakrrhi953kcsd4q3ivk2lr18x74s3g4ma1d"))))
4767 (build-system python-build-system)
4768 (arguments
4769 ;; python2 only, see https://github.com/fhcrc/seqmagick/issues/56
4770 `(#:python ,python-2
4771 #:phases
4772 (modify-phases %standard-phases
4773 ;; Current test in setup.py does not work as of 0.6.1,
4774 ;; so use nose to run tests instead for now. See
4775 ;; https://github.com/fhcrc/seqmagick/issues/55
4776 (replace 'check (lambda _ (zero? (system* "nosetests")))))))
4777 (inputs
7bba95b7
BW		 4778 ;; biopython-1.66 is required due to
4779 ;; https://github.com/fhcrc/seqmagick/issues/59
4780 ;; When that issue is resolved the 'python2-biopython-1.66' package
4781 ;; should be removed.
4782 `(("python-biopython" ,python2-biopython-1.66)))
d708b7a9 4783 (native-inputs
f3b98f4f 4784 `(("python-nose" ,python2-nose)))
d708b7a9
BW		 4785 (home-page "http://github.com/fhcrc/seqmagick")
4786 (synopsis "Tools for converting and modifying sequence files")
4787 (description
4788 "Bioinformaticians often have to convert sequence files between formats
4789and do little manipulations on them, and it's not worth writing scripts for
4790that. Seqmagick is a utility to expose the file format conversion in
4791BioPython in a convenient way. Instead of having a big mess of scripts, there
4792is one that takes arguments.")
4793 (license license:gpl3)))
4794
66daf78c
BW		 4795(define-public seqtk
4796 (package
4797 (name "seqtk")
4798 (version "1.2")
4799 (source (origin
4800 (method url-fetch)
4801 (uri (string-append
4802 "https://github.com/lh3/seqtk/archive/v"
4803 version ".tar.gz"))
4804 (file-name (string-append name "-" version ".tar.gz"))
4805 (sha256
4806 (base32
4807 "0ywdyzpmfiz2wp6ampbzqg4y8bj450nfgqarpamg045b8mk32lxx"))
4808 (modules '((guix build utils)))
4809 (snippet
4810 '(begin
4811 ;; Remove extraneous header files, as is done in the seqtk
4812 ;; master branch.
4813 (for-each (lambda (file) (delete-file file))
4814 (list "ksort.h" "kstring.h" "kvec.h"))
4815 #t))))
4816 (build-system gnu-build-system)
4817 (arguments
4818 `(#:phases
4819 (modify-phases %standard-phases
4820 (delete 'configure)
4821 (replace 'check
4822 ;; There are no tests, so we just run a sanity check.
4823 (lambda _ (zero? (system* "./seqtk" "seq"))))
4824 (replace 'install
4825 (lambda* (#:key outputs #:allow-other-keys)
4826 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
4827 (install-file "seqtk" bin)))))))
4828 (inputs
4829 `(("zlib" ,zlib)))
4830 (home-page "https://github.com/lh3/seqtk")
4831 (synopsis "Toolkit for processing biological sequences in FASTA/Q format")
4832 (description
4833 "Seqtk is a fast and lightweight tool for processing sequences in the
4834FASTA or FASTQ format. It parses both FASTA and FASTQ files which can be
4835optionally compressed by gzip.")
4836 (license license:expat)))
4837
5f7e17be
BW		 4838(define-public snap-aligner
4839 (package
4840 (name "snap-aligner")
4841 (version "1.0beta.18")
4842 (source (origin
4843 (method url-fetch)
4844 (uri (string-append
4845 "https://github.com/amplab/snap/archive/v"
4846 version ".tar.gz"))
4847 (file-name (string-append name "-" version ".tar.gz"))
4848 (sha256
4849 (base32
4850 "1vnsjwv007k1fl1q7d681kbwn6bc66cgw6h16hym6gvyy71qv2ly"))))
4851 (build-system gnu-build-system)
4852 (arguments
4853 '(#:phases
4854 (modify-phases %standard-phases
4855 (delete 'configure)
4856 (replace 'check (lambda _ (zero? (system* "./unit_tests"))))
4857 (replace 'install
4858 (lambda* (#:key outputs #:allow-other-keys)
4859 (let* ((out (assoc-ref outputs "out"))
4860 (bin (string-append out "/bin")))
5f7e17be
BW		 4861 (install-file "snap-aligner" bin)
4862 (install-file "SNAPCommand" bin)
4863 #t))))))
4864 (native-inputs
4865 `(("zlib" ,zlib)))
4866 (home-page "http://snap.cs.berkeley.edu/")
4867 (synopsis "Short read DNA sequence aligner")
4868 (description
4869 "SNAP is a fast and accurate aligner for short DNA reads. It is
4870optimized for modern read lengths of 100 bases or higher, and takes advantage
4871of these reads to align data quickly through a hash-based indexing scheme.")
3e6fdd5f
EF		 4872 ;; 32-bit systems are not supported by the unpatched code.
4873 ;; Following the bug reports https://github.com/amplab/snap/issues/68 and
4874 ;; https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=812378 we see that
4875 ;; systems without a lot of memory cannot make good use of this program.
4876 (supported-systems '("x86_64-linux"))
5f7e17be
BW		 4877 (license license:asl2.0)))
4878
bcadaf00
BW		 4879(define-public sortmerna
4880 (package
4881 (name "sortmerna")
849485f5 4882 (version "2.1b")
bcadaf00
BW		 4883 (source
4884 (origin
4885 (method url-fetch)
4886 (uri (string-append
4887 "https://github.com/biocore/sortmerna/archive/"
4888 version ".tar.gz"))
4889 (file-name (string-append name "-" version ".tar.gz"))
4890 (sha256
4891 (base32
849485f5 4892 "1ghaghvd82af9j5adavxh77g7hm247d1r69m3fbi6f1jdivj5ldk"))))
bcadaf00
BW		 4893 (build-system gnu-build-system)
4894 (outputs '("out" ;for binaries
4895 "db")) ;for sequence databases
4896 (arguments
4897 `(#:phases
4898 (modify-phases %standard-phases
4899 (replace 'install
4900 (lambda* (#:key outputs #:allow-other-keys)
4901 (let* ((out (assoc-ref outputs "out"))
4902 (bin (string-append out "/bin"))
4903 (db (assoc-ref outputs "db"))
4904 (share
4905 (string-append db "/share/sortmerna/rRNA_databases")))
4906 (install-file "sortmerna" bin)
4907 (install-file "indexdb_rna" bin)
4908 (for-each (lambda (file)
4909 (install-file file share))
4910 (find-files "rRNA_databases" ".*fasta"))
4911 #t))))))
849485f5
BW		 4912 (inputs
4913 `(("zlib" ,zlib)))
bcadaf00
BW		 4914 (home-page "http://bioinfo.lifl.fr/RNA/sortmerna")
4915 (synopsis "Biological sequence analysis tool for NGS reads")
4916 (description
4917 "SortMeRNA is a biological sequence analysis tool for filtering, mapping
4918and operational taxonomic unit (OTU) picking of next generation
4919sequencing (NGS) reads. The core algorithm is based on approximate seeds and
4920allows for fast and sensitive analyses of nucleotide sequences. The main
4921application of SortMeRNA is filtering rRNA from metatranscriptomic data.")
4922 (license license:lgpl3)))
4923
ce7155d5
RW		 4924(define-public star
4925 (package
4926 (name "star")
3bce0f14 4927 (version "2.5.2a")
ce7155d5
RW		 4928 (source (origin
4929 (method url-fetch)
3062d750
RW		 4930 (uri (string-append "https://github.com/alexdobin/STAR/archive/"
4931 version ".tar.gz"))
4932 (file-name (string-append name "-" version ".tar.gz"))
ce7155d5
RW		 4933 (sha256
4934 (base32
3bce0f14 4935 "0xjlsm4p9flln111hv4xx7xy94c2nl53zvdvbk9winmiradjsdra"))
ce7155d5
RW		 4936 (modules '((guix build utils)))
4937 (snippet
3062d750
RW		 4938 '(begin
4939 (substitute* "source/Makefile"
4940 (("/bin/rm") "rm"))
4941 ;; Remove pre-built binaries and bundled htslib sources.
4942 (delete-file-recursively "bin/MacOSX_x86_64")
4943 (delete-file-recursively "bin/Linux_x86_64")
4944 (delete-file-recursively "source/htslib")
4945 #t))))
ce7155d5
RW		 4946 (build-system gnu-build-system)
4947 (arguments
4948 '(#:tests? #f ;no check target
4949 #:make-flags '("STAR")
4950 #:phases
c0266e8d
RW		 4951 (modify-phases %standard-phases
4952 (add-after 'unpack 'enter-source-dir
4953 (lambda _ (chdir "source") #t))
3062d750
RW		 4954 (add-after 'enter-source-dir 'do-not-use-bundled-htslib
4955 (lambda _
4956 (substitute* "Makefile"
4957 (("(Depend.list: \\$\\(SOURCES\\) parametersDefault\\.xxd) htslib"
4958 _ prefix) prefix))
4959 (substitute* '("BAMfunctions.cpp"
4960 "signalFromBAM.h"
4961 "bam_cat.h"
4962 "bam_cat.c"
4963 "STAR.cpp"
4964 "bamRemoveDuplicates.cpp")
4965 (("#include \"htslib/([^\"]+\\.h)\"" _ header)
4966 (string-append "#include <" header ">")))
4967 (substitute* "IncludeDefine.h"
4968 (("\"htslib/(htslib/[^\"]+.h)\"" _ header)
4969 (string-append "<" header ">")))
4970 #t))
c0266e8d
RW		 4971 (replace 'install
4972 (lambda* (#:key outputs #:allow-other-keys)
4973 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
4974 (install-file "STAR" bin))
4975 #t))
4976 (delete 'configure))))
ce7155d5
RW		 4977 (native-inputs
4978 `(("vim" ,vim))) ; for xxd
4979 (inputs
3062d750
RW		 4980 `(("htslib" ,htslib)
4981 ("zlib" ,zlib)))
ce7155d5
RW		 4982 (home-page "https://github.com/alexdobin/STAR")
4983 (synopsis "Universal RNA-seq aligner")
4984 (description
4985 "The Spliced Transcripts Alignment to a Reference (STAR) software is
4986based on a previously undescribed RNA-seq alignment algorithm that uses
4987sequential maximum mappable seed search in uncompressed suffix arrays followed
4988by seed clustering and stitching procedure. In addition to unbiased de novo
4989detection of canonical junctions, STAR can discover non-canonical splices and
4990chimeric (fusion) transcripts, and is also capable of mapping full-length RNA
4991sequences.")
c11f79a4
BW		 4992 ;; Only 64-bit systems are supported according to the README.
4993 (supported-systems '("x86_64-linux" "mips64el-linux"))
ce7155d5
RW		 4994 ;; STAR is licensed under GPLv3 or later; htslib is MIT-licensed.
4995 (license license:gpl3+)))
de07c0db 4996
dbf4ed7c
RW		 4997(define-public subread
4998 (package
4999 (name "subread")
c180533b 5000 (version "1.5.1")
dbf4ed7c
RW		 5001 (source (origin
5002 (method url-fetch)
de67e922
LF		 5003 (uri (string-append "mirror://sourceforge/subread/subread-"
5004 version "/subread-" version "-source.tar.gz"))
dbf4ed7c
RW		 5005 (sha256
5006 (base32
c180533b 5007 "0gn5zhbvllks0mmdg3qlmsbg91p2mpdc2wixwfqpi85yzfrh8hcy"))))
dbf4ed7c
RW		 5008 (build-system gnu-build-system)
5009 (arguments
5010 `(#:tests? #f ;no "check" target
104c1986
RW		 5011 ;; The CC and CCFLAGS variables are set to contain a lot of x86_64
5012 ;; optimizations by default, so we override these flags such that x86_64
5013 ;; flags are only added when the build target is an x86_64 system.
5014 #:make-flags
5015 (list (let ((system ,(or (%current-target-system)
5016 (%current-system)))
5017 (flags '("-ggdb" "-fomit-frame-pointer"
5018 "-ffast-math" "-funroll-loops"
5019 "-fmessage-length=0"
5020 "-O9" "-Wall" "-DMAKE_FOR_EXON"
5021 "-DMAKE_STANDALONE"
5022 "-DSUBREAD_VERSION=\\\"${SUBREAD_VERSION}\\\""))
5023 (flags64 '("-mmmx" "-msse" "-msse2" "-msse3")))
5024 (if (string-prefix? "x86_64" system)
5025 (string-append "CCFLAGS=" (string-join (append flags flags64)))
5026 (string-append "CCFLAGS=" (string-join flags))))
5027 "-f" "Makefile.Linux"
5028 "CC=gcc ${CCFLAGS}")
dbf4ed7c
RW		 5029 #:phases
5030 (alist-cons-after
5031 'unpack 'enter-dir
5032 (lambda _ (chdir "src") #t)
5033 (alist-replace
5034 'install
5035 (lambda* (#:key outputs #:allow-other-keys)
5036 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5037 (mkdir-p bin)
5038 (copy-recursively "../bin" bin)))
5039 ;; no "configure" script
5040 (alist-delete 'configure %standard-phases)))))
5041 (inputs `(("zlib" ,zlib)))
5042 (home-page "http://bioinf.wehi.edu.au/subread-package/")
5043 (synopsis "Tool kit for processing next-gen sequencing data")
5044 (description
5045 "The subread package contains the following tools: subread aligner, a
5046general-purpose read aligner; subjunc aligner: detecting exon-exon junctions
5047and mapping RNA-seq reads; featureCounts: counting mapped reads for genomic
5048features; exactSNP: a SNP caller that discovers SNPs by testing signals
5049against local background noises.")
5050 (license license:gpl3+)))
5051
d15d981e
RW		 5052(define-public stringtie
5053 (package
5054 (name "stringtie")
5055 (version "1.2.1")
5056 (source (origin
5057 (method url-fetch)
5058 (uri (string-append "http://ccb.jhu.edu/software/stringtie/dl/"
5059 "stringtie-" version ".tar.gz"))
5060 (sha256
5061 (base32
5062 "1cqllsc1maq4kh92isi8yadgzbmnf042hlnalpk3y59aph1z3bfz"))
5063 (modules '((guix build utils)))
5064 (snippet
5065 '(begin
5066 (delete-file-recursively "samtools-0.1.18")
5067 #t))))
5068 (build-system gnu-build-system)
5069 (arguments
5070 `(#:tests? #f ;no test suite
5071 #:phases
5072 (modify-phases %standard-phases
5073 ;; no configure script
5074 (delete 'configure)
5075 (add-before 'build 'use-system-samtools
5076 (lambda _
5077 (substitute* "Makefile"
5078 (("stringtie: \\$\\{BAM\\}/libbam\\.a")
5079 "stringtie: "))
5080 (substitute* '("gclib/GBam.h"
5081 "gclib/GBam.cpp")
5082 (("#include \"(bam|sam|kstring).h\"" _ header)
5083 (string-append "#include <samtools/" header ".h>")))
5084 #t))
5085 (replace 'install
5086 (lambda* (#:key outputs #:allow-other-keys)
5087 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5088 (install-file "stringtie" bin)
5089 #t))))))
5090 (inputs
5091 `(("samtools" ,samtools-0.1)
5092 ("zlib" ,zlib)))
5093 (home-page "http://ccb.jhu.edu/software/stringtie/")
5094 (synopsis "Transcript assembly and quantification for RNA-Seq data")
5095 (description
5096 "StringTie is a fast and efficient assembler of RNA-Seq sequence
5097alignments into potential transcripts. It uses a novel network flow algorithm
5098as well as an optional de novo assembly step to assemble and quantitate
5099full-length transcripts representing multiple splice variants for each gene
5100locus. Its input can include not only the alignments of raw reads used by
5101other transcript assemblers, but also alignments of longer sequences that have
5102been assembled from those reads. To identify differentially expressed genes
5103between experiments, StringTie's output can be processed either by the
5104Cuffdiff or Ballgown programs.")
5105 (license license:artistic2.0)))
5106
de07c0db
RW		 5107(define-public vcftools
5108 (package
5109 (name "vcftools")
9b36e256 5110 (version "0.1.14")
de07c0db
RW		 5111 (source (origin
5112 (method url-fetch)
5113 (uri (string-append
9b36e256
RJ		 5114 "https://github.com/vcftools/vcftools/releases/download/v"
5115 version "/vcftools-" version ".tar.gz"))
de07c0db
RW		 5116 (sha256
5117 (base32
9b36e256 5118 "10l5c07z9p4i9pr4gl54b2c9h6ndhqlbq1rashg2zcgwkbfrkmvn"))))
de07c0db
RW		 5119 (build-system gnu-build-system)
5120 (arguments
5121 `(#:tests? #f ; no "check" target
5122 #:make-flags (list
7c3958e1 5123 "CFLAGS=-O2" ; override "-m64" flag
de07c0db
RW		 5124 (string-append "PREFIX=" (assoc-ref %outputs "out"))
5125 (string-append "MANDIR=" (assoc-ref %outputs "out")
9b36e256
RJ		 5126 "/share/man/man1"))))
5127 (native-inputs
5128 `(("pkg-config" ,pkg-config)))
de07c0db
RW		 5129 (inputs
5130 `(("perl" ,perl)
5131 ("zlib" ,zlib)))
9b36e256 5132 (home-page "https://vcftools.github.io/")
de07c0db
RW		 5133 (synopsis "Tools for working with VCF files")
5134 (description
5135 "VCFtools is a program package designed for working with VCF files, such
5136as those generated by the 1000 Genomes Project. The aim of VCFtools is to
5137provide easily accessible methods for working with complex genetic variation
5138data in the form of VCF files.")
