HCoop / jackhill / guix / guix.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
gnu: moka-icon-theme: Add faba-icon-theme as propagated input.
[jackhill/guix/guix.git] / gnu / packages / bioinformatics.scm
CommitLineData
4e10a221 1;;; GNU Guix --- Functional package management for GNU
6193563a 2;;; Copyright © 2014, 2015, 2016, 2017 Ricardo Wurmus <rekado@elephly.net>
a9f754d7 3;;; Copyright © 2015, 2016, 2017 Ben Woodcroft <donttrustben@gmail.com>
8921841d 4;;; Copyright © 2015, 2016 Pjotr Prins <pjotr.guix@thebird.nl>
a5002ae7 5;;; Copyright © 2015 Andreas Enge <andreas@enge.fr>
a0a71439 6;;; Copyright © 2016 Roel Janssen <roel@gnu.org>
ce7e361f 7;;; Copyright © 2016, 2017 Efraim Flashner <efraim@flashner.co.il>
318c0aee 8;;; Copyright © 2016 Marius Bakke <mbakke@fastmail.com>
3fffabce 9;;; Copyright © 2016 Raoul Bonnal <ilpuccio.febo@gmail.com>
ce4dfde0 10;;; Copyright © 2017 Tobias Geerinckx-Rice <me@tobias.gr>
4e10a221
RW		 11;;;
12;;; This file is part of GNU Guix.
13;;;
14;;; GNU Guix is free software; you can redistribute it and/or modify it
15;;; under the terms of the GNU General Public License as published by
16;;; the Free Software Foundation; either version 3 of the License, or (at
17;;; your option) any later version.
18;;;
19;;; GNU Guix is distributed in the hope that it will be useful, but
20;;; WITHOUT ANY WARRANTY; without even the implied warranty of
21;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22;;; GNU General Public License for more details.
23;;;
24;;; You should have received a copy of the GNU General Public License
25;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
26
27(define-module (gnu packages bioinformatics)
28 #:use-module ((guix licenses) #:prefix license:)
29 #:use-module (guix packages)
8e913213 30 #:use-module (guix utils)
4e10a221 31 #:use-module (guix download)
2c16316e 32 #:use-module (guix git-download)
ec946638 33 #:use-module (guix hg-download)
10b4a969 34 #:use-module (guix build-system ant)
4e10a221 35 #:use-module (guix build-system gnu)
d7678942 36 #:use-module (guix build-system cmake)
c033f5d6 37 #:use-module (guix build-system ocaml)
365c8153 38 #:use-module (guix build-system perl)
8622a072 39 #:use-module (guix build-system python)
a5002ae7 40 #:use-module (guix build-system r)
9c38b540 41 #:use-module (guix build-system ruby)
d3517eda 42 #:use-module (guix build-system trivial)
4e10a221 43 #:use-module (gnu packages)
a2950fa4 44 #:use-module (gnu packages autotools)
684bf7c7 45 #:use-module (gnu packages algebra)
d3517eda 46 #:use-module (gnu packages base)
318c0aee 47 #:use-module (gnu packages bash)
a0a71439 48 #:use-module (gnu packages bison)
e4e5a4d8 49 #:use-module (gnu packages boost)
4e10a221 50 #:use-module (gnu packages compression)
82c370de 51 #:use-module (gnu packages cpio)
7cb61550 52 #:use-module (gnu packages cran)
1baee943 53 #:use-module (gnu packages curl)
99828fa7 54 #:use-module (gnu packages documentation)
94820951 55 #:use-module (gnu packages databases)
d29150b5 56 #:use-module (gnu packages datastructures)
75dd2424 57 #:use-module (gnu packages file)
99268755 58 #:use-module (gnu packages flex)
02f35bb5 59 #:use-module (gnu packages gawk)
2409f37f 60 #:use-module (gnu packages gcc)
66e40e00 61 #:use-module (gnu packages gd)
97b9da68 62 #:use-module (gnu packages gtk)
b16728b0 63 #:use-module (gnu packages glib)
18f5d2a7 64 #:use-module (gnu packages graph)
db7a3444 65 #:use-module (gnu packages groff)
50937297 66 #:use-module (gnu packages guile)
89984be4 67 #:use-module (gnu packages haskell)
66e40e00 68 #:use-module (gnu packages image)
97b9da68 69 #:use-module (gnu packages imagemagick)
15a3c3d4 70 #:use-module (gnu packages java)
5ded35d8 71 #:use-module (gnu packages ldc)
51c64999 72 #:use-module (gnu packages linux)
ec946638 73 #:use-module (gnu packages logging)
36742f43 74 #:use-module (gnu packages machine-learning)
db7a3444 75 #:use-module (gnu packages man)
c833ab55 76 #:use-module (gnu packages maths)