5139 ;; The license is declared as LGPLv3 in the README and
9b36e256 5140 ;; at https://vcftools.github.io/license.html
de07c0db 5141 (license license:lgpl3)))
9c38b540 5142
35aa90a1
RW		 5143(define-public infernal
5144 (package
5145 (name "infernal")
5146 (version "1.1.2")
5147 (source (origin
5148 (method url-fetch)
5149 (uri (string-append "http://eddylab.org/software/infernal/"
5150 "infernal-" version ".tar.gz"))
5151 (sha256
5152 (base32
5153 "0sr2hiz3qxfwqpz3whxr6n82p3x27336v3f34iqznp10hks2935c"))))
5154 (build-system gnu-build-system)
5155 (native-inputs
5156 `(("perl" ,perl))) ; for tests
5157 (home-page "http://eddylab.org/infernal/")
5158 (synopsis "Inference of RNA alignments")
5159 (description "Infernal (\"INFERence of RNA ALignment\") is a tool for
5160searching DNA sequence databases for RNA structure and sequence similarities.
5161It is an implementation of a special case of profile stochastic context-free
5162grammars called @dfn{covariance models} (CMs). A CM is like a sequence
5163profile, but it scores a combination of sequence consensus and RNA secondary
5164structure consensus, so in many cases, it is more capable of identifying RNA
5165homologs that conserve their secondary structure more than their primary
5166sequence.")
48409ef2
EF		 5167 ;; Infernal 1.1.2 requires VMX or SSE capability for parallel instructions.
5168 (supported-systems '("i686-linux" "x86_64-linux"))
35aa90a1
RW		 5169 (license license:bsd-3)))
5170
7b3df1e5
BW		 5171(define-public r-vegan
5172 (package
5173 (name "r-vegan")
2c575bb9 5174 (version "2.4-1")
7b3df1e5
BW		 5175 (source
5176 (origin
5177 (method url-fetch)
5178 (uri (cran-uri "vegan" version))
5179 (sha256
5180 (base32
2c575bb9 5181 "0i0c7rc0nzgbysd1nlxzxd2rvy75qcnw3yc7nggzqjzzj5d7yzsd"))))
7b3df1e5
BW		 5182 (build-system r-build-system)
5183 (arguments
5184 `(#:phases
5185 (modify-phases %standard-phases
5186 (add-after 'unpack 'revert-test-deletion
5187 ;; The distributed sources do not include tests with the CRAN
5188 ;; package. Here we revert the commit
5189 ;; `591d0e8ba1deaaf82445474ec6619c0b43db4e63' which deletes these
5190 ;; tests. There are plans to not delete tests in future as
5191 ;; documented at https://github.com/vegandevs/vegan/issues/181.
5192 (lambda* (#:key inputs #:allow-other-keys)
5193 (zero?
5194 (system* "patch" "-R" "-p1" "-i"
5195 (assoc-ref inputs "r-vegan-delete-tests-patch"))))))))
5196 (native-inputs
5197 `(("gfortran" ,gfortran)
5198 ("r-knitr" ,r-knitr)
5199 ("r-vegan-delete-tests-patch"
5200 ,(origin
5201 (method url-fetch)
5202 (uri (string-append
5203 "https://github.com/vegandevs/vegan/commit/"
5204 "591d0e8ba1deaaf82445474ec6619c0b43db4e63.patch"))
5205 (sha256
5206 (base32
5207 "0b1bi7y4jjdl3ph721vm9apm51dr2z9piwvhy4355sf2b4kyyj5a"))))))
5208 (propagated-inputs
5209 `(("r-cluster" ,r-cluster)
5210 ("r-lattice" ,r-lattice)
5211 ("r-mgcv" ,r-mgcv)
5212 ("r-permute" ,r-permute)))
5213 (home-page "https://cran.r-project.org/web/packages/vegan")
5214 (synopsis "Functions for community ecology")
5215 (description
5216 "The vegan package provides tools for descriptive community ecology. It
5217has most basic functions of diversity analysis, community ordination and
5218dissimilarity analysis. Most of its multivariate tools can be used for other
5219data types as well.")
5220 (license license:gpl2+)))
5221
8c6de588
RW		 5222(define-public r-annotate
5223 (package
5224 (name "r-annotate")
9beba74f 5225 (version "1.52.0")
8c6de588
RW		 5226 (source
5227 (origin
5228 (method url-fetch)
5229 (uri (bioconductor-uri "annotate" version))
5230 (sha256
5231 (base32
9beba74f 5232 "1fd2csq7dcs2gwndgwdx2nwkymz8gsmlnqqzv3p0vjjsvvq5n2a8"))))
8c6de588
RW		 5233 (build-system r-build-system)
5234 (propagated-inputs
5235 `(("r-annotationdbi" ,r-annotationdbi)
5236 ("r-biobase" ,r-biobase)
5237 ("r-biocgenerics" ,r-biocgenerics)
5238 ("r-dbi" ,r-dbi)
5239 ("r-xml" ,r-xml)
5240 ("r-xtable" ,r-xtable)))
5241 (home-page
5242 "http://bioconductor.org/packages/annotate")
5243 (synopsis "Annotation for microarrays")
5244 (description "This package provides R enviroments for the annotation of
5245microarrays.")
5246 (license license:artistic2.0)))
5247
07a664cd
RW		 5248(define-public r-geneplotter
5249 (package
5250 (name "r-geneplotter")
6faa2a48 5251 (version "1.52.0")
07a664cd
RW		 5252 (source
5253 (origin
5254 (method url-fetch)
5255 (uri (bioconductor-uri "geneplotter" version))
5256 (sha256
5257 (base32
6faa2a48 5258 "1p6yvxi243irhjxwm97hp73abhwampj0myyf8z00ij166674pc7h"))))
07a664cd
RW		 5259 (build-system r-build-system)
5260 (propagated-inputs
5261 `(("r-annotate" ,r-annotate)
5262 ("r-annotationdbi" ,r-annotationdbi)
5263 ("r-biobase" ,r-biobase)
5264 ("r-biocgenerics" ,r-biocgenerics)
5265 ("r-lattice" ,r-lattice)
5266 ("r-rcolorbrewer" ,r-rcolorbrewer)))
5267 (home-page "http://bioconductor.org/packages/geneplotter")
5268 (synopsis "Graphics functions for genomic data")
5269 (description
5270 "This package provides functions for plotting genomic data.")
5271 (license license:artistic2.0)))
5272
2301fd3e
RW		 5273(define-public r-genefilter
5274 (package
5275 (name "r-genefilter")
1e66e719 5276 (version "1.56.0")
2301fd3e
RW		 5277 (source
5278 (origin
5279 (method url-fetch)
5280 (uri (bioconductor-uri "genefilter" version))
5281 (sha256
5282 (base32
1e66e719 5283 "1vzgciqd09csqcw9qync8blsv51ylrd86a65iadgyy6j26g01fwd"))))
2301fd3e
RW		 5284 (build-system r-build-system)
5285 (native-inputs
5286 `(("gfortran" ,gfortran)))
5287 (propagated-inputs
5288 `(("r-annotate" ,r-annotate)
5289 ("r-annotationdbi" ,r-annotationdbi)
5290 ("r-biobase" ,r-biobase)
5291 ("r-s4vectors" ,r-s4vectors)))
5292 (home-page "http://bioconductor.org/packages/genefilter")
5293 (synopsis "Filter genes from high-throughput experiments")
5294 (description
5295 "This package provides basic functions for filtering genes from
5296high-throughput sequencing experiments.")
5297 (license license:artistic2.0)))
5298
ad34f0ac
RW		 5299(define-public r-deseq2
5300 (package
5301 (name "r-deseq2")
e1db8dd8 5302 (version "1.14.0")
ad34f0ac
RW		 5303 (source
5304 (origin
5305 (method url-fetch)
5306 (uri (bioconductor-uri "DESeq2" version))
5307 (sha256
5308 (base32
e1db8dd8 5309 "0kq06jy4xg5ii3a9l62f17kirsfx0gsiwq6mhiy985cqzpdn893g"))))
ad34f0ac
RW		 5310 (properties `((upstream-name . "DESeq2")))
5311 (build-system r-build-system)
5312 (arguments
5313 `(#:phases
5314 (modify-phases %standard-phases
5315 (add-after 'unpack 'link-against-armadillo
5316 (lambda _
5317 (substitute* "src/Makevars"
5318 (("PKG_LIBS =" prefix)
5319 (string-append prefix "-larmadillo"))))))))
5320 (propagated-inputs
5321 `(("r-biobase" ,r-biobase)
5322 ("r-biocgenerics" ,r-biocgenerics)
5323 ("r-biocparallel" ,r-biocparallel)
5324 ("r-genefilter" ,r-genefilter)
5325 ("r-geneplotter" ,r-geneplotter)
5326 ("r-genomicranges" ,r-genomicranges)
5327 ("r-ggplot2" ,r-ggplot2)
5328 ("r-hmisc" ,r-hmisc)
5329 ("r-iranges" ,r-iranges)
5330 ("r-locfit" ,r-locfit)
5331 ("r-rcpp" ,r-rcpp)
5332 ("r-rcpparmadillo" ,r-rcpparmadillo)
5333 ("r-s4vectors" ,r-s4vectors)
5334 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5335 (home-page "http://bioconductor.org/packages/DESeq2")
5336 (synopsis "Differential gene expression analysis")
5337 (description
5338 "This package provides functions to estimate variance-mean dependence in
5339count data from high-throughput nucleotide sequencing assays and test for
5340differential expression based on a model using the negative binomial
5341distribution.")
5342 (license license:lgpl3+)))
5343
e8163773
RW		 5344(define-public r-annotationforge
5345 (package
5346 (name "r-annotationforge")
55cd914c 5347 (version "1.16.0")
e8163773
RW		 5348 (source
5349 (origin
5350 (method url-fetch)
5351 (uri (bioconductor-uri "AnnotationForge" version))
5352 (sha256
5353 (base32
55cd914c 5354 "02msyb9p3hywrryx00zpjkjl126mrv827i1ah1092s0cplm6xxvf"))))
e8163773
RW		 5355 (properties
5356 `((upstream-name . "AnnotationForge")))
5357 (build-system r-build-system)
5358 (propagated-inputs
5359 `(("r-annotationdbi" ,r-annotationdbi)
5360 ("r-biobase" ,r-biobase)
5361 ("r-biocgenerics" ,r-biocgenerics)
5362 ("r-dbi" ,r-dbi)
55cd914c 5363 ("r-rcurl" ,r-rcurl)
e8163773
RW		 5364 ("r-rsqlite" ,r-rsqlite)
5365 ("r-s4vectors" ,r-s4vectors)
5366 ("r-xml" ,r-xml)))
5367 (home-page "http://bioconductor.org/packages/AnnotationForge")
5368 (synopsis "Code for building annotation database packages")
5369 (description
5370 "This package provides code for generating Annotation packages and their
5371databases. Packages produced are intended to be used with AnnotationDbi.")
5372 (license license:artistic2.0)))
5373
cd9e7dc7
RW		 5374(define-public r-rbgl
5375 (package
5376 (name "r-rbgl")
700c780d 5377 (version "1.50.0")
cd9e7dc7
RW		 5378 (source
5379 (origin
5380 (method url-fetch)
5381 (uri (bioconductor-uri "RBGL" version))
5382 (sha256
5383 (base32
700c780d 5384 "1q14m8w6ih56v680kf3d9wh1qbgp7af33kz3cxafdf1vvzx9km08"))))
cd9e7dc7
RW		 5385 (properties `((upstream-name . "RBGL")))
5386 (build-system r-build-system)
5387 (propagated-inputs `(("r-graph" ,r-graph)))
5388 (home-page "http://www.bioconductor.org/packages/RBGL")
5389 (synopsis "Interface to the Boost graph library")
5390 (description
5391 "This package provides a fairly extensive and comprehensive interface to
5392the graph algorithms contained in the Boost library.")
5393 (license license:artistic2.0)))
5394
ad740ff8
RW		 5395(define-public r-gseabase
5396 (package
5397 (name "r-gseabase")
558d02c3 5398 (version "1.36.0")
ad740ff8
RW		 5399 (source
5400 (origin
5401 (method url-fetch)
5402 (uri (bioconductor-uri "GSEABase" version))
5403 (sha256
5404 (base32
558d02c3 5405 "0l2x7yj7lfb0m2dmsav5ib026dikpgl4crdckrnj776yy08lgxpj"))))
ad740ff8
RW		 5406 (properties `((upstream-name . "GSEABase")))
5407 (build-system r-build-system)
5408 (propagated-inputs
5409 `(("r-annotate" ,r-annotate)
5410 ("r-annotationdbi" ,r-annotationdbi)
5411 ("r-biobase" ,r-biobase)
5412 ("r-biocgenerics" ,r-biocgenerics)
5413 ("r-graph" ,r-graph)
5414 ("r-xml" ,r-xml)))
5415 (home-page "http://bioconductor.org/packages/GSEABase")
5416 (synopsis "Gene set enrichment data structures and methods")
5417 (description
5418 "This package provides classes and methods to support @dfn{Gene Set
5419Enrichment Analysis} (GSEA).")
5420 (license license:artistic2.0)))
5421
1a1931f7
RW		 5422(define-public r-category
5423 (package
5424 (name "r-category")
816599c1 5425 (version "2.40.0")
1a1931f7
RW		 5426 (source
5427 (origin
5428 (method url-fetch)
5429 (uri (bioconductor-uri "Category" version))
5430 (sha256
5431 (base32
816599c1 5432 "16ncwz7b4y48k0p3fvbrbmvf7nfz63li9ysgcl8kp9kl4hg7llng"))))
1a1931f7
RW		 5433 (properties `((upstream-name . "Category")))
5434 (build-system r-build-system)
5435 (propagated-inputs
5436 `(("r-annotate" ,r-annotate)
5437 ("r-annotationdbi" ,r-annotationdbi)
5438 ("r-biobase" ,r-biobase)
5439 ("r-biocgenerics" ,r-biocgenerics)
5440 ("r-genefilter" ,r-genefilter)
5441 ("r-graph" ,r-graph)
5442 ("r-gseabase" ,r-gseabase)
5443 ("r-matrix" ,r-matrix)
5444 ("r-rbgl" ,r-rbgl)
5445 ("r-rsqlite" ,r-rsqlite)))
5446 (home-page "http://bioconductor.org/packages/Category")
5447 (synopsis "Category analysis")
5448 (description
5449 "This package provides a collection of tools for performing category
5450analysis.")
5451 (license license:artistic2.0)))
5452
89f40c5e
RW		 5453(define-public r-gostats
5454 (package
5455 (name "r-gostats")
ae6899b6 5456 (version "2.40.0")
89f40c5e
RW		 5457 (source
5458 (origin
5459 (method url-fetch)
5460 (uri (bioconductor-uri "GOstats" version))
5461 (sha256
5462 (base32
ae6899b6 5463 "0g2czm94zhzx92z7y2r4mjfxhwml7bhab2db6820ks8nkw1zvr9n"))))
89f40c5e
RW		 5464 (properties `((upstream-name . "GOstats")))
5465 (build-system r-build-system)
5466 (propagated-inputs
5467 `(("r-annotate" ,r-annotate)
5468 ("r-annotationdbi" ,r-annotationdbi)
5469 ("r-annotationforge" ,r-annotationforge)
5470 ("r-biobase" ,r-biobase)
5471 ("r-category" ,r-category)
5472 ("r-go-db" ,r-go-db)
5473 ("r-graph" ,r-graph)
5474 ("r-rbgl" ,r-rbgl)))
5475 (home-page "http://bioconductor.org/packages/GOstats")
5476 (synopsis "Tools for manipulating GO and microarrays")
5477 (description
5478 "This package provides a set of tools for interacting with GO and
5479microarray data. A variety of basic manipulation tools for graphs, hypothesis
5480testing and other simple calculations.")
5481 (license license:artistic2.0)))
5482
cb99d457
RW		 5483(define-public r-shortread
5484 (package
5485 (name "r-shortread")
b87b41a7 5486 (version "1.32.0")
cb99d457
RW		 5487 (source
5488 (origin
5489 (method url-fetch)
5490 (uri (bioconductor-uri "ShortRead" version))
5491 (sha256
5492 (base32
b87b41a7 5493 "0mjdlg92x5qw4x2djc4dv5lxwl7ai6ix56nnf86zr07jk8vc7yls"))))
cb99d457
RW		 5494 (properties `((upstream-name . "ShortRead")))
5495 (build-system r-build-system)
5496 (inputs
5497 `(("zlib" ,zlib)))
5498 (propagated-inputs
5499 `(("r-biobase" ,r-biobase)
5500 ("r-biocgenerics" ,r-biocgenerics)
5501 ("r-biocparallel" ,r-biocparallel)
5502 ("r-biostrings" ,r-biostrings)
5503 ("r-genomeinfodb" ,r-genomeinfodb)
5504 ("r-genomicalignments" ,r-genomicalignments)
5505 ("r-genomicranges" ,r-genomicranges)
5506 ("r-hwriter" ,r-hwriter)
5507 ("r-iranges" ,r-iranges)
5508 ("r-lattice" ,r-lattice)
5509 ("r-latticeextra" ,r-latticeextra)
5510 ("r-rsamtools" ,r-rsamtools)
5511 ("r-s4vectors" ,r-s4vectors)
5512 ("r-xvector" ,r-xvector)
5513 ("r-zlibbioc" ,r-zlibbioc)))
5514 (home-page "http://bioconductor.org/packages/ShortRead")
5515 (synopsis "FASTQ input and manipulation tools")
5516 (description
5517 "This package implements sampling, iteration, and input of FASTQ files.
5518It includes functions for filtering and trimming reads, and for generating a
5519quality assessment report. Data are represented as
5520@code{DNAStringSet}-derived objects, and easily manipulated for a diversity of
5521purposes. The package also contains legacy support for early single-end,
5522ungapped alignment formats.")
5523 (license license:artistic2.0)))
5524
7f903d73
RW		 5525(define-public r-systempiper
5526 (package
5527 (name "r-systempiper")
1265d387 5528 (version "1.8.1")
7f903d73
RW		 5529 (source
5530 (origin
5531 (method url-fetch)
5532 (uri (bioconductor-uri "systemPipeR" version))
5533 (sha256
5534 (base32
1265d387 5535 "0hyi841w8fm2yzpm6lwqi3jz5kc8ny8dy5p29dxynzaw5bpjw56d"))))
7f903d73
RW		 5536 (properties `((upstream-name . "systemPipeR")))
5537 (build-system r-build-system)
5538 (propagated-inputs
5539 `(("r-annotate" ,r-annotate)
5540 ("r-batchjobs" ,r-batchjobs)
5541 ("r-biocgenerics" ,r-biocgenerics)
5542 ("r-biostrings" ,r-biostrings)
5543 ("r-deseq2" ,r-deseq2)
5544 ("r-edger" ,r-edger)
5545 ("r-genomicfeatures" ,r-genomicfeatures)
5546 ("r-genomicranges" ,r-genomicranges)
5547 ("r-ggplot2" ,r-ggplot2)
5548 ("r-go-db" ,r-go-db)
5549 ("r-gostats" ,r-gostats)
5550 ("r-limma" ,r-limma)
5551 ("r-pheatmap" ,r-pheatmap)
5552 ("r-rjson" ,r-rjson)
5553 ("r-rsamtools" ,r-rsamtools)
5554 ("r-shortread" ,r-shortread)
5555 ("r-summarizedexperiment" ,r-summarizedexperiment)
5556 ("r-variantannotation" ,r-variantannotation)))
5557 (home-page "https://github.com/tgirke/systemPipeR")
5558 (synopsis "Next generation sequencing workflow and reporting environment")
5559 (description
5560 "This R package provides tools for building and running automated
5561end-to-end analysis workflows for a wide range of @dfn{next generation
5562sequence} (NGS) applications such as RNA-Seq, ChIP-Seq, VAR-Seq and Ribo-Seq.
5563Important features include a uniform workflow interface across different NGS
5564applications, automated report generation, and support for running both R and
5565command-line software, such as NGS aligners or peak/variant callers, on local
5566computers or compute clusters. Efficient handling of complex sample sets and
5567experimental designs is facilitated by a consistently implemented sample
5568annotation infrastructure.")
5569 (license license:artistic2.0)))
5570
684f29bd
RW		 5571(define-public r-grohmm
5572 (package
5573 (name "r-grohmm")
c9b5046a 5574 (version "1.8.0")
684f29bd
RW		 5575 (source
5576 (origin
5577 (method url-fetch)
5578 (uri (bioconductor-uri "groHMM" version))
5579 (sha256
5580 (base32
c9b5046a 5581 "0d91nyhqbi5hv3mgmr2z0g29wg2md26g0hyv5mgapmz20cd9zi4y"))))
684f29bd
RW		 5582 (properties `((upstream-name . "groHMM")))
5583 (build-system r-build-system)
5584 (propagated-inputs
5585 `(("r-genomeinfodb" ,r-genomeinfodb)
5586 ("r-genomicalignments" ,r-genomicalignments)
5587 ("r-genomicranges" ,r-genomicranges)
5588 ("r-iranges" ,r-iranges)
5589 ("r-rtracklayer" ,r-rtracklayer)
5590 ("r-s4vectors" ,r-s4vectors)))
5591 (home-page "https://github.com/Kraus-Lab/groHMM")
5592 (synopsis "GRO-seq analysis pipeline")
5593 (description
5594 "This package provides a pipeline for the analysis of GRO-seq data.")
5595 (license license:gpl3+)))
5596
f3cfe451
RW		 5597(define-public r-txdb-hsapiens-ucsc-hg19-knowngene
5598 (package
5599 (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
5600 (version "3.2.2")
5601 (source (origin
5602 (method url-fetch)
5603 ;; We cannot use bioconductor-uri here because this tarball is
5604 ;; located under "data/annotation/" instead of "bioc/".
5605 (uri (string-append "http://bioconductor.org/packages/"
5606 "release/data/annotation/src/contrib"
5607 "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
5608 version ".tar.gz"))
5609 (sha256
5610 (base32
5611 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
5612 (properties
5613 `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
5614 (build-system r-build-system)
5615 ;; As this package provides little more than a very large data file it
5616 ;; doesn't make sense to build substitutes.
5617 (arguments `(#:substitutable? #f))
5618 (propagated-inputs
5619 `(("r-genomicfeatures" ,r-genomicfeatures)))
5620 (home-page
5621 "http://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
5622 (synopsis "Annotation package for human genome in TxDb format")
5623 (description
5624 "This package provides an annotation database of Homo sapiens genome
5625data. It is derived from the UCSC hg19 genome and based on the \"knownGene\"
5626track. The database is exposed as a @code{TxDb} object.")
5627 (license license:artistic2.0)))
5628
a2950fa4
BW		 5629(define-public vsearch
5630 (package
5631 (name "vsearch")
a225db52 5632 (version "2.3.4")
a2950fa4
BW		 5633 (source
5634 (origin
5635 (method url-fetch)
5636 (uri (string-append
5637 "https://github.com/torognes/vsearch/archive/v"
5638 version ".tar.gz"))
5639 (file-name (string-append name "-" version ".tar.gz"))
5640 (sha256
5641 (base32
a225db52 5642 "1xyraxmhyx62mxx8z7c8waygvcijwkh48ms1ar60w2cv2y2sn4al"))
a2950fa4
BW		 5643 (modules '((guix build utils)))
5644 (snippet
5645 '(begin
5646 ;; Remove bundled cityhash and '-mtune=native'.
5647 (substitute* "src/Makefile.am"
5648 (("^AM_CXXFLAGS=-I\\$\\{srcdir\\}/cityhash \
5649-O3 -mtune=native -Wall -Wsign-compare")
5650 (string-append "AM_CXXFLAGS=-lcityhash"
5651 " -O3 -Wall -Wsign-compare"))
cf6edaba 5652 (("^__top_builddir__bin_vsearch_SOURCES = city.h \\\\")
a2950fa4 5653 "__top_builddir__bin_vsearch_SOURCES = \\")
cf6edaba
BW		 5654 (("^city.h \\\\") "\\")
5655 (("^citycrc.h \\\\") "\\")
5656 (("^libcityhash_a.*") "")
5657 (("noinst_LIBRARIES = libcpu_sse2.a libcpu_ssse3.a \
5658libcityhash.a")
5659 "noinst_LIBRARIES = libcpu_sse2.a libcpu_ssse3.a")
5660 (("__top_builddir__bin_vsearch_LDADD = libcpu_ssse3.a \
5661libcpu_sse2.a libcityhash.a")
5662 "__top_builddir__bin_vsearch_LDADD = libcpu_ssse3.a \
5663libcpu_sse2.a -lcityhash"))
a2950fa4 5664 (substitute* "src/vsearch.h"
cf6edaba
BW		 5665 (("^\\#include \"city.h\"") "#include <city.h>")
5666 (("^\\#include \"citycrc.h\"") "#include <citycrc.h>"))
5667 (delete-file "src/city.h")
5668 (delete-file "src/citycrc.h")
5669 (delete-file "src/city.cc")
a2950fa4
BW		 5670 #t))))
5671 (build-system gnu-build-system)
5672 (arguments
5673 `(#:phases
5674 (modify-phases %standard-phases
5675 (add-before 'configure 'autogen
5676 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
5677 (inputs
5678 `(("zlib" ,zlib)
5679 ("bzip2" ,bzip2)
5680 ("cityhash" ,cityhash)))
5681 (native-inputs
5682 `(("autoconf" ,autoconf)
5683 ("automake" ,automake)))
5684 (synopsis "Sequence search tools for metagenomics")
5685 (description
5686 "VSEARCH supports DNA sequence searching, clustering, chimera detection,
5687dereplication, pairwise alignment, shuffling, subsampling, sorting and
5688masking. The tool takes advantage of parallelism in the form of SIMD
5689vectorization as well as multiple threads to perform accurate alignments at
5690high speed. VSEARCH uses an optimal global aligner (full dynamic programming
5691Needleman-Wunsch).")
5692 (home-page "https://github.com/torognes/vsearch")
6f04e515
BW		 5693 ;; vsearch uses non-portable SSE intrinsics so building fails on other
5694 ;; platforms.
5695 (supported-systems '("x86_64-linux"))
a2950fa4
BW		 5696 ;; Dual licensed; also includes public domain source.
5697 (license (list license:gpl3 license:bsd-2))))
5698
07837874
RW		 5699(define-public pardre
5700 (package
5701 (name "pardre")
7922ab8f
BW		 5702 ;; The source of 1.1.5 changed in place, so we append "-1" to the version.
5703 (version "1.1.5-1")
07837874
RW		 5704 (source
5705 (origin
5706 (method url-fetch)
5707 (uri (string-append "mirror://sourceforge/pardre/ParDRe-rel"
7922ab8f 5708 "1.1.5" ".tar.gz"))
07837874
RW		 5709 (sha256
5710 (base32
7922ab8f 5711 "17j73nc0viq4f6qj50nrndsrif5d6b71q8fl87m54psiv0ilns2b"))))
07837874
RW		 5712 (build-system gnu-build-system)
5713 (arguments
5714 `(#:tests? #f ; no tests included
5715 #:phases
5716 (modify-phases %standard-phases
5717 (delete 'configure)
5718 (replace 'install
5719 (lambda* (#:key outputs #:allow-other-keys)
5720 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
07837874
RW		 5721 (install-file "ParDRe" bin)
5722 #t))))))
5723 (inputs
5724 `(("openmpi" ,openmpi)
5725 ("zlib" ,zlib)))
5726 (synopsis "Parallel tool to remove duplicate DNA reads")
5727 (description
5728 "ParDRe is a parallel tool to remove duplicate genetic sequence reads.
5729Duplicate reads can be seen as identical or nearly identical sequences with
5730some mismatches. This tool lets users avoid the analysis of unnecessary
5731reads, reducing the time of subsequent procedures with the
5732dataset (e.g. assemblies, mappings, etc.). The tool is implemented with MPI
5733in order to exploit the parallel capabilities of multicore clusters. It is
5734faster than multithreaded counterparts (end of 2015) for the same number of
5735cores and, thanks to the message-passing technology, it can be executed on
5736clusters.")
5737 (home-page "https://sourceforge.net/projects/pardre/")
5738 (license license:gpl3+)))
5739
e4a44a6a
BW		 5740(define-public ruby-bio-kseq
5741 (package
5742 (name "ruby-bio-kseq")
5743 (version "0.0.2")
5744 (source
5745 (origin
5746 (method url-fetch)
5747 (uri (rubygems-uri "bio-kseq" version))
5748 (sha256
5749 (base32
5750 "1xyaha46khb5jc6wzkbf7040jagac49jbimn0vcrzid0j8jdikrz"))))
5751 (build-system ruby-build-system)
5752 (arguments
5753 `(#:test-target "spec"))
5754 (native-inputs
5755 `(("bundler" ,bundler)
5756 ("ruby-rspec" ,ruby-rspec)
5757 ("ruby-rake-compiler" ,ruby-rake-compiler)))
5758 (inputs
5759 `(("zlib" ,zlib)))
5760 (synopsis "Ruby bindings for the kseq.h FASTA/Q parser")
5761 (description
5762 "@code{Bio::Kseq} provides ruby bindings to the @code{kseq.h} FASTA and
5763FASTQ parsing code. It provides a fast iterator over sequences and their
5764quality scores.")
5765 (home-page "https://github.com/gusevfe/bio-kseq")
5766 (license license:expat)))
5767
9c38b540
PP		 5768(define-public bio-locus
5769 (package
5770 (name "bio-locus")
5771 (version "0.0.7")
5772 (source
5773 (origin
5774 (method url-fetch)
5775 (uri (rubygems-uri "bio-locus" version))
5776 (sha256
5777 (base32
5778 "02vmrxyimkj9sahsp4zhfhnmbvz6dbbqz1y01vglf8cbwvkajfl0"))))
5779 (build-system ruby-build-system)
5780 (native-inputs
5781 `(("ruby-rspec" ,ruby-rspec)))
5782 (synopsis "Tool for fast querying of genome locations")
5783 (description
5784 "Bio-locus is a tabix-like tool for fast querying of genome
5785locations. Many file formats in bioinformatics contain records that
5786start with a chromosome name and a position for a SNP, or a start-end
5787position for indels. Bio-locus allows users to store this chr+pos or
5788chr+pos+alt information in a database.")
5789 (home-page "https://github.com/pjotrp/bio-locus")
5790 (license license:expat)))
edb15985 5791
b2bddb07
PP		 5792(define-public bio-blastxmlparser
5793 (package
5794 (name "bio-blastxmlparser")
5795 (version "2.0.4")
5796 (source (origin
5797 (method url-fetch)
5798 (uri (rubygems-uri "bio-blastxmlparser" version))
5799 (sha256
5800 (base32
5801 "1wf4qygcmdjgcqm6flmvsagfr1gs9lf63mj32qv3z1f481zc5692"))))
5802 (build-system ruby-build-system)
5803 (propagated-inputs
5804 `(("ruby-bio-logger" ,ruby-bio-logger)
5805 ("ruby-nokogiri" ,ruby-nokogiri)))
5806 (inputs
5807 `(("ruby-rspec" ,ruby-rspec)))
5808 (synopsis "Fast big data BLAST XML parser and library")
5809 (description
5810 "Very fast parallel big-data BLAST XML file parser which can be used as
5811command line utility. Use blastxmlparser to: Parse BLAST XML; filter output;
5812generate FASTA, JSON, YAML, RDF, JSON-LD, HTML, CSV, tabular output etc.")
5813 (home-page "http://github.com/pjotrp/blastxmlparser")
5814 (license license:expat)))
5815
edb15985
PP		 5816(define-public bioruby
5817 (package
5818 (name "bioruby")
5819 (version "1.5.0")
5820 (source
5821 (origin
5822 (method url-fetch)
5823 (uri (rubygems-uri "bio" version))
5824 (sha256
5825 (base32
5826 "01k2fyjl5fpx4zn8g6gqiqvsg2j1fgixrs9p03vzxckynxdq3wmc"))))
5827 (build-system ruby-build-system)
5828 (propagated-inputs
5829 `(("ruby-libxml" ,ruby-libxml)))
5830 (native-inputs
5831 `(("which" ,which))) ; required for test phase
5832 (arguments
5833 `(#:phases
5834 (modify-phases %standard-phases
5835 (add-before 'build 'patch-test-command
5836 (lambda _
5837 (substitute* '("test/functional/bio/test_command.rb")
5838 (("/bin/sh") (which "sh")))
5839 (substitute* '("test/functional/bio/test_command.rb")
5840 (("/bin/ls") (which "ls")))
5841 (substitute* '("test/functional/bio/test_command.rb")
5842 (("which") (which "which")))
5843 (substitute* '("test/functional/bio/test_command.rb",
5844 "test/data/command/echoarg2.sh")
5845 (("/bin/echo") (which "echo")))
5846 #t)))))
5847 (synopsis "Ruby library, shell and utilities for bioinformatics")
5848 (description "BioRuby comes with a comprehensive set of Ruby development
5849tools and libraries for bioinformatics and molecular biology. BioRuby has
5850components for sequence analysis, pathway analysis, protein modelling and
5851phylogenetic analysis; it supports many widely used data formats and provides
5852easy access to databases, external programs and public web services, including
5853BLAST, KEGG, GenBank, MEDLINE and GO.")
5854 (home-page "http://bioruby.org/")
5855 ;; Code is released under Ruby license, except for setup
5856 ;; (LGPLv2.1+) and scripts in samples (which have GPL2 and GPL2+)
5857 (license (list license:ruby license:lgpl2.1+ license:gpl2+ ))))
a5002ae7 5858
9fba89e8
RW		 5859(define-public r-acsnminer
5860 (package
5861 (name "r-acsnminer")
0b54b4c9 5862 (version "0.16.8.25")
9fba89e8
RW		 5863 (source (origin
5864 (method url-fetch)
5865 (uri (cran-uri "ACSNMineR" version))
5866 (sha256
5867 (base32
0b54b4c9 5868 "0gh604s8qall6zfjlwcg2ilxjvz08dplf9k5g47idhv43scm748l"))))
9fba89e8
RW		 5869 (properties `((upstream-name . "ACSNMineR")))
5870 (build-system r-build-system)
5871 (propagated-inputs
5872 `(("r-ggplot2" ,r-ggplot2)
5873 ("r-gridextra" ,r-gridextra)))
5874 (home-page "http://cran.r-project.org/web/packages/ACSNMineR")
5875 (synopsis "Gene enrichment analysis")
5876 (description
5877 "This package provides tools to compute and represent gene set enrichment
5878or depletion from your data based on pre-saved maps from the @dfn{Atlas of
5879Cancer Signalling Networks} (ACSN) or user imported maps. The gene set
5880enrichment can be run with hypergeometric test or Fisher exact test, and can
5881use multiple corrections. Visualization of data can be done either by
5882barplots or heatmaps.")
5883 (license license:gpl2+)))
5884
d29b25c4
RW		 5885(define-public r-biocgenerics
5886 (package
5887 (name "r-biocgenerics")
f0b297e1 5888 (version "0.20.0")
d29b25c4
RW		 5889 (source (origin
5890 (method url-fetch)
5891 (uri (bioconductor-uri "BiocGenerics" version))
5892 (sha256
5893 (base32
f0b297e1 5894 "06szdz7dfs1iyv5zdl4fjzad18nnf1zf3wvglc6c6yd9mrqlf7vk"))))
d29b25c4 5895 (properties
1d216b6e 5896 `((upstream-name . "BiocGenerics")))
d29b25c4
RW		 5897 (build-system r-build-system)
5898 (home-page "http://bioconductor.org/packages/BiocGenerics")
5899 (synopsis "S4 generic functions for Bioconductor")
5900 (description
5901 "This package provides S4 generic functions needed by many Bioconductor
5902packages.")
5903 (license license:artistic2.0)))
5904
eb24341f
RJ		 5905(define-public r-biocinstaller
5906 (package
5907 (name "r-biocinstaller")
5c7b04e0 5908 (version "1.24.0")
eb24341f
RJ		 5909 (source (origin
5910 (method url-fetch)
5911 (uri (bioconductor-uri "BiocInstaller" version))
5912 (sha256
5913 (base32
5c7b04e0 5914 "0y1y5wmy6lzjqx3hdg15n91d417ccjj8dbvdkhmp99bs5aijwcpn"))))
eb24341f
RJ		 5915 (properties
5916 `((upstream-name . "BiocInstaller")))
5917 (build-system r-build-system)
5918 (home-page "http://bioconductor.org/packages/BiocInstaller")
5919 (synopsis "Install Bioconductor packages")
5920 (description "This package is used to install and update R packages from
5921Bioconductor, CRAN, and Github.")