6c2b26e2 77 #:use-module (gnu packages mpi)
4e10a221 78 #:use-module (gnu packages ncurses)
c033f5d6 79 #:use-module (gnu packages ocaml)
81f3e0c1 80 #:use-module (gnu packages pcre)
ceb62d54 81 #:use-module (gnu packages parallel)
66e40e00 82 #:use-module (gnu packages pdf)
4e10a221
RW		 83 #:use-module (gnu packages perl)
84 #:use-module (gnu packages pkg-config)
bfe3c685 85 #:use-module (gnu packages popt)
e4e5a4d8 86 #:use-module (gnu packages protobuf)
346a829a 87 #:use-module (gnu packages python)
ec946638 88 #:use-module (gnu packages readline)
9c38b540 89 #:use-module (gnu packages ruby)
84be3b99 90 #:use-module (gnu packages serialization)
94820951 91 #:use-module (gnu packages shells)
c833ab55 92 #:use-module (gnu packages statistics)
aa163424 93 #:use-module (gnu packages swig)
d7678942 94 #:use-module (gnu packages tbb)
97b9da68 95 #:use-module (gnu packages tex)
db7a3444 96 #:use-module (gnu packages texinfo)
2127cedb 97 #:use-module (gnu packages textutils)
43c565d2 98 #:use-module (gnu packages time)
a2950fa4 99 #:use-module (gnu packages tls)
ce7155d5 100 #:use-module (gnu packages vim)
365c8153 101 #:use-module (gnu packages web)
c833ab55 102 #:use-module (gnu packages xml)
66e40e00 103 #:use-module (gnu packages xorg)
2c9232ae 104 #:use-module (srfi srfi-1)
ce7e361f 105 #:use-module (ice-9 match))
4e10a221 106
036cd0cb
RJPB		 107(define-public r-ape
108 (package
109 (name "r-ape")
b12c6a66 110 (version "5.0")
036cd0cb
RJPB		 111 (source
112 (origin
113 (method url-fetch)
114 (uri (cran-uri "ape" version))
115 (sha256
116 (base32
b12c6a66 117 "0q59pmxawz498cb9mv5m49lhiwxib8ak94yyydz7qg8b6lpd4bn3"))))
036cd0cb 118 (build-system r-build-system)
aeb64f3c
RW		 119 (propagated-inputs
120 `(("r-lattice" ,r-lattice)
b12c6a66
RW		 121 ("r-nlme" ,r-nlme)
122 ("r-rcpp" ,r-rcpp)))
036cd0cb
RJPB		 123 (home-page "http://ape-package.ird.fr/")
124 (synopsis "Analyses of phylogenetics and evolution")
125 (description
126 "This package provides functions for reading, writing, plotting, and
127manipulating phylogenetic trees, analyses of comparative data in a
128phylogenetic framework, ancestral character analyses, analyses of
129diversification and macroevolution, computing distances from DNA sequences,
130and several other tools.")
131 (license license:gpl2+)))
132
8dc797fa
BW		 133(define-public aragorn
134 (package
135 (name "aragorn")
e990c81d 136 (version "1.2.38")
8dc797fa
BW		 137 (source (origin
138 (method url-fetch)
139 (uri (string-append
140 "http://mbio-serv2.mbioekol.lu.se/ARAGORN/Downloads/aragorn"
141 version ".tgz"))
142 (sha256
143 (base32
e990c81d 144 "09i1rg716smlbnixfm7q1ml2mfpaa2fpn3hwjg625ysmfwwy712b"))))
8dc797fa
BW		 145 (build-system gnu-build-system)
146 (arguments
147 `(#:tests? #f ; there are no tests
148 #:phases
149 (modify-phases %standard-phases
150 (delete 'configure)
151 (replace 'build
152 (lambda _
153 (zero? (system* "gcc"
154 "-O3"
155 "-ffast-math"
156 "-finline-functions"
157 "-o"
158 "aragorn"
159 (string-append "aragorn" ,version ".c")))))
160 (replace 'install
161 (lambda* (#:key outputs #:allow-other-keys)
162 (let* ((out (assoc-ref outputs "out"))
163 (bin (string-append out "/bin"))
164 (man (string-append out "/share/man/man1")))
165 (mkdir-p bin)
f3860753 166 (install-file "aragorn" bin)
8dc797fa 167 (mkdir-p man)
f3860753 168 (install-file "aragorn.1" man))
8dc797fa
BW		 169 #t)))))
170 (home-page "http://mbio-serv2.mbioekol.lu.se/ARAGORN")
171 (synopsis "Detect tRNA, mtRNA and tmRNA genes in nucleotide sequences")
172 (description
173 "Aragorn identifies transfer RNA, mitochondrial RNA and
174transfer-messenger RNA from nucleotide sequences, based on homology to known
175tRNA consensus sequences and RNA structure. It also outputs the secondary
176structure of the predicted RNA.")
177 (license license:gpl2)))
178
a12ba6e8
BW		 179(define-public bamm
180 (package
181 (name "bamm")
4b6da268 182 (version "1.7.3")
a12ba6e8
BW		 183 (source (origin
184 (method url-fetch)
185 ;; BamM is not available on pypi.