5922 (license license:artistic2.0)))
5923
207ce8fb
RJ		 5924(define-public r-biocviews
5925 (package
5926 (name "r-biocviews")
5927 (version "1.42.0")
5928 (source (origin
5929 (method url-fetch)
5930 (uri (bioconductor-uri "biocViews" version))
5931 (sha256
5932 (base32
5933 "07rjk10b91pkriyq297w86199r2d3sfji3ggs9mq2gyalsa8y4b6"))))
5934 (properties
5935 `((upstream-name . "biocViews")))
5936 (build-system r-build-system)
5937 (propagated-inputs
5938 `(("r-biobase" ,r-biobase)
5939 ("r-graph" ,r-graph)
5940 ("r-rbgl" ,r-rbgl)
5941 ("r-rcurl" ,r-rcurl)
5942 ("r-xml" ,r-xml)
5943 ("r-knitr" ,r-knitr)
5944 ("r-runit" ,r-runit)))
5945 (home-page "http://bioconductor.org/packages/biocViews")
5946 (synopsis "Bioconductor package categorization helper")
5947 (description "The purpose of biocViews is to create HTML pages that
5948categorize packages in a Bioconductor package repository according to keywords,
5949also known as views, in a controlled vocabulary.")
5950 (license license:artistic2.0)))
5951
99df12cd
RJ		 5952(define-public r-biocstyle
5953 (package
5954 (name "r-biocstyle")
5955 (version "2.2.0")
5956 (source (origin
5957 (method url-fetch)
5958 (uri (bioconductor-uri "BiocStyle" version))
5959 (sha256
5960 (base32
5961 "0qbk23fz8cn260isd9xlh9lxfj4adar6iqzai01c4kz0p31f45za"))))
5962 (properties
5963 `((upstream-name . "BiocStyle")))
5964 (build-system r-build-system)
5965 (home-page "http://bioconductor.org/packages/BiocStyle")
5966 (synopsis "Bioconductor formatting styles")
5967 (description "This package provides standard formatting styles for
5968Bioconductor PDF and HTML documents. Package vignettes illustrate use and
5969functionality.")
5970 (license license:artistic2.0)))
5971
4644644a
RJ		 5972(define-public r-bioccheck
5973 (package
5974 (name "r-bioccheck")
5975 (version "1.10.0")
5976 (source (origin
5977 (method url-fetch)
5978 (uri (bioconductor-uri "BiocCheck" version))
5979 (sha256
5980 (base32
5981 "1rfy37xg1nc2cmgbclvzsi7sgmdcdjiahsx9crgx3yaw7kxgiack"))))
5982 (properties
5983 `((upstream-name . "BiocCheck")))
5984 (build-system r-build-system)
5985 (arguments
5986 '(#:phases
5987 (modify-phases %standard-phases
5988 ;; This package can be used by calling BiocCheck(<package>) from
5989 ;; within R, or by running R CMD BiocCheck <package>. This phase
5990 ;; makes sure the latter works. For this to work, the BiocCheck
5991 ;; script must be somewhere on the PATH (not the R bin directory).
5992 (add-after 'install 'install-bioccheck-subcommand
5993 (lambda* (#:key outputs #:allow-other-keys)
5994 (let* ((out (assoc-ref outputs "out"))
5995 (dest-dir (string-append out "/bin"))
5996 (script-dir
5997 (string-append out "/site-library/BiocCheck/script/")))
5998 (mkdir-p dest-dir)
5999 (symlink (string-append script-dir "/checkBadDeps.R")
6000 (string-append dest-dir "/checkBadDeps.R"))
6001 (symlink (string-append script-dir "/BiocCheck")
6002 (string-append dest-dir "/BiocCheck")))
6003 #t)))))
6004 (native-inputs
6005 `(("which" ,which)))
6006 (propagated-inputs
6007 `(("r-graph" ,r-graph)
6008 ("r-knitr" ,r-knitr)
6009 ("r-httr" ,r-httr)
6010 ("r-optparse" ,r-optparse)
6011 ("r-devtools" ,r-devtools)
6012 ("r-biocinstaller" ,r-biocinstaller)
6013 ("r-biocviews" ,r-biocviews)))
6014 (home-page "http://bioconductor.org/packages/BiocCheck")
6015 (synopsis "Executes Bioconductor-specific package checks")
6016 (description "This package contains tools to perform additional quality
6017checks on R packages that are to be submitted to the Bioconductor repository.")
6018 (license license:artistic2.0)))
6019
2acaaee5
RJ		 6020(define-public r-getopt
6021 (package
6022 (name "r-getopt")
6023 (version "1.20.0")
6024 (source
6025 (origin
6026 (method url-fetch)
6027 (uri (cran-uri "getopt" version))
6028 (sha256
6029 (base32
6030 "00f57vgnzmg7cz80rjmjz1556xqcmx8nhrlbbhaq4w7gl2ibl87r"))))
6031 (build-system r-build-system)
6032 (home-page "https://github.com/trevorld/getopt")
6033 (synopsis "Command-line option processor for R")
6034 (description
6035 "This package is designed to be used with Rscript to write shebang
6036scripts that accept short and long options. Many users will prefer to
6037use the packages @code{optparse} or @code{argparse} which add extra
6038features like automatically generated help options and usage texts,
6039support for default values, positional argument support, etc.")
6040 (license license:gpl2+)))
6041
c79ad57a
RJ		 6042(define-public r-optparse
6043 (package
6044 (name "r-optparse")
6045 (version "1.3.2")
6046 (source
6047 (origin
6048 (method url-fetch)
6049 (uri (cran-uri "optparse" version))
6050 (sha256
6051 (base32
6052 "1g8as89r91xxi5j5azsd6vrfrhg84mnfx2683j7pacdp8s33radw"))))
6053 (build-system r-build-system)
6054 (propagated-inputs
6055 `(("r-getopt" ,r-getopt)))
6056 (home-page
6057 "https://github.com/trevorld/optparse")
6058 (synopsis "Command line option parser")
6059 (description
6060 "This package provides a command line parser inspired by Python's
6061@code{optparse} library to be used with Rscript to write shebang scripts
6062that accept short and long options.")
6063 (license license:gpl2+)))
6064
247d498a
RJ		 6065(define-public r-dnacopy
6066 (package
6067 (name "r-dnacopy")
22c5d736 6068 (version "1.48.0")
247d498a
RJ		 6069 (source (origin
6070 (method url-fetch)
6071 (uri (bioconductor-uri "DNAcopy" version))
6072 (sha256
6073 (base32
22c5d736 6074 "1idyvfvy7xx8k9vk00y4k3819qmip8iqm809j3vpxabmsn7r9zyh"))))
247d498a
RJ		 6075 (properties
6076 `((upstream-name . "DNAcopy")))
6077 (build-system r-build-system)
6078 (inputs
6079 `(("gfortran" ,gfortran)))
5697fdc3 6080 (home-page "https://bioconductor.org/packages/DNAcopy")
247d498a
RJ		 6081 (synopsis "Implementation of a circular binary segmentation algorithm")
6082 (description "This package implements the circular binary segmentation (CBS)
6083algorithm to segment DNA copy number data and identify genomic regions with
6084abnormal copy number.")
6085 (license license:gpl2+)))
6086
7485129e
RW		 6087(define-public r-s4vectors
6088 (package
6089 (name "r-s4vectors")
b120c5f3 6090 (version "0.12.0")
7485129e
RW		 6091 (source (origin
6092 (method url-fetch)
6093 (uri (bioconductor-uri "S4Vectors" version))
6094 (sha256
6095 (base32
b120c5f3 6096 "0m0npc0vhmcwcxws7v2f8k4hvvrjvnlrsr94klxf4a8m4xw2xzzk"))))
7485129e 6097 (properties
1d216b6e 6098 `((upstream-name . "S4Vectors")))
7485129e
RW		 6099 (build-system r-build-system)
6100 (propagated-inputs
6101 `(("r-biocgenerics" ,r-biocgenerics)))
6102 (home-page "http://bioconductor.org/packages/S4Vectors")
6103 (synopsis "S4 implementation of vectors and lists")
6104 (description
6105 "The S4Vectors package defines the @code{Vector} and @code{List} virtual
6106classes and a set of generic functions that extend the semantic of ordinary
6107vectors and lists in R. Package developers can easily implement vector-like
6108or list-like objects as concrete subclasses of @code{Vector} or @code{List}.
6109In addition, a few low-level concrete subclasses of general interest (e.g.
6110@code{DataFrame}, @code{Rle}, and @code{Hits}) are implemented in the
6111S4Vectors package itself.")
6112 (license license:artistic2.0)))
6113
274da826
RW		 6114(define-public r-seqinr
6115 (package
6116 (name "r-seqinr")
a46cc723 6117 (version "3.3-3")
274da826
RW		 6118 (source
6119 (origin
6120 (method url-fetch)
6121 (uri (cran-uri "seqinr" version))
6122 (sha256
6123 (base32
a46cc723 6124 "0rk4yba8km26c0rh1f4h474zsb5n6kjmqsi55bnzr6p8pymp18hj"))))
274da826
RW		 6125 (build-system r-build-system)
6126 (propagated-inputs
3b851cd4
RW		 6127 `(("r-ade4" ,r-ade4)
6128 ("r-segmented" ,r-segmented)))
274da826
RW		 6129 (inputs
6130 `(("zlib" ,zlib)))
6131 (home-page "http://seqinr.r-forge.r-project.org/")
6132 (synopsis "Biological sequences retrieval and analysis")
6133 (description
6134 "This package provides tools for exploratory data analysis and data
6135visualization of biological sequence (DNA and protein) data. It also includes
6136utilities for sequence data management under the ACNUC system.")
6137 (license license:gpl2+)))
6138
78addcb0
RW		 6139(define-public r-iranges
6140 (package
6141 (name "r-iranges")
7f354daf 6142 (version "2.8.0")
78addcb0
RW		 6143 (source (origin
6144 (method url-fetch)
6145 (uri (bioconductor-uri "IRanges" version))
6146 (sha256
6147 (base32
7f354daf 6148 "0cdl1sfd3cvf93lnz91fdk64fbg1mnd5g958dwh1il8r358hqq3f"))))
78addcb0 6149 (properties
1d216b6e 6150 `((upstream-name . "IRanges")))
78addcb0
RW		 6151 (build-system r-build-system)
6152 (propagated-inputs
6153 `(("r-biocgenerics" ,r-biocgenerics)
6154 ("r-s4vectors" ,r-s4vectors)))
6155 (home-page "http://bioconductor.org/packages/IRanges")
6156 (synopsis "Infrastructure for manipulating intervals on sequences")
6157 (description
6158 "This package provides efficient low-level and highly reusable S4 classes
6159for storing ranges of integers, RLE vectors (Run-Length Encoding), and, more
6160generally, data that can be organized sequentially (formally defined as
6161@code{Vector} objects), as well as views on these @code{Vector} objects.
6162Efficient list-like classes are also provided for storing big collections of
6163instances of the basic classes. All classes in the package use consistent
6164naming and share the same rich and consistent \"Vector API\" as much as
6165possible.")
6166 (license license:artistic2.0)))
6167
bf7764b7
RW		 6168(define-public r-genomeinfodb
6169 (package
6170 (name "r-genomeinfodb")
dece310a 6171 (version "1.10.0")
bf7764b7
RW		 6172 (source (origin
6173 (method url-fetch)
6174 (uri (bioconductor-uri "GenomeInfoDb" version))
6175 (sha256
6176 (base32
dece310a 6177 "0nhg4bk38gzvf3mvnbqgisbbhfv1kzjld27z1z9knnlkplkiyyyv"))))
bf7764b7 6178 (properties
1d216b6e 6179 `((upstream-name . "GenomeInfoDb")))
bf7764b7
RW		 6180 (build-system r-build-system)
6181 (propagated-inputs
6182 `(("r-biocgenerics" ,r-biocgenerics)
6183 ("r-iranges" ,r-iranges)
6184 ("r-s4vectors" ,r-s4vectors)))
6185 (home-page "http://bioconductor.org/packages/GenomeInfoDb")
6186 (synopsis "Utilities for manipulating chromosome identifiers")
6187 (description
6188 "This package contains data and functions that define and allow
6189translation between different chromosome sequence naming conventions (e.g.,
6190\"chr1\" versus \"1\"), including a function that attempts to place sequence
6191names in their natural, rather than lexicographic, order.")
6192 (license license:artistic2.0)))
6193
744004a3
RJ		 6194(define-public r-edger
6195 (package
6196 (name "r-edger")
ed305423 6197 (version "3.16.1")
744004a3
RJ		 6198 (source (origin
6199 (method url-fetch)
6200 (uri (bioconductor-uri "edgeR" version))
6201 (sha256
6202 (base32
ed305423 6203 "1r6hhwkqp13m022hjajzr1lnjsbai0yjhykwn0kp1f0la990a808"))))
744004a3
RJ		 6204 (properties `((upstream-name . "edgeR")))
6205 (build-system r-build-system)
6206 (propagated-inputs
5e48005f
RW		 6207 `(("r-limma" ,r-limma)
6208 ("r-locfit" ,r-locfit)))
744004a3
RJ		 6209 (home-page "http://bioinf.wehi.edu.au/edgeR")
6210 (synopsis "EdgeR does empirical analysis of digital gene expression data")
6211 (description "This package can do differential expression analysis of
6212RNA-seq expression profiles with biological replication. It implements a range
6213of statistical methodology based on the negative binomial distributions,
6214including empirical Bayes estimation, exact tests, generalized linear models
6215and quasi-likelihood tests. It be applied to differential signal analysis of
6216other types of genomic data that produce counts, including ChIP-seq, SAGE and
6217CAGE.")
6218 (license license:gpl2+)))
6219
b669d9c4
RJ		 6220(define-public r-variantannotation
6221 (package
6222 (name "r-variantannotation")
37d96f1d 6223 (version "1.20.0")
b669d9c4
RJ		 6224 (source (origin
6225 (method url-fetch)
6226 (uri (bioconductor-uri "VariantAnnotation" version))
6227 (sha256
6228 (base32
37d96f1d 6229 "1lwzfgahz8ipwli73kcfqb18y6adi129hap1gnycnj3980m54i8q"))))
b669d9c4
RJ		 6230 (properties
6231 `((upstream-name . "VariantAnnotation")))
6232 (inputs
6233 `(("zlib" ,zlib)))
6234 (propagated-inputs
6235 `(("r-annotationdbi" ,r-annotationdbi)
37d96f1d 6236 ("r-biobase" ,r-biobase)
b669d9c4 6237 ("r-biocgenerics" ,r-biocgenerics)
37d96f1d 6238 ("r-biostrings" ,r-biostrings)
b669d9c4
RJ		 6239 ("r-bsgenome" ,r-bsgenome)
6240 ("r-dbi" ,r-dbi)
6241 ("r-genomeinfodb" ,r-genomeinfodb)
6242 ("r-genomicfeatures" ,r-genomicfeatures)
6243 ("r-genomicranges" ,r-genomicranges)
37d96f1d 6244 ("r-iranges" ,r-iranges)
b669d9c4
RJ		 6245 ("r-summarizedexperiment" ,r-summarizedexperiment)
6246 ("r-rsamtools" ,r-rsamtools)
37d96f1d
RW		 6247 ("r-rtracklayer" ,r-rtracklayer)
6248 ("r-s4vectors" ,r-s4vectors)
6249 ("r-xvector" ,r-xvector)
b669d9c4
RJ		 6250 ("r-zlibbioc" ,r-zlibbioc)))
6251 (build-system r-build-system)
6252 (home-page "https://bioconductor.org/packages/VariantAnnotation")
6253 (synopsis "Package for annotation of genetic variants")
6254 (description "This R package can annotate variants, compute amino acid
6255coding changes and predict coding outcomes.")
6256 (license license:artistic2.0)))
6257
7d4224d7
RJ		 6258(define-public r-limma
6259 (package
6260 (name "r-limma")
5b3b2d2f 6261 (version "3.30.2")
7d4224d7
RJ		 6262 (source (origin
6263 (method url-fetch)
6264 (uri (bioconductor-uri "limma" version))
6265 (sha256
6266 (base32
5b3b2d2f 6267 "04jris7wk2lxksrrvrjsysznsdb2k04lfgrnp18ic49sazva0hfy"))))
7d4224d7
RJ		 6268 (build-system r-build-system)
6269 (home-page "http://bioinf.wehi.edu.au/limma")
6270 (synopsis "Package for linear models for microarray and RNA-seq data")
6271 (description "This package can be used for the analysis of gene expression
6272studies, especially the use of linear models for analysing designed experiments
6273and the assessment of differential expression. The analysis methods apply to
6274different technologies, including microarrays, RNA-seq, and quantitative PCR.")
6275 (license license:gpl2+)))
6276
0e7d5560
RW		 6277(define-public r-xvector
6278 (package
6279 (name "r-xvector")
1d048589 6280 (version "0.14.0")
0e7d5560
RW		 6281 (source (origin
6282 (method url-fetch)
6283 (uri (bioconductor-uri "XVector" version))
6284 (sha256
6285 (base32
1d048589 6286 "09lbqxpqr80g0kw77mpz0p1a8cq706j33kz8194wp71il67cdzi7"))))
0e7d5560 6287 (properties
1d216b6e 6288 `((upstream-name . "XVector")))
0e7d5560
RW		 6289 (build-system r-build-system)
6290 (arguments
6291 `(#:phases
6292 (modify-phases %standard-phases
6293 (add-after 'unpack 'use-system-zlib
6294 (lambda _
6295 (substitute* "DESCRIPTION"
6296 (("zlibbioc, ") ""))
6297 (substitute* "NAMESPACE"
6298 (("import\\(zlibbioc\\)") ""))
6299 #t)))))
6300 (inputs
6301 `(("zlib" ,zlib)))
6302 (propagated-inputs
6303 `(("r-biocgenerics" ,r-biocgenerics)
6304 ("r-iranges" ,r-iranges)
6305 ("r-s4vectors" ,r-s4vectors)))
6306 (home-page "http://bioconductor.org/packages/XVector")
6307 (synopsis "Representation and manpulation of external sequences")
6308 (description
6309 "This package provides memory efficient S4 classes for storing sequences
6310\"externally\" (behind an R external pointer, or on disk).")
6311 (license license:artistic2.0)))
6312
e2cd1d0f
RW		 6313(define-public r-genomicranges
6314 (package
6315 (name "r-genomicranges")
92a740af 6316 (version "1.26.1")
e2cd1d0f
RW		 6317 (source (origin
6318 (method url-fetch)
6319 (uri (bioconductor-uri "GenomicRanges" version))
6320 (sha256
6321 (base32
92a740af 6322 "039nxccg9i2an8q2wni79x8dr9p1fcfcqvih9hg9w243pczg2g3c"))))
e2cd1d0f 6323 (properties
1d216b6e 6324 `((upstream-name . "GenomicRanges")))
e2cd1d0f
RW		 6325 (build-system r-build-system)
6326 (propagated-inputs
6327 `(("r-biocgenerics" ,r-biocgenerics)
6328 ("r-genomeinfodb" ,r-genomeinfodb)
92a740af
RW		 6329 ("r-iranges" ,r-iranges)
6330 ("r-s4vectors" ,r-s4vectors)
e2cd1d0f
RW		 6331 ("r-xvector" ,r-xvector)))
6332 (home-page "http://bioconductor.org/packages/GenomicRanges")
6333 (synopsis "Representation and manipulation of genomic intervals")
6334 (description
6335 "This package provides tools to efficiently represent and manipulate
6336genomic annotations and alignments is playing a central role when it comes to
6337analyzing high-throughput sequencing data (a.k.a. NGS data). The
6338GenomicRanges package defines general purpose containers for storing and
6339manipulating genomic intervals and variables defined along a genome.")
6340 (license license:artistic2.0)))
6341
555e3399
RW		 6342(define-public r-biobase
6343 (package
6344 (name "r-biobase")
464870ca 6345 (version "2.34.0")
555e3399
RW		 6346 (source (origin
6347 (method url-fetch)
6348 (uri (bioconductor-uri "Biobase" version))
6349 (sha256
6350 (base32
464870ca 6351 "0js9j9wqls8f571ifl9ylllbb9a9hwf7b7drf2grwb1fl31ldazl"))))
555e3399
RW		 6352 (properties
6353 `((upstream-name . "Biobase")))
6354 (build-system r-build-system)
6355 (propagated-inputs
6356 `(("r-biocgenerics" ,r-biocgenerics)))
6357 (home-page "http://bioconductor.org/packages/Biobase")
6358 (synopsis "Base functions for Bioconductor")
6359 (description
6360 "This package provides functions that are needed by many other packages
6361on Bioconductor or which replace R functions.")
6362 (license license:artistic2.0)))
6363
8b7bce74
RW		 6364(define-public r-annotationdbi
6365 (package
6366 (name "r-annotationdbi")
e45dc057 6367 (version "1.36.0")
8b7bce74
RW		 6368 (source (origin
6369 (method url-fetch)
6370 (uri (bioconductor-uri "AnnotationDbi" version))
6371 (sha256
6372 (base32
e45dc057 6373 "0ydrqw1k1j5p6w76bwc753cx545c055x88q87wzya93858synj6r"))))
8b7bce74
RW		 6374 (properties
6375 `((upstream-name . "AnnotationDbi")))
6376 (build-system r-build-system)
6377 (propagated-inputs
6378 `(("r-biobase" ,r-biobase)
6379 ("r-biocgenerics" ,r-biocgenerics)
6380 ("r-dbi" ,r-dbi)
6381 ("r-iranges" ,r-iranges)
6382 ("r-rsqlite" ,r-rsqlite)
6383 ("r-s4vectors" ,r-s4vectors)))
6384 (home-page "http://bioconductor.org/packages/AnnotationDbi")
6385 (synopsis "Annotation database interface")
6386 (description
6387 "This package provides user interface and database connection code for
6388annotation data packages using SQLite data storage.")
6389 (license license:artistic2.0)))
6390
c465fa72
RW		 6391(define-public r-biomart
6392 (package
6393 (name "r-biomart")
94053331 6394 (version "2.30.0")
c465fa72
RW		 6395 (source (origin
6396 (method url-fetch)
6397 (uri (bioconductor-uri "biomaRt" version))
6398 (sha256
6399 (base32
94053331 6400 "1x0flcghq71784q2l02j0g4f9jkmyb14f6i307n6c59d6ji7h7x6"))))
c465fa72
RW		 6401 (properties
6402 `((upstream-name . "biomaRt")))
6403 (build-system r-build-system)
6404 (propagated-inputs
6405 `(("r-annotationdbi" ,r-annotationdbi)
6406 ("r-rcurl" ,r-rcurl)
6407 ("r-xml" ,r-xml)))
6408 (home-page "http://bioconductor.org/packages/biomaRt")
6409 (synopsis "Interface to BioMart databases")
6410 (description
6411 "biomaRt provides an interface to a growing collection of databases
6412implementing the @url{BioMart software suite, http://www.biomart.org}. The
6413package enables retrieval of large amounts of data in a uniform way without
6414the need to know the underlying database schemas or write complex SQL queries.
6415Examples of BioMart databases are Ensembl, COSMIC, Uniprot, HGNC, Gramene,
6416Wormbase and dbSNP mapped to Ensembl. These major databases give biomaRt
6417users direct access to a diverse set of data and enable a wide range of
6418powerful online queries from gene annotation to database mining.")
6419 (license license:artistic2.0)))
6420
e91d362e
RW		 6421(define-public r-biocparallel
6422 (package
6423 (name "r-biocparallel")
c15ae9b0 6424 (version "1.8.1")
e91d362e
RW		 6425 (source (origin
6426 (method url-fetch)
6427 (uri (bioconductor-uri "BiocParallel" version))
6428 (sha256
6429 (base32
c15ae9b0 6430 "123i928rwi4h4sy4fpysv6pinw5nns0sm3myxi2ghqhm34ws8gyl"))))
e91d362e
RW		 6431 (properties
6432 `((upstream-name . "BiocParallel")))
6433 (build-system r-build-system)
6434 (propagated-inputs
6435 `(("r-futile-logger" ,r-futile-logger)
6436 ("r-snow" ,r-snow)))
6437 (home-page "http://bioconductor.org/packages/BiocParallel")
6438 (synopsis "Bioconductor facilities for parallel evaluation")
6439 (description
6440 "This package provides modified versions and novel implementation of
6441functions for parallel evaluation, tailored to use with Bioconductor
6442objects.")
6443 (license (list license:gpl2+ license:gpl3+))))
6444
bf159353
RW		 6445(define-public r-biostrings
6446 (package
6447 (name "r-biostrings")
47d39b6e 6448 (version "2.42.0")
bf159353
RW		 6449 (source (origin
6450 (method url-fetch)
6451 (uri (bioconductor-uri "Biostrings" version))
6452 (sha256
6453 (base32
47d39b6e 6454 "08z8lkz3axa94wkf144a931ry6vf6cc25avi1ywr84ln2k5czz9f"))))
bf159353
RW		 6455 (properties
6456 `((upstream-name . "Biostrings")))
6457 (build-system r-build-system)
6458 (propagated-inputs
6459 `(("r-biocgenerics" ,r-biocgenerics)
6460 ("r-iranges" ,r-iranges)
6461 ("r-s4vectors" ,r-s4vectors)
6462 ("r-xvector" ,r-xvector)))
6463 (home-page "http://bioconductor.org/packages/Biostrings")
6464 (synopsis "String objects and algorithms for biological sequences")
6465 (description
6466 "This package provides memory efficient string containers, string
6467matching algorithms, and other utilities, for fast manipulation of large
6468biological sequences or sets of sequences.")
6469 (license license:artistic2.0)))
6470
f8d74f70
RW		 6471(define-public r-rsamtools
6472 (package
6473 (name "r-rsamtools")
c0ccef41 6474 (version "1.26.1")
f8d74f70
RW		 6475 (source (origin
6476 (method url-fetch)
6477 (uri (bioconductor-uri "Rsamtools" version))
6478 (sha256
6479 (base32
c0ccef41 6480 "0pf4f6brf4bl5zgjrah0f38qslazrs49ayqgyh0xfqgrh63yx4ck"))))
f8d74f70
RW		 6481 (properties
6482 `((upstream-name . "Rsamtools")))
6483 (build-system r-build-system)
6484 (arguments
6485 `(#:phases
6486 (modify-phases %standard-phases
6487 (add-after 'unpack 'use-system-zlib
6488 (lambda _
6489 (substitute* "DESCRIPTION"
6490 (("zlibbioc, ") ""))
6491 (substitute* "NAMESPACE"
6492 (("import\\(zlibbioc\\)") ""))
6493 #t)))))
6494 (inputs
6495 `(("zlib" ,zlib)))
6496 (propagated-inputs
6497 `(("r-biocgenerics" ,r-biocgenerics)
6498 ("r-biocparallel" ,r-biocparallel)
6499 ("r-biostrings" ,r-biostrings)
6500 ("r-bitops" ,r-bitops)
6501 ("r-genomeinfodb" ,r-genomeinfodb)
6502 ("r-genomicranges" ,r-genomicranges)
6503 ("r-iranges" ,r-iranges)
6504 ("r-s4vectors" ,r-s4vectors)
6505 ("r-xvector" ,r-xvector)))
6506 (home-page "http://bioconductor.org/packages/release/bioc/html/Rsamtools.html")
6507 (synopsis "Interface to samtools, bcftools, and tabix")
6508 (description
6509 "This package provides an interface to the 'samtools', 'bcftools', and
6510'tabix' utilities for manipulating SAM (Sequence Alignment / Map), FASTA,
6511binary variant call (BCF) and compressed indexed tab-delimited (tabix)
6512files.")
6513 (license license:expat)))
6514
6e76dda2
RW		 6515(define-public r-summarizedexperiment
6516 (package
6517 (name "r-summarizedexperiment")
08da08d2 6518 (version "1.4.0")
6e76dda2
RW		 6519 (source (origin
6520 (method url-fetch)
6521 (uri (bioconductor-uri "SummarizedExperiment" version))
6522 (sha256
6523 (base32
08da08d2 6524 "1kbj8sg2ik9f8d6g95wz0py62jldg01qy5rsdpg1cxw95nf7dzi3"))))
6e76dda2
RW		 6525 (properties
6526 `((upstream-name . "SummarizedExperiment")))
6527 (build-system r-build-system)
6528 (propagated-inputs
6529 `(("r-biobase" ,r-biobase)
6530 ("r-biocgenerics" ,r-biocgenerics)
6531 ("r-genomeinfodb" ,r-genomeinfodb)
6532 ("r-genomicranges" ,r-genomicranges)
6533 ("r-iranges" ,r-iranges)
6534 ("r-s4vectors" ,r-s4vectors)))
6535 (home-page "http://bioconductor.org/packages/SummarizedExperiment")
6536 (synopsis "Container for representing genomic ranges by sample")
6537 (description
6538 "The SummarizedExperiment container contains one or more assays, each
6539represented by a matrix-like object of numeric or other mode. The rows
6540typically represent genomic ranges of interest and the columns represent
6541samples.")
6542 (license license:artistic2.0)))
6543
d8a828af
RW		 6544(define-public r-genomicalignments
6545 (package
6546 (name "r-genomicalignments")
4ab32f0c 6547 (version "1.10.0")
d8a828af
RW		 6548 (source (origin
6549 (method url-fetch)
6550 (uri (bioconductor-uri "GenomicAlignments" version))
6551 (sha256
6552 (base32
4ab32f0c 6553 "11vb0a0zd36i4yhg4mfijv787v0nihn6pkjj6q7rfy19gwy61xlc"))))
d8a828af
RW		 6554 (properties
6555 `((upstream-name . "GenomicAlignments")))
6556 (build-system r-build-system)
6557 (propagated-inputs
6558 `(("r-biocgenerics" ,r-biocgenerics)
6559 ("r-biocparallel" ,r-biocparallel)
6560 ("r-biostrings" ,r-biostrings)
6561 ("r-genomeinfodb" ,r-genomeinfodb)
6562 ("r-genomicranges" ,r-genomicranges)
6563 ("r-iranges" ,r-iranges)
6564 ("r-rsamtools" ,r-rsamtools)
6565 ("r-s4vectors" ,r-s4vectors)
6566 ("r-summarizedexperiment" ,r-summarizedexperiment)))
6567 (home-page "http://bioconductor.org/packages/GenomicAlignments")
6568 (synopsis "Representation and manipulation of short genomic alignments")
6569 (description
6570 "This package provides efficient containers for storing and manipulating
6571short genomic alignments (typically obtained by aligning short reads to a
6572reference genome). This includes read counting, computing the coverage,
6573junction detection, and working with the nucleotide content of the
6574alignments.")
6575 (license license:artistic2.0)))
6576
317755ff
RW		 6577(define-public r-rtracklayer
6578 (package
6579 (name "r-rtracklayer")
16a8bd31 6580 (version "1.34.1")
317755ff
RW		 6581 (source (origin
6582 (method url-fetch)
6583 (uri (bioconductor-uri "rtracklayer" version))
6584 (sha256
6585 (base32
16a8bd31 6586 "0x59k2fd0iaqi93gy6bm58p2j2z90z1b7a6w5b4c098y98n096rc"))))
317755ff
RW		 6587 (build-system r-build-system)
6588 (arguments
6589 `(#:phases
6590 (modify-phases %standard-phases
6591 (add-after 'unpack 'use-system-zlib
6592 (lambda _
6593 (substitute* "DESCRIPTION"
6594 (("zlibbioc, ") ""))
6595 (substitute* "NAMESPACE"
6596 (("import\\(zlibbioc\\)") ""))
6597 #t)))))
6598 (inputs
6599 `(("zlib" ,zlib)))
6600 (propagated-inputs
6601 `(("r-biocgenerics" ,r-biocgenerics)
6602 ("r-biostrings" ,r-biostrings)
6603 ("r-genomeinfodb" ,r-genomeinfodb)
6604 ("r-genomicalignments" ,r-genomicalignments)
6605 ("r-genomicranges" ,r-genomicranges)
6606 ("r-iranges" ,r-iranges)
6607 ("r-rcurl" ,r-rcurl)
6608 ("r-rsamtools" ,r-rsamtools)
6609 ("r-s4vectors" ,r-s4vectors)
6610 ("r-xml" ,r-xml)
6611 ("r-xvector" ,r-xvector)))
6612 (home-page "http://bioconductor.org/packages/rtracklayer")
6613 (synopsis "R interface to genome browsers and their annotation tracks")
6614 (description
6615 "rtracklayer is an extensible framework for interacting with multiple
6616genome browsers (currently UCSC built-in) and manipulating annotation tracks
6617in various formats (currently GFF, BED, bedGraph, BED15, WIG, BigWig and 2bit
6618built-in). The user may export/import tracks to/from the supported browsers,
6619as well as query and modify the browser state, such as the current viewport.")
6620 (license license:artistic2.0)))
6621
2fd7c049
RW		 6622(define-public r-genomicfeatures
6623 (package
6624 (name "r-genomicfeatures")
b17aac08 6625 (version "1.26.0")
2fd7c049
RW		 6626 (source (origin
6627 (method url-fetch)
6628 (uri (bioconductor-uri "GenomicFeatures" version))
6629 (sha256
6630 (base32
b17aac08 6631 "0z8spi2knwzwi10c38vr7xlvi3ah9faj7m1lka880mmxkl9cai4k"))))
2fd7c049
RW		 6632 (properties
6633 `((upstream-name . "GenomicFeatures")))
6634 (build-system r-build-system)
6635 (propagated-inputs
6636 `(("r-annotationdbi" ,r-annotationdbi)
6637 ("r-biobase" ,r-biobase)
6638 ("r-biocgenerics" ,r-biocgenerics)
6639 ("r-biomart" ,r-biomart)
6640 ("r-biostrings" ,r-biostrings)
6641 ("r-dbi" ,r-dbi)
6642 ("r-genomeinfodb" ,r-genomeinfodb)
6643 ("r-genomicranges" ,r-genomicranges)
6644 ("r-iranges" ,r-iranges)
6645 ("r-rcurl" ,r-rcurl)
6646 ("r-rsqlite" ,r-rsqlite)
6647 ("r-rtracklayer" ,r-rtracklayer)
6648 ("r-s4vectors" ,r-s4vectors)
6649 ("r-xvector" ,r-xvector)))
6650 (home-page "http://bioconductor.org/packages/GenomicFeatures")
6651 (synopsis "Tools for working with transcript centric annotations")
6652 (description
6653 "This package provides a set of tools and methods for making and
6654manipulating transcript centric annotations. With these tools the user can
6655easily download the genomic locations of the transcripts, exons and cds of a
6656given organism, from either the UCSC Genome Browser or a BioMart
6657database (more sources will be supported in the future). This information is
6658then stored in a local database that keeps track of the relationship between
6659transcripts, exons, cds and genes. Flexible methods are provided for
6660extracting the desired features in a convenient format.")
6661 (license license:artistic2.0)))
6662
fb25d880
RW		 6663(define-public r-go-db
6664 (package
6665 (name "r-go-db")
d1b1587c 6666 (version "3.4.0")
fb25d880
RW		 6667 (source (origin
6668 (method url-fetch)
f82c8c3c
PP		 6669 (uri (string-append "http://www.bioconductor.org/packages/"
6670 "release/data/annotation/src/contrib/GO.db_"
6671 version ".tar.gz"))
fb25d880
RW		 6672 (sha256
6673 (base32
d1b1587c 6674 "02cj8kqi5w39jwcs8gp1dgj08sah262ppxnkz4h3qd0w191y8yyl"))))
fb25d880
RW		 6675 (properties
6676 `((upstream-name . "GO.db")))
6677 (build-system r-build-system)
3141b83d
RW		 6678 (propagated-inputs
6679 `(("r-annotationdbi" ,r-annotationdbi)))
fb25d880
RW		 6680 (home-page "http://bioconductor.org/packages/GO.db")
6681 (synopsis "Annotation maps describing the entire Gene Ontology")
6682 (description
6683 "The purpose of this GO.db annotation package is to provide detailed
6684information about the latest version of the Gene Ontologies.")