186 (uri (string-append
4b6da268 187 "https://github.com/Ecogenomics/BamM/archive/"
a12ba6e8
BW		 188 version ".tar.gz"))
189 (file-name (string-append name "-" version ".tar.gz"))
190 (sha256
191 (base32
4b6da268 192 "1f35yxp4pc8aadsvbpg6r4kg2jh4fkjci0iby4iyljm6980sac0s"))
a12ba6e8
BW		 193 (modules '((guix build utils)))
194 (snippet
195 `(begin
196 ;; Delete bundled htslib.
197 (delete-file-recursively "c/htslib-1.3.1")
198 #t))))
199 (build-system python-build-system)
200 (arguments
201 `(#:python ,python-2 ; BamM is Python 2 only.
202 ;; Do not use bundled libhts. Do use the bundled libcfu because it has
203 ;; been modified from its original form.
204 #:configure-flags
205 (let ((htslib (assoc-ref %build-inputs "htslib")))
206 (list "--with-libhts-lib" (string-append htslib "/lib")
207 "--with-libhts-inc" (string-append htslib "/include/htslib")))
208 #:phases
209 (modify-phases %standard-phases
210 (add-after 'unpack 'autogen
211 (lambda _
212 (with-directory-excursion "c"
213 (let ((sh (which "sh")))
214 ;; Use autogen so that 'configure' works.
215 (substitute* "autogen.sh" (("/bin/sh") sh))
216 (setenv "CONFIG_SHELL" sh)
217 (substitute* "configure" (("/bin/sh") sh))
218 (zero? (system* "./autogen.sh"))))))
219 (delete 'build)
220 ;; Run tests after installation so compilation only happens once.
221 (delete 'check)
222 (add-after 'install 'wrap-executable
223 (lambda* (#:key outputs #:allow-other-keys)
224 (let* ((out (assoc-ref outputs "out"))
225 (path (getenv "PATH")))
226 (wrap-program (string-append out "/bin/bamm")
227 `("PATH" ":" prefix (,path))))
228 #t))
229 (add-after 'wrap-executable 'post-install-check
230 (lambda* (#:key inputs outputs #:allow-other-keys)
231 (setenv "PATH"
232 (string-append (assoc-ref outputs "out")
233 "/bin:"
234 (getenv "PATH")))
235 (setenv "PYTHONPATH"
236 (string-append
237 (assoc-ref outputs "out")
238 "/lib/python"
239 (string-take (string-take-right
240 (assoc-ref inputs "python") 5) 3)
241 "/site-packages:"
242 (getenv "PYTHONPATH")))
243 ;; There are 2 errors printed, but they are safe to ignore:
244 ;; 1) [E::hts_open_format] fail to open file ...
245 ;; 2) samtools view: failed to open ...
246 (zero? (system* "nosetests")))))))
247 (native-inputs
248 `(("autoconf" ,autoconf)
249 ("automake" ,automake)
250 ("libtool" ,libtool)
251 ("zlib" ,zlib)
252 ("python-nose" ,python2-nose)
f3b98f4f 253 ("python-pysam" ,python2-pysam)))
a12ba6e8 254 (inputs
bca2c576 255 `(("htslib" ,htslib-1.3) ; At least one test fails on htslib-1.4+.
a12ba6e8
BW		 256 ("samtools" ,samtools)
257 ("bwa" ,bwa)
258 ("grep" ,grep)
259 ("sed" ,sed)
260 ("coreutils" ,coreutils)))
261 (propagated-inputs
262 `(("python-numpy" ,python2-numpy)))
263 (home-page "http://ecogenomics.github.io/BamM/")
264 (synopsis "Metagenomics-focused BAM file manipulator")
265 (description
266 "BamM is a C library, wrapped in python, to efficiently generate and
267parse BAM files, specifically for the analysis of metagenomic data. For
268instance, it implements several methods to assess contig-wise read coverage.")
269 (license license:lgpl3+)))
270
9794180d
RW		 271(define-public bamtools
272 (package
273 (name "bamtools")
48d66a9c 274 (version "2.4.1")
9794180d
RW		 275 (source (origin
276 (method url-fetch)
277 (uri (string-append
278 "https://github.com/pezmaster31/bamtools/archive/v"
279 version ".tar.gz"))
280 (file-name (string-append name "-" version ".tar.gz"))
281 (sha256
282 (base32
48d66a9c 283 "0jr024kcrhjb82cm69i7p5fcg5375zlc1h3qh2n1v368hcd0qflk"))))
9794180d 284 (build-system cmake-build-system)
4702cec2
RW		 285 (arguments
286 `(#:tests? #f ;no "check" target
287 #:phases
288 (modify-phases %standard-phases
289 (add-before
290 'configure 'set-ldflags
291 (lambda* (#:key outputs #:allow-other-keys)
292 (setenv "LDFLAGS"
293 (string-append
294 "-Wl,-rpath="
295 (assoc-ref outputs