6685 (license license:artistic2.0)))
6686
d1dbde6a
RW		 6687(define-public r-graph
6688 (package
6689 (name "r-graph")
3587b464 6690 (version "1.52.0")
d1dbde6a
RW		 6691 (source (origin
6692 (method url-fetch)
6693 (uri (bioconductor-uri "graph" version))
6694 (sha256
6695 (base32
3587b464 6696 "0g3dk5vsdp489fmyg8mifczmzgqrjlakkkr8i96dj15gghp3l135"))))
d1dbde6a
RW		 6697 (build-system r-build-system)
6698 (propagated-inputs
6699 `(("r-biocgenerics" ,r-biocgenerics)))
6700 (home-page "http://bioconductor.org/packages/graph")
6701 (synopsis "Handle graph data structures in R")
6702 (description
6703 "This package implements some simple graph handling capabilities for R.")
6704 (license license:artistic2.0)))
6705
d547ce5e
RW		 6706(define-public r-topgo
6707 (package
6708 (name "r-topgo")
30ec4de7 6709 (version "2.26.0")
d547ce5e
RW		 6710 (source (origin
6711 (method url-fetch)
6712 (uri (bioconductor-uri "topGO" version))
6713 (sha256
6714 (base32
30ec4de7 6715 "0j6sgvam4lk9348ag6pypcbkv93x4fk0di8ivhr23mz2s2yqzwrx"))))
d547ce5e
RW		 6716 (properties
6717 `((upstream-name . "topGO")))
6718 (build-system r-build-system)
6719 (propagated-inputs
6720 `(("r-annotationdbi" ,r-annotationdbi)
30ec4de7 6721 ("r-dbi" ,r-dbi)
d547ce5e
RW		 6722 ("r-biobase" ,r-biobase)
6723 ("r-biocgenerics" ,r-biocgenerics)
6724 ("r-go-db" ,r-go-db)
e48e88c2 6725 ("r-matrixstats" ,r-matrixstats)
6d415db2 6726 ("r-graph" ,r-graph)
d547ce5e
RW		 6727 ("r-sparsem" ,r-sparsem)))
6728 (home-page "http://bioconductor.org/packages/topGO")
6729 (synopsis "Enrichment analysis for gene ontology")
6730 (description
6731 "The topGO package provides tools for testing @dfn{gene ontology} (GO)
6732terms while accounting for the topology of the GO graph. Different test
6733statistics and different methods for eliminating local similarities and
6734dependencies between GO terms can be implemented and applied.")
6735 ;; Any version of the LGPL applies.
6736 (license license:lgpl2.1+)))
6737
c63cef66
RW		 6738(define-public r-bsgenome
6739 (package
6740 (name "r-bsgenome")
3d74c044 6741 (version "1.42.0")
c63cef66
RW		 6742 (source (origin
6743 (method url-fetch)
6744 (uri (bioconductor-uri "BSgenome" version))
6745 (sha256
6746 (base32
3d74c044 6747 "0hxwc02h5mzhkrk60d1jmlsfjf0ai9jxdc0128kj1sg4r2k1q94y"))))
c63cef66
RW		 6748 (properties
6749 `((upstream-name . "BSgenome")))
6750 (build-system r-build-system)
6751 (propagated-inputs
6752 `(("r-biocgenerics" ,r-biocgenerics)
6753 ("r-biostrings" ,r-biostrings)
6754 ("r-genomeinfodb" ,r-genomeinfodb)
6755 ("r-genomicranges" ,r-genomicranges)
6756 ("r-iranges" ,r-iranges)
6757 ("r-rsamtools" ,r-rsamtools)
6758 ("r-rtracklayer" ,r-rtracklayer)
6759 ("r-s4vectors" ,r-s4vectors)
6760 ("r-xvector" ,r-xvector)))
6761 (home-page "http://bioconductor.org/packages/BSgenome")
6762 (synopsis "Infrastructure for Biostrings-based genome data packages")
6763 (description
6764 "This package provides infrastructure shared by all Biostrings-based
6765genome data packages and support for efficient SNP representation.")
6766 (license license:artistic2.0)))
6767
c43a011d
RW		 6768(define-public r-impute
6769 (package
6770 (name "r-impute")
acf6f7e0 6771 (version "1.48.0")
c43a011d
RW		 6772 (source (origin
6773 (method url-fetch)
6774 (uri (bioconductor-uri "impute" version))
6775 (sha256
6776 (base32
acf6f7e0 6777 "1164zvnikbjd0ybdn9xwn520rlmdjd824vmhnl83zgv3v9lzp9bm"))))
c43a011d
RW		 6778 (inputs
6779 `(("gfortran" ,gfortran)))
6780 (build-system r-build-system)
6781 (home-page "http://bioconductor.org/packages/impute")
6782 (synopsis "Imputation for microarray data")
6783 (description
6784 "This package provides a function to impute missing gene expression
6785microarray data, using nearest neighbor averaging.")
6786 (license license:gpl2+)))
6787
03ea5a35
RW		 6788(define-public r-seqpattern
6789 (package
6790 (name "r-seqpattern")
e92dd6f5 6791 (version "1.6.0")
03ea5a35
RW		 6792 (source (origin
6793 (method url-fetch)
6794 (uri (bioconductor-uri "seqPattern" version))
6795 (sha256
6796 (base32
e92dd6f5 6797 "0lsa5pz36xapi3yiv78k3z286a5md5sm5g21pgfyg8zmhmkxr7y8"))))
03ea5a35
RW		 6798 (properties
6799 `((upstream-name . "seqPattern")))
6800 (build-system r-build-system)
6801 (propagated-inputs
6802 `(("r-biostrings" ,r-biostrings)
6803 ("r-genomicranges" ,r-genomicranges)
6804 ("r-iranges" ,r-iranges)
e92dd6f5 6805 ("r-kernsmooth" ,r-kernsmooth)
03ea5a35
RW		 6806 ("r-plotrix" ,r-plotrix)))
6807 (home-page "http://bioconductor.org/packages/seqPattern")
6808 (synopsis "Visualising oligonucleotide patterns and motif occurrences")
6809 (description
6810 "This package provides tools to visualize oligonucleotide patterns and
6811sequence motif occurrences across a large set of sequences centred at a common
6812reference point and sorted by a user defined feature.")
6813 (license license:gpl3+)))
6814
cb933df6
RW		 6815(define-public r-genomation
6816 (package
6817 (name "r-genomation")
51c3c490 6818 (version "1.6.0")
cb933df6
RW		 6819 (source (origin
6820 (method url-fetch)
6821 (uri (bioconductor-uri "genomation" version))
6822 (sha256
6823 (base32
51c3c490 6824 "1m4mz7wihj8yqivwkzw68div8ybk4rjsai3ffki7xp7sh21ax03y"))))
cb933df6
RW		 6825 (build-system r-build-system)
6826 (propagated-inputs
6827 `(("r-biostrings" ,r-biostrings)
6828 ("r-bsgenome" ,r-bsgenome)
6829 ("r-data-table" ,r-data-table)
6830 ("r-genomeinfodb" ,r-genomeinfodb)
6831 ("r-genomicalignments" ,r-genomicalignments)
6832 ("r-genomicranges" ,r-genomicranges)
6833 ("r-ggplot2" ,r-ggplot2)
6834 ("r-gridbase" ,r-gridbase)
6835 ("r-impute" ,r-impute)
6836 ("r-iranges" ,r-iranges)
6837 ("r-matrixstats" ,r-matrixstats)
6838 ("r-plotrix" ,r-plotrix)
6839 ("r-plyr" ,r-plyr)
51c3c490 6840 ("r-rcpp" ,r-rcpp)
cb933df6
RW		 6841 ("r-readr" ,r-readr)
6842 ("r-reshape2" ,r-reshape2)
51c3c490 6843 ("r-rhtslib" ,r-rhtslib)
cb933df6
RW		 6844 ("r-rsamtools" ,r-rsamtools)
6845 ("r-rtracklayer" ,r-rtracklayer)
51c3c490
RW		 6846 ("r-runit" ,r-runit)
6847 ("r-s4vectors" ,r-s4vectors)
cb933df6 6848 ("r-seqpattern" ,r-seqpattern)))
51c3c490
RW		 6849 (inputs
6850 `(("zlib" ,zlib)))
cb933df6
RW		 6851 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
6852 (synopsis "Summary, annotation and visualization of genomic data")
6853 (description
6854 "This package provides a package for summary and annotation of genomic
6855intervals. Users can visualize and quantify genomic intervals over
6856pre-defined functional regions, such as promoters, exons, introns, etc. The
6857genomic intervals represent regions with a defined chromosome position, which
6858may be associated with a score, such as aligned reads from HT-seq experiments,
6859TF binding sites, methylation scores, etc. The package can use any tabular
6860genomic feature data as long as it has minimal information on the locations of
6861genomic intervals. In addition, it can use BAM or BigWig files as input.")
6862 (license license:artistic2.0)))
6863
64efa307
RW		 6864(define-public r-genomationdata
6865 (package
6866 (name "r-genomationdata")
59198f8f 6867 (version "1.6.0")
64efa307
RW		 6868 (source (origin
6869 (method url-fetch)
6870 ;; We cannot use bioconductor-uri here because this tarball is
6871 ;; located under "data/annotation/" instead of "bioc/".
6872 (uri (string-append "https://bioconductor.org/packages/"
6873 "release/data/experiment/src/contrib/"
6874 "genomationData_" version ".tar.gz"))
6875 (sha256
6876 (base32
59198f8f 6877 "16dqwb7wx1igx77zdbcskx5m1hs4g4gp2hl56zzm70hcagnlkz8y"))))
64efa307
RW		 6878 (build-system r-build-system)
6879 ;; As this package provides little more than large data files, it doesn't
6880 ;; make sense to build substitutes.
6881 (arguments `(#:substitutable? #f))
6882 (native-inputs
6883 `(("r-knitr" ,r-knitr)))
6884 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
6885 (synopsis "Experimental data for use with the genomation package")
6886 (description
6887 "This package contains experimental genetic data for use with the
6888genomation package. Included are Chip Seq, Methylation and Cage data,
6889downloaded from Encode.")
6890 (license license:gpl3+)))
6891
486da491
RW		 6892(define-public r-org-hs-eg-db
6893 (package
6894 (name "r-org-hs-eg-db")
83f9a6fb 6895 (version "3.4.0")
486da491
RW		 6896 (source (origin
6897 (method url-fetch)
6898 ;; We cannot use bioconductor-uri here because this tarball is
6899 ;; located under "data/annotation/" instead of "bioc/".
6900 (uri (string-append "http://www.bioconductor.org/packages/"
6901 "release/data/annotation/src/contrib/"
6902 "org.Hs.eg.db_" version ".tar.gz"))
6903 (sha256
6904 (base32
83f9a6fb 6905 "19mg64pw8zcvb9yxzzyf7caz1kvdrkfsj1hd84bzq7crrh8kc4y6"))))
486da491
RW		 6906 (properties
6907 `((upstream-name . "org.Hs.eg.db")))
6908 (build-system r-build-system)
6909 (propagated-inputs
6910 `(("r-annotationdbi" ,r-annotationdbi)))
6911 (home-page "http://www.bioconductor.org/packages/org.Hs.eg.db/")
6912 (synopsis "Genome wide annotation for Human")
6913 (description
6914 "This package provides mappings from Entrez gene identifiers to various
6915annotations for the human genome.")
6916 (license license:artistic2.0)))
6917
fefedf98
RW		 6918(define-public r-org-ce-eg-db
6919 (package
6920 (name "r-org-ce-eg-db")
f04a3eff 6921 (version "3.4.0")
fefedf98
RW		 6922 (source (origin
6923 (method url-fetch)
6924 ;; We cannot use bioconductor-uri here because this tarball is
6925 ;; located under "data/annotation/" instead of "bioc/".
6926 (uri (string-append "http://www.bioconductor.org/packages/"
6927 "release/data/annotation/src/contrib/"
6928 "org.Ce.eg.db_" version ".tar.gz"))
6929 (sha256
6930 (base32
f04a3eff 6931 "12llfzrrc09kj2wzbisdspv38qzkzgpsbn8kv7qkwg746k3pq436"))))
fefedf98
RW		 6932 (properties
6933 `((upstream-name . "org.Ce.eg.db")))
6934 (build-system r-build-system)
6935 (propagated-inputs
6936 `(("r-annotationdbi" ,r-annotationdbi)))
6937 (home-page "http://www.bioconductor.org/packages/org.Ce.eg.db/")
6938 (synopsis "Genome wide annotation for Worm")
6939 (description
6940 "This package provides mappings from Entrez gene identifiers to various
6941annotations for the genome of the model worm Caenorhabditis elegans.")
6942 (license license:artistic2.0)))
6943
16c53a1e
RW		 6944(define-public r-org-dm-eg-db
6945 (package
6946 (name "r-org-dm-eg-db")
0cfac6f2 6947 (version "3.4.0")
16c53a1e
RW		 6948 (source (origin
6949 (method url-fetch)
6950 ;; We cannot use bioconductor-uri here because this tarball is
6951 ;; located under "data/annotation/" instead of "bioc/".
6952 (uri (string-append "http://www.bioconductor.org/packages/"
6953 "release/data/annotation/src/contrib/"
6954 "org.Dm.eg.db_" version ".tar.gz"))
6955 (sha256
6956 (base32
0cfac6f2 6957 "1vzbphbrh1cf7xi5cksia9xy9a9l42js2z2qsajvjxvddiphrb7j"))))
16c53a1e
RW		 6958 (properties
6959 `((upstream-name . "org.Dm.eg.db")))
6960 (build-system r-build-system)
6961 (propagated-inputs
6962 `(("r-annotationdbi" ,r-annotationdbi)))
6963 (home-page "http://www.bioconductor.org/packages/org.Dm.eg.db/")
6964 (synopsis "Genome wide annotation for Fly")
6965 (description
6966 "This package provides mappings from Entrez gene identifiers to various
6967annotations for the genome of the model fruit fly Drosophila melanogaster.")
6968 (license license:artistic2.0)))
6969
e761beb9
RW		 6970(define-public r-org-mm-eg-db
6971 (package
6972 (name "r-org-mm-eg-db")
7c45670b 6973 (version "3.4.0")
e761beb9
RW		 6974 (source (origin
6975 (method url-fetch)
6976 ;; We cannot use bioconductor-uri here because this tarball is
6977 ;; located under "data/annotation/" instead of "bioc/".
6978 (uri (string-append "http://www.bioconductor.org/packages/"
6979 "release/data/annotation/src/contrib/"
6980 "org.Mm.eg.db_" version ".tar.gz"))
6981 (sha256
6982 (base32
7c45670b 6983 "1lykjqjaf01fmgg3cvfcvwd5xjq6zc5vbxnm5r4l32fzvl89q50c"))))
e761beb9
RW		 6984 (properties
6985 `((upstream-name . "org.Mm.eg.db")))
6986 (build-system r-build-system)
6987 (propagated-inputs
6988 `(("r-annotationdbi" ,r-annotationdbi)))
6989 (home-page "http://www.bioconductor.org/packages/org.Mm.eg.db/")
6990 (synopsis "Genome wide annotation for Mouse")
6991 (description
6992 "This package provides mappings from Entrez gene identifiers to various
6993annotations for the genome of the model mouse Mus musculus.")
6994 (license license:artistic2.0)))
6995
936e7d67
RW		 6996(define-public r-seqlogo
6997 (package
6998 (name "r-seqlogo")
21d6c7a3 6999 (version "1.40.0")
936e7d67
RW		 7000 (source
7001 (origin
7002 (method url-fetch)
7003 (uri (bioconductor-uri "seqLogo" version))
7004 (sha256
7005 (base32
21d6c7a3 7006 "18bajdl75h3039559d81rgllqqvnq8ygsfxfx081xphxs0v6xggy"))))
936e7d67
RW		 7007 (properties `((upstream-name . "seqLogo")))
7008 (build-system r-build-system)
7009 (home-page "http://bioconductor.org/packages/seqLogo")
7010 (synopsis "Sequence logos for DNA sequence alignments")
7011 (description
7012 "seqLogo takes the position weight matrix of a DNA sequence motif and
7013plots the corresponding sequence logo as introduced by Schneider and
7014Stephens (1990).")
7015 (license license:lgpl2.0+)))
7016
c90a4baf
RW		 7017(define-public r-bsgenome-hsapiens-ucsc-hg19
7018 (package
7019 (name "r-bsgenome-hsapiens-ucsc-hg19")
7020 (version "1.4.0")
7021 (source (origin
7022 (method url-fetch)
7023 ;; We cannot use bioconductor-uri here because this tarball is
7024 ;; located under "data/annotation/" instead of "bioc/".
7025 (uri (string-append "http://www.bioconductor.org/packages/"
7026 "release/data/annotation/src/contrib/"
7027 "BSgenome.Hsapiens.UCSC.hg19_"
7028 version ".tar.gz"))
7029 (sha256
7030 (base32
7031 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
7032 (properties
7033 `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
7034 (build-system r-build-system)
7035 ;; As this package provides little more than a very large data file it
7036 ;; doesn't make sense to build substitutes.
7037 (arguments `(#:substitutable? #f))
7038 (propagated-inputs
7039 `(("r-bsgenome" ,r-bsgenome)))
7040 (home-page
7041 "http://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
7042 (synopsis "Full genome sequences for Homo sapiens")
7043 (description
7044 "This package provides full genome sequences for Homo sapiens as provided
7045by UCSC (hg19, February 2009) and stored in Biostrings objects.")
7046 (license license:artistic2.0)))
7047
a3e90287
RW		 7048(define-public r-bsgenome-mmusculus-ucsc-mm9
7049 (package
7050 (name "r-bsgenome-mmusculus-ucsc-mm9")
7051 (version "1.4.0")
7052 (source (origin
7053 (method url-fetch)
7054 ;; We cannot use bioconductor-uri here because this tarball is
7055 ;; located under "data/annotation/" instead of "bioc/".
7056 (uri (string-append "http://www.bioconductor.org/packages/"
7057 "release/data/annotation/src/contrib/"
7058 "BSgenome.Mmusculus.UCSC.mm9_"
7059 version ".tar.gz"))
7060 (sha256
7061 (base32
7062 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
7063 (properties
7064 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
7065 (build-system r-build-system)
7066 ;; As this package provides little more than a very large data file it
7067 ;; doesn't make sense to build substitutes.
7068 (arguments `(#:substitutable? #f))
7069 (propagated-inputs
7070 `(("r-bsgenome" ,r-bsgenome)))
7071 (home-page
7072 "http://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
7073 (synopsis "Full genome sequences for Mouse")
7074 (description
7075 "This package provides full genome sequences for Mus musculus (Mouse) as
7076provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
7077 (license license:artistic2.0)))
7078
4714d521
RW		 7079(define-public r-bsgenome-mmusculus-ucsc-mm10
7080 (package
7081 (name "r-bsgenome-mmusculus-ucsc-mm10")
7082 (version "1.4.0")
7083 (source (origin
7084 (method url-fetch)
7085 ;; We cannot use bioconductor-uri here because this tarball is
7086 ;; located under "data/annotation/" instead of "bioc/".
7087 (uri (string-append "http://www.bioconductor.org/packages/"
7088 "release/data/annotation/src/contrib/"
7089 "BSgenome.Mmusculus.UCSC.mm10_"
7090 version ".tar.gz"))
7091 (sha256
7092 (base32
7093 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
7094 (properties
7095 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
7096 (build-system r-build-system)
7097 ;; As this package provides little more than a very large data file it
7098 ;; doesn't make sense to build substitutes.
7099 (arguments `(#:substitutable? #f))
7100 (propagated-inputs
7101 `(("r-bsgenome" ,r-bsgenome)))
7102 (home-page
7103 "http://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
7104 (synopsis "Full genome sequences for Mouse")
7105 (description
7106 "This package provides full genome sequences for Mus
7107musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
7108in Biostrings objects.")
7109 (license license:artistic2.0)))
7110
943bd627
RW		 7111(define-public r-bsgenome-celegans-ucsc-ce6
7112 (package
7113 (name "r-bsgenome-celegans-ucsc-ce6")
7114 (version "1.4.0")
7115 (source (origin
7116 (method url-fetch)
7117 ;; We cannot use bioconductor-uri here because this tarball is
7118 ;; located under "data/annotation/" instead of "bioc/".
7119 (uri (string-append "http://www.bioconductor.org/packages/"
7120 "release/data/annotation/src/contrib/"
7121 "BSgenome.Celegans.UCSC.ce6_"
7122 version ".tar.gz"))
7123 (sha256
7124 (base32
7125 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
7126 (properties
7127 `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
7128 (build-system r-build-system)
7129 ;; As this package provides little more than a very large data file it
7130 ;; doesn't make sense to build substitutes.
7131 (arguments `(#:substitutable? #f))
7132 (propagated-inputs
7133 `(("r-bsgenome" ,r-bsgenome)))
7134 (home-page
7135 "http://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
7136 (synopsis "Full genome sequences for Worm")
7137 (description
7138 "This package provides full genome sequences for Caenorhabditis
7139elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
7140objects.")
7141 (license license:artistic2.0)))
7142
fc47c7d6
RW		 7143(define-public r-bsgenome-celegans-ucsc-ce10
7144 (package
7145 (name "r-bsgenome-celegans-ucsc-ce10")
7146 (version "1.4.0")
7147 (source (origin
7148 (method url-fetch)
7149 ;; We cannot use bioconductor-uri here because this tarball is
7150 ;; located under "data/annotation/" instead of "bioc/".
7151 (uri (string-append "http://www.bioconductor.org/packages/"
7152 "release/data/annotation/src/contrib/"
7153 "BSgenome.Celegans.UCSC.ce10_"
7154 version ".tar.gz"))
7155 (sha256
7156 (base32
7157 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
7158 (properties
7159 `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
7160 (build-system r-build-system)
7161 ;; As this package provides little more than a very large data file it
7162 ;; doesn't make sense to build substitutes.
7163 (arguments `(#:substitutable? #f))
7164 (propagated-inputs
7165 `(("r-bsgenome" ,r-bsgenome)))
7166 (home-page
7167 "http://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
7168 (synopsis "Full genome sequences for Worm")
7169 (description
7170 "This package provides full genome sequences for Caenorhabditis
7171elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
7172objects.")
7173 (license license:artistic2.0)))
7174
6dc60998
RW		 7175(define-public r-bsgenome-dmelanogaster-ucsc-dm3
7176 (package
7177 (name "r-bsgenome-dmelanogaster-ucsc-dm3")
7178 (version "1.4.0")
7179 (source (origin
7180 (method url-fetch)
7181 ;; We cannot use bioconductor-uri here because this tarball is
7182 ;; located under "data/annotation/" instead of "bioc/".
7183 (uri (string-append "http://www.bioconductor.org/packages/"
7184 "release/data/annotation/src/contrib/"
7185 "BSgenome.Dmelanogaster.UCSC.dm3_"
7186 version ".tar.gz"))
7187 (sha256
7188 (base32
7189 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
7190 (properties
7191 `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
7192 (build-system r-build-system)
7193 ;; As this package provides little more than a very large data file it
7194 ;; doesn't make sense to build substitutes.
7195 (arguments `(#:substitutable? #f))
7196 (propagated-inputs
7197 `(("r-bsgenome" ,r-bsgenome)))
7198 (home-page
7199 "http://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
7200 (synopsis "Full genome sequences for Fly")
7201 (description
7202 "This package provides full genome sequences for Drosophila
7203melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
7204Biostrings objects.")
7205 (license license:artistic2.0)))
7206
ae2462f7
RW		 7207(define-public r-motifrg
7208 (package
7209 (name "r-motifrg")
809251e1 7210 (version "1.18.0")
ae2462f7
RW		 7211 (source
7212 (origin
7213 (method url-fetch)
7214 (uri (bioconductor-uri "motifRG" version))
7215 (sha256
7216 (base32
809251e1 7217 "1pa97aj6c5f3gx4bgriw110764dj3m9h104ddi8rv2bpy41yd98d"))))
ae2462f7
RW		 7218 (properties `((upstream-name . "motifRG")))
7219 (build-system r-build-system)
7220 (propagated-inputs
7221 `(("r-biostrings" ,r-biostrings)
7222 ("r-bsgenome" ,r-bsgenome)
7223 ("r-bsgenome.hsapiens.ucsc.hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7224 ("r-iranges" ,r-iranges)
7225 ("r-seqlogo" ,r-seqlogo)
7226 ("r-xvector" ,r-xvector)))
7227 (home-page "http://bioconductor.org/packages/motifRG")
7228 (synopsis "Discover motifs in high throughput sequencing data")
7229 (description
7230 "This package provides tools for discriminative motif discovery in high
7231throughput genetic sequencing data sets using regression methods.")
7232 (license license:artistic2.0)))
7233
a5002ae7
AE		 7234(define-public r-qtl
7235 (package
7236 (name "r-qtl")
e1c2ad67 7237 (version "1.39-5")
a5002ae7
AE		 7238 (source
7239 (origin
7240 (method url-fetch)
7241 (uri (string-append "mirror://cran/src/contrib/qtl_"
7242 version ".tar.gz"))
7243 (sha256
7244 (base32
e1c2ad67 7245 "1grwgvyv7x0dgay1858bg7qf4wk47gpnq7qkqpcda9cn0h970d6f"))))
a5002ae7
AE		 7246 (build-system r-build-system)
7247 (home-page "http://rqtl.org/")
7248 (synopsis "R package for analyzing QTL experiments in genetics")
7249 (description "R/qtl is an extension library for the R statistics
7250system. It is used to analyze experimental crosses for identifying
7251genes contributing to variation in quantitative traits (so-called
7252quantitative trait loci, QTLs).
7253
7254Using a hidden Markov model, R/qtl allows to estimate genetic maps, to
7255identify genotyping errors, and to perform single-QTL and two-QTL,
7256two-dimensional genome scans.")
7257 (license license:gpl3)))
d1e32822 7258
9e3ba31c
RJ		 7259(define-public r-zlibbioc
7260 (package
7261 (name "r-zlibbioc")
da22da86 7262 (version "1.20.0")
9e3ba31c
RJ		 7263 (source (origin
7264 (method url-fetch)
7265 (uri (bioconductor-uri "zlibbioc" version))
7266 (sha256
7267 (base32
da22da86 7268 "0hbk90q5hl0fycfvy5nxxa4hxgglag9lzp7i0fg849bqygg5nbyq"))))
9e3ba31c
RJ		 7269 (properties
7270 `((upstream-name . "zlibbioc")))
7271 (build-system r-build-system)
7272 (home-page "https://bioconductor.org/packages/zlibbioc")
7273 (synopsis "Provider for zlib-1.2.5 to R packages")
7274 (description "This package uses the source code of zlib-1.2.5 to create
7275libraries for systems that do not have these available via other means.")
7276 (license license:artistic2.0)))
7277
e619a5c2
RW		 7278(define-public r-r4rna
7279 (package
7280 (name "r-r4rna")
7281 (version "0.1.4")
7282 (source
7283 (origin
7284 (method url-fetch)
7285 (uri (string-append "http://www.e-rna.org/r-chie/files/R4RNA_"
7286 version ".tar.gz"))
7287 (sha256
7288 (base32
7289 "1p0i78wh76jfgmn9jphbwwaz6yy6pipzfg08xs54cxavxg2j81p5"))))
7290 (build-system r-build-system)
7291 (propagated-inputs
7292 `(("r-optparse" ,r-optparse)
7293 ("r-rcolorbrewer" ,r-rcolorbrewer)))
7294 (home-page "http://www.e-rna.org/r-chie/index.cgi")
7295 (synopsis "Analysis framework for RNA secondary structure")
7296 (description
7297 "The R4RNA package aims to be a general framework for the analysis of RNA
7298secondary structure and comparative analysis in R.")
7299 (license license:gpl3+)))
7300
52765a63
RW		 7301(define-public r-rhtslib
7302 (package
7303 (name "r-rhtslib")
53ca52f0 7304 (version "1.6.0")
52765a63
RW		 7305 (source
7306 (origin
7307 (method url-fetch)
7308 (uri (bioconductor-uri "Rhtslib" version))
7309 (sha256
7310 (base32
53ca52f0 7311 "1vk3ng61dhi3pbia1lp3gl3mlr3i1vb2lkq83qb53i9dzz128wh9"))))
52765a63
RW		 7312 (properties `((upstream-name . "Rhtslib")))
7313 (build-system r-build-system)
7314 (propagated-inputs
7315 `(("r-zlibbioc" ,r-zlibbioc)))
7316 (inputs
7317 `(("zlib" ,zlib)))
53ca52f0
RW		 7318 (native-inputs
7319 `(("autoconf" ,autoconf)))
52765a63
RW		 7320 (home-page "https://github.com/nhayden/Rhtslib")
7321 (synopsis "High-throughput sequencing library as an R package")
7322 (description
7323 "This package provides the HTSlib C library for high-throughput
7324nucleotide sequence analysis. The package is primarily useful to developers
7325of other R packages who wish to make use of HTSlib.")
7326 (license license:lgpl2.0+)))
7327
fe02c4c9
RW		 7328(define-public r-bamsignals
7329 (package
7330 (name "r-bamsignals")
e357bec8 7331 (version "1.6.0")
fe02c4c9
RW		 7332 (source
7333 (origin
7334 (method url-fetch)
7335 (uri (bioconductor-uri "bamsignals" version))
7336 (sha256
7337 (base32
e357bec8 7338 "1k42gvk5mgq4la1fp0in3an2zfdz69h6522jsqhmk0f6i75kg4mb"))))
fe02c4c9
RW		 7339 (build-system r-build-system)
7340 (propagated-inputs
7341 `(("r-biocgenerics" ,r-biocgenerics)
7342 ("r-genomicranges" ,r-genomicranges)
7343 ("r-iranges" ,r-iranges)
7344 ("r-rcpp" ,r-rcpp)
7345 ("r-rhtslib" ,r-rhtslib)
7346 ("r-zlibbioc" ,r-zlibbioc)))
7347 (inputs
7348 `(("zlib" ,zlib)))
7349 (home-page "http://bioconductor.org/packages/bamsignals")
7350 (synopsis "Extract read count signals from bam files")
7351 (description
7352 "This package allows to efficiently obtain count vectors from indexed bam
7353files. It counts the number of nucleotide sequence reads in given genomic
7354ranges and it computes reads profiles and coverage profiles. It also handles
7355paired-end data.")
7356 (license license:gpl2+)))
7357
89984be4
RW		 7358(define-public r-rcas
7359 (package
7360 (name "r-rcas")
ed8fbe5e 7361 (version "1.0.0")
89984be4
RW		 7362 (source (origin
7363 (method url-fetch)
7364 (uri (string-append "https://github.com/BIMSBbioinfo/RCAS/archive/v"
7365 version ".tar.gz"))
7366 (file-name (string-append name "-" version ".tar.gz"))
7367 (sha256
7368 (base32
ed8fbe5e 7369 "1h7di822ihgkhmmmlfbfz3c2dkjyjxl307i6mx8w0cwjqbna1kp6"))))
89984be4
RW		 7370 (build-system r-build-system)
7371 (native-inputs
7372 `(("r-knitr" ,r-knitr)
7373 ("r-testthat" ,r-testthat)
7374 ;; During vignette building knitr checks that "pandoc-citeproc"
7375 ;; is in the PATH.
7376 ("ghc-pandoc-citeproc" ,ghc-pandoc-citeproc)))
7377 (propagated-inputs
7378 `(("r-data-table" ,r-data-table)
7379 ("r-biomart" ,r-biomart)
7380 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
7381 ("r-org-ce-eg-db" ,r-org-ce-eg-db)
7382 ("r-org-dm-eg-db" ,r-org-dm-eg-db)
7383 ("r-org-mm-eg-db" ,r-org-mm-eg-db)
7384 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7385 ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
7386 ("r-bsgenome-celegans-ucsc-ce10" ,r-bsgenome-celegans-ucsc-ce10)
7387 ("r-bsgenome-dmelanogaster-ucsc-dm3" ,r-bsgenome-dmelanogaster-ucsc-dm3)
7388 ("r-topgo" ,r-topgo)
7389 ("r-dt" ,r-dt)
7390 ("r-plotly" ,r-plotly)
7391 ("r-motifrg" ,r-motifrg)
7392 ("r-genomation" ,r-genomation)
7393 ("r-genomicfeatures" ,r-genomicfeatures)
7394 ("r-rtracklayer" ,r-rtracklayer)
7395 ("r-rmarkdown" ,r-rmarkdown)))
7396 (synopsis "RNA-centric annotation system")
7397 (description
7398 "RCAS aims to be a standalone RNA-centric annotation system that provides
7399intuitive reports and publication-ready graphics. This package provides the R
7400library implementing most of the pipeline's features.")
7401 (home-page "https://github.com/BIMSBbioinfo/RCAS")
7402 (license license:expat)))
7403
50937297
RW		 7404(define-public rcas-web
7405 (package
7406 (name "rcas-web")
7407 (version "0.0.3")
7408 (source
7409 (origin
7410 (method url-fetch)
7411 (uri (string-append "https://github.com/BIMSBbioinfo/rcas-web/"
7412 "releases/download/v" version
7413 "/rcas-web-" version ".tar.gz"))
7414 (sha256
7415 (base32
7416 "0d3my0g8i7js59n184zzzjdki7hgmhpi4rhfvk7i6jsw01ba04qq"))))
7417 (build-system gnu-build-system)
7418 (arguments
7419 `(#:phases
7420 (modify-phases %standard-phases
7421 (add-after 'install 'wrap-executable
7422 (lambda* (#:key inputs outputs #:allow-other-keys)
7423 (let* ((out (assoc-ref outputs "out"))
7424 (json (assoc-ref inputs "guile-json"))
7425 (redis (assoc-ref inputs "guile-redis"))
7426 (path (string-append
7427 json "/share/guile/site/2.2:"
7428 redis "/share/guile/site/2.2")))
7429 (wrap-program (string-append out "/bin/rcas-web")
7430 `("GUILE_LOAD_PATH" ":" = (,path))
7431 `("GUILE_LOAD_COMPILED_PATH" ":" = (,path))
7432 `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
7433 #t)))))
7434 (inputs
7435 `(("r" ,r)
7436 ("r-rcas" ,r-rcas)
7437 ("guile-next" ,guile-next)
7438 ("guile-json" ,guile2.2-json)
7439 ("guile-redis" ,guile2.2-redis)))
7440 (native-inputs
7441 `(("pkg-config" ,pkg-config)))
7442 (home-page "https://github.com/BIMSBbioinfo/rcas-web")
7443 (synopsis "Web interface for RNA-centric annotation system (RCAS)")
7444 (description "This package provides a simple web interface for the
7445@dfn{RNA-centric annotation system} (RCAS).")
7446 (license license:agpl3+)))
7447
7500e42b
RJ		 7448(define-public r-mutationalpatterns
7449 (package
7450 (name "r-mutationalpatterns")
7451 (version "1.0.0")
7452 (source
7453 (origin
7454 (method url-fetch)
7455 (uri (bioconductor-uri "MutationalPatterns" version))
7456 (sha256
7457 (base32
7458 "1a3c2bm0xx0q4gf98jiw74msmdf2fr8rbsdysd5ww9kqlzmsbr17"))))
7459 (build-system r-build-system)
7460 (propagated-inputs
7461 `(("r-biocgenerics" ,r-biocgenerics)
7462 ("r-biostrings" ,r-biostrings)
7463 ("r-genomicranges" ,r-genomicranges)
7464 ("r-genomeinfodb" ,r-genomeinfodb)
7465 ("r-ggplot2" ,r-ggplot2)
7466 ("r-gridextra" ,r-gridextra)
7467 ("r-iranges" ,r-iranges)
7468 ("r-nmf" ,r-nmf)
7469 ("r-plyr" ,r-plyr)
7470 ("r-pracma" ,r-pracma)
7471 ("r-reshape2" ,r-reshape2)
7472 ("r-summarizedexperiment" ,r-summarizedexperiment)
7473 ("r-variantannotation" ,r-variantannotation)))
7474 (home-page "http://bioconductor.org/packages/MutationalPatterns/")
7475 (synopsis "Extract and visualize mutational patterns in genomic data")
7476 (description "This package provides an extensive toolset for the
7477characterization and visualization of a wide range of mutational patterns
7478in SNV base substitution data.")
7479 (license license:expat)))
7480
d7160529
RW		 7481(define-public r-wgcna
7482 (package
7483 (name "r-wgcna")
7484 (version "1.51")
7485 (source
7486 (origin
7487 (method url-fetch)
7488 (uri (cran-uri "WGCNA" version))
7489 (sha256
7490 (base32
7491 "0hzvnhw76vwg8bl8x368f0c5szpwb8323bmrb3bir93i5bmfjsxx"))))
7492 (properties `((upstream-name . "WGCNA")))
7493 (build-system r-build-system)
7494 (propagated-inputs
7495 `(("r-annotationdbi" ,r-annotationdbi)
7496 ("r-doparallel" ,r-doparallel)
7497 ("r-dynamictreecut" ,r-dynamictreecut)
7498 ("r-fastcluster" ,r-fastcluster)
7499 ("r-foreach" ,r-foreach)
7500 ("r-go-db" ,r-go-db)
7501 ("r-hmisc" ,r-hmisc)
7502 ("r-impute" ,r-impute)
7503 ("r-matrixstats" ,r-matrixstats)
7504 ("r-preprocesscore" ,r-preprocesscore)))
7505 (home-page
7506 "http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/Rpackages/WGCNA/")
7507 (synopsis "Weighted correlation network analysis")
7508 (description
7509 "This package provides functions necessary to perform Weighted
7510Correlation Network Analysis on high-dimensional data. It includes functions
7511for rudimentary data cleaning, construction and summarization of correlation
7512networks, module identification and functions for relating both variables and
7513modules to sample traits. It also includes a number of utility functions for
7514data manipulation and visualization.")
7515 (license license:gpl2+)))
7516
c827f202
RW		 7517(define-public r-chipkernels
7518 (let ((commit "c9cfcacb626b1221094fb3490ea7bac0fd625372")
7519 (revision "1"))
7520 (package
7521 (name "r-chipkernels")
7522 (version (string-append "1.1-" revision "." (string-take commit 9)))
7523 (source
7524 (origin
7525 (method git-fetch)
7526 (uri (git-reference
7527 (url "https://github.com/ManuSetty/ChIPKernels.git")
7528 (commit commit)))
7529 (file-name (string-append name "-" version))
7530 (sha256
7531 (base32
7532 "14bj5qhjm1hsm9ay561nfbqi9wxsa7y487df2idsaaf6z10nw4v0"))))
7533 (build-system r-build-system)
7534 (propagated-inputs
7535 `(("r-iranges" ,r-iranges)
7536 ("r-xvector" ,r-xvector)
7537 ("r-biostrings" ,r-biostrings)
7538 ("r-bsgenome" ,r-bsgenome)
7539 ("r-gtools" ,r-gtools)
7540 ("r-genomicranges" ,r-genomicranges)
7541 ("r-sfsmisc" ,r-sfsmisc)
7542 ("r-kernlab" ,r-kernlab)
7543 ("r-s4vectors" ,r-s4vectors)
7544 ("r-biocgenerics" ,r-biocgenerics)))
7545 (home-page "https://github.com/ManuSetty/ChIPKernels")
7546 (synopsis "Build string kernels for DNA Sequence analysis")
7547 (description "ChIPKernels is an R package for building different string
7548kernels used for DNA Sequence analysis. A dictionary of the desired kernel
7549must be built and this dictionary can be used for determining kernels for DNA
7550Sequences.")
7551 (license license:gpl2+))))
7552
2d9fb170
RW		 7553(define-public r-seqgl
7554 (package
7555 (name "r-seqgl")
7556 (version "1.1.4")
7557 (source
7558 (origin
7559 (method url-fetch)
7560 (uri (string-append "https://github.com/ManuSetty/SeqGL/"
7561 "archive/" version ".tar.gz"))
7562 (file-name (string-append name "-" version ".tar.gz"))
7563 (sha256
7564 (base32
7565 "0pnk1p3sci5yipyc8xnb6jbmydpl80fld927xgnbcv104hy8h8yh"))))
7566 (build-system r-build-system)
7567 (propagated-inputs
7568 `(("r-biostrings" ,r-biostrings)
7569 ("r-chipkernels" ,r-chipkernels)
7570 ("r-genomicranges" ,r-genomicranges)
7571 ("r-spams" ,r-spams)
7572 ("r-wgcna" ,r-wgcna)
7573 ("r-fastcluster" ,r-fastcluster)))
7574 (home-page "https://github.com/ManuSetty/SeqGL")
7575 (synopsis "Group lasso for Dnase/ChIP-seq data")
7576 (description "SeqGL is a group lasso based algorithm to extract
7577transcription factor sequence signals from ChIP, DNase and ATAC-seq profiles.
7578This package presents a method which uses group lasso to discriminate between
7579bound and non bound genomic regions to accurately identify transcription
7580factors bound at the specific regions.")
7581 (license license:gpl2+)))
7582
bd3be46e
RW		 7583(define-public r-gkmsvm
7584 (package
7585 (name "r-gkmsvm")
7586 (version "0.71.0")
7587 (source
7588 (origin
7589 (method url-fetch)
7590 (uri (cran-uri "gkmSVM" version))
7591 (sha256
7592 (base32
7593 "1zpxgxmf2nd5j5wn00ps6kfxr8wxh7d1swr1rr4spq7sj5z5z0k0"))))
7594 (properties `((upstream-name . "gkmSVM")))
7595 (build-system r-build-system)
7596 (propagated-inputs
7597 `(("r-biocgenerics" ,r-biocgenerics)
7598 ("r-biostrings" ,r-biostrings)
7599 ("r-genomeinfodb" ,r-genomeinfodb)
7600 ("r-genomicranges" ,r-genomicranges)
7601 ("r-iranges" ,r-iranges)
7602 ("r-kernlab" ,r-kernlab)
7603 ("r-rcpp" ,r-rcpp)
7604 ("r-rocr" ,r-rocr)
7605 ("r-rtracklayer" ,r-rtracklayer)
7606 ("r-s4vectors" ,r-s4vectors)
7607 ("r-seqinr" ,r-seqinr)))
7608 (home-page "http://cran.r-project.org/web/packages/gkmSVM")
7609 (synopsis "Gapped-kmer support vector machine")
7610 (description
7611 "This R package provides tools for training gapped-kmer SVM classifiers
7612for DNA and protein sequences. This package supports several sequence
7613kernels, including: gkmSVM, kmer-SVM, mismatch kernel and wildcard kernel.")
7614 (license license:gpl2+)))
7615
66e40e00
RW		 7616(define-public emboss
7617 (package
7618 (name "emboss")
7619 (version "6.5.7")
7620 (source (origin
7621 (method url-fetch)
7622 (uri (string-append "ftp://emboss.open-bio.org/pub/EMBOSS/old/"
7623 (version-major+minor version) ".0/"
7624 "EMBOSS-" version ".tar.gz"))
7625 (sha256
7626 (base32
7627 "0vsmz96gc411yj2iyzdrsmg4l2n1nhgmp7vrgzlxx3xixv9xbf0q"))))
7628 (build-system gnu-build-system)
7629 (arguments
7630 `(#:configure-flags
7631 (list (string-append "--with-hpdf="
7632 (assoc-ref %build-inputs "libharu")))
7633 #:phases
7634 (modify-phases %standard-phases
7635 (add-after 'unpack 'fix-checks
7636 (lambda _
7637 ;; The PNGDRIVER tests check for the presence of libgd, libpng
7638 ;; and zlib, but assume that they are all found at the same
7639 ;; prefix.
7640 (substitute* "configure.in"
7641 (("CHECK_PNGDRIVER")
7642 "LIBS=\"$LIBS -lgd -lpng -lz -lm\"
7643AC_DEFINE([PLD_png], [1], [Define to 1 if PNG support is available])
7644AM_CONDITIONAL(AMPNG, true)"))
7645 #t))
7646 (add-after 'unpack 'disable-update-check
7647 (lambda _
7648 ;; At build time there is no connection to the Internet, so
7649 ;; looking for updates will not work.
7650 (substitute* "Makefile.am"
7651 (("\\$\\(bindir\\)/embossupdate") ""))
7652 #t))
7653 (add-before 'configure 'autogen
7654 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
7655 (inputs
7656 `(("perl" ,perl)
7657 ("libpng" ,libpng)
7658 ("gd" ,gd)
7659 ("libx11" ,libx11)
7660 ("libharu" ,libharu)
7661 ("zlib" ,zlib)))
7662 (native-inputs
7663 `(("autoconf" ,autoconf)
7664 ("automake" ,automake)
7665 ("libtool" ,libtool)
7666 ("pkg-config" ,pkg-config)))
7667 (home-page "http://emboss.sourceforge.net")
7668 (synopsis "Molecular biology analysis suite")
7669 (description "EMBOSS is the \"European Molecular Biology Open Software
7670Suite\". EMBOSS is an analysis package specially developed for the needs of
7671the molecular biology (e.g. EMBnet) user community. The software
7672automatically copes with data in a variety of formats and even allows
7673transparent retrieval of sequence data from the web. It also provides a
7674number of libraries for the development of software in the field of molecular
7675biology. EMBOSS also integrates a range of currently available packages and
7676tools for sequence analysis into a seamless whole.")
7677 (license license:gpl2+)))
7678
1f1b20b8
RW		 7679(define-public bits
7680 (let ((revision "1")
7681 (commit "3cc4567896d9d6442923da944beb704750a08d2d"))
7682 (package
7683 (name "bits")
7684 ;; The version is 2.13.0 even though no release archives have been
7685 ;; published as yet.
7686 (version (string-append "2.13.0-" revision "." (string-take commit 9)))
7687 (source (origin
7688 (method git-fetch)
7689 (uri (git-reference
7690 (url "https://github.com/arq5x/bits.git")
7691 (commit commit)))
7692 (file-name (string-append name "-" version "-checkout"))
7693 (sha256
7694 (base32
7695 "17n2kffk4kmhivd8c98g2vr6y1s23vbg4sxlxs689wni66797hbs"))))
7696 (build-system gnu-build-system)
7697 (arguments
7698 `(#:tests? #f ;no tests included
7699 #:phases
7700 (modify-phases %standard-phases
7701 (delete 'configure)
7702 (add-after 'unpack 'remove-cuda
7703 (lambda _
7704 (substitute* "Makefile"
7705 ((".*_cuda") "")
7706 (("(bits_test_intersections) \\\\" _ match) match))
7707 #t))
7708 (replace 'install
7709 (lambda* (#:key outputs #:allow-other-keys)
7710 (copy-recursively
7711 "bin" (string-append (assoc-ref outputs "out") "/bin"))
7712 #t)))))
7713 (inputs
7714 `(("gsl" ,gsl)
7715 ("zlib" ,zlib)))
7716 (home-page "https://github.com/arq5x/bits")
7717 (synopsis "Implementation of binary interval search algorithm")
7718 (description "This package provides an implementation of the
7719BITS (Binary Interval Search) algorithm, an approach to interval set
7720intersection. It is especially suited for the comparison of diverse genomic
7721datasets and the exploration of large datasets of genome
7722intervals (e.g. genes, sequence alignments).")
7723 (license license:gpl2))))
7724
e62ffce5 7725(define-public piranha
883302da
RW		 7726 ;; There is no release tarball for the latest version. The latest commit is
7727 ;; older than one year at the time of this writing.
7728 (let ((revision "1")
7729 (commit "0466d364b71117d01e4471b74c514436cc281233"))
7730 (package
7731 (name "piranha")
7732 (version (string-append "1.2.1-" revision "." (string-take commit 9)))
7733 (source (origin
7734 (method git-fetch)
7735 (uri (git-reference
7736 (url "https://github.com/smithlabcode/piranha.git")
7737 (commit commit)))
7738 (sha256
7739 (base32
7740 "117dc0zf20c61jam69sk4abl57ah6yi6i7qra7d7y5zrbgk12q5n"))))
7741 (build-system gnu-build-system)
7742 (arguments
7743 `(#:test-target "test"
7744 #:phases
7745 (modify-phases %standard-phases
7746 (add-after 'unpack 'copy-smithlab-cpp
7747 (lambda* (#:key inputs #:allow-other-keys)
e62ffce5 7748 (for-each (lambda (file)
883302da
RW		 7749 (install-file file "./src/smithlab_cpp/"))
7750 (find-files (assoc-ref inputs "smithlab-cpp")))
7751 #t))
7752 (add-after 'install 'install-to-store
7753 (lambda* (#:key outputs #:allow-other-keys)
7754 (let* ((out (assoc-ref outputs "out"))
7755 (bin (string-append out "/bin")))
883302da
RW		 7756 (for-each (lambda (file)
7757 (install-file file bin))
7758 (find-files "bin" ".*")))
7759 #t)))
7760 #:configure-flags
7761 (list (string-append "--with-bam_tools_headers="
7762 (assoc-ref %build-inputs "bamtools") "/include/bamtools")
7763 (string-append "--with-bam_tools_library="
7764 (assoc-ref %build-inputs "bamtools") "/lib/bamtools"))))
7765 (inputs
7766 `(("bamtools" ,bamtools)
7767 ("samtools" ,samtools-0.1)
7768 ("gsl" ,gsl)
7769 ("smithlab-cpp"
7770 ,(let ((commit "3723e2db438c51501d0423429ff396c3035ba46a"))
7771 (origin
7772 (method git-fetch)
7773 (uri (git-reference
7774 (url "https://github.com/smithlabcode/smithlab_cpp.git")
7775 (commit commit)))
7776 (file-name (string-append "smithlab_cpp-" commit "-checkout"))
7777 (sha256
7778 (base32
7779 "0l4gvbwslw5ngziskja41c00x1r06l3yidv7y0xw9djibhykzy0g")))))))
7780 (native-inputs
7781 `(("python" ,python-2)))
7782 (home-page "https://github.com/smithlabcode/piranha")
7783 (synopsis "Peak-caller for CLIP-seq and RIP-seq data")
7784 (description
7785 "Piranha is a peak-caller for genomic data produced by CLIP-seq and
e62ffce5
RW		 7786RIP-seq experiments. It takes input in BED or BAM format and identifies
7787regions of statistically significant read enrichment. Additional covariates
7788may optionally be provided to further inform the peak-calling process.")
883302da 7789 (license license:gpl3+))))
e62ffce5 7790
d1e32822
RW		 7791(define-public pepr
7792 (package
7793 (name "pepr")
7794 (version "1.0.9")
7795 (source (origin
7796 (method url-fetch)
7797 (uri (string-append "https://pypi.python.org/packages/source/P"
7798 "/PePr/PePr-" version ".tar.gz"))
7799 (sha256
7800 (base32
7801 "0qxjfdpl1b1y53nccws2d85f6k74zwmx8y8sd9rszcqhfayx6gdx"))))
7802 (build-system python-build-system)
7803 (arguments
7804 `(#:python ,python-2 ; python2 only
b41a05ce 7805 #:tests? #f)) ; no tests included
d1e32822
RW		 7806 (propagated-inputs
7807 `(("python2-numpy" ,python2-numpy)
7808 ("python2-scipy" ,python2-scipy)
7809 ("python2-pysam" ,python2-pysam)))
7810 (home-page "https://code.google.com/p/pepr-chip-seq/")
7811 (synopsis "Peak-calling and prioritization pipeline for ChIP-Seq data")
7812 (description
7813 "PePr is a ChIP-Seq peak calling or differential binding analysis tool
7814that is primarily designed for data with biological replicates. It uses a
7815negative binomial distribution to model the read counts among the samples in
7816the same group, and look for consistent differences between ChIP and control
7817group or two ChIP groups run under different conditions.")
7818 (license license:gpl3+)))
6b49a37e
RJ		 7819
7820(define-public filevercmp
7821 (let ((commit "1a9b779b93d0b244040274794d402106907b71b7"))
7822 (package
7823 (name "filevercmp")
7824 (version (string-append "0-1." (string-take commit 7)))
7825 (source (origin
7826 (method url-fetch)
7827 (uri (string-append "https://github.com/ekg/filevercmp/archive/"
7828 commit ".tar.gz"))
7829 (file-name (string-append name "-" version ".tar.gz"))
7830 (sha256
7831 (base32 "0yp5jswf5j2pqc6517x277s4s6h1ss99v57kxw9gy0jkfl3yh450"))))
7832 (build-system gnu-build-system)
7833 (arguments
7834 `(#:tests? #f ; There are no tests to run.
7835 #:phases
7836 (modify-phases %standard-phases
7837 (delete 'configure) ; There is no configure phase.
7838 (replace 'install
7839 (lambda* (#:key outputs #:allow-other-keys)
7840 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
7841 (install-file "filevercmp" bin)))))))
7842 (home-page "https://github.com/ekg/filevercmp")
7843 (synopsis "This program compares version strings")
7844 (description "This program compares version strings. It intends to be a
7845replacement for strverscmp.")
7846 (license license:gpl3+))))
5fb5dffb
RW		 7847
7848(define-public multiqc
7849 (package
7850 (name "multiqc")
7851 (version "0.6")
7852 (source
7853 (origin
7854 (method url-fetch)
7855 (uri (pypi-uri "multiqc" version))
7856 (sha256
7857 (base32
7858 "0avw11h63ldpxy5pizc3wl1wa01ha7q10wb240nggsjz3jaqvyiy"))))
7859 (build-system python-build-system)
7860 (propagated-inputs
7861 `(("python-jinja2" ,python-jinja2)
7862 ("python-simplejson" ,python-simplejson)
7863 ("python-pyyaml" ,python-pyyaml)
7864 ("python-click" ,python-click)
7865 ("python-matplotlib" ,python-matplotlib)
7866 ("python-numpy" ,python-numpy)))
5fb5dffb
RW		 7867 (home-page "http://multiqc.info")
7868 (synopsis "Aggregate bioinformatics analysis reports")
7869 (description
7870 "MultiQC is a tool to aggregate bioinformatics results across many
7871samples into a single report. It contains modules for a large number of
7872common bioinformatics tools.")
7873 (license license:gpl3)))
jackhill's copy of GNU Guix: https://guix.gnu.org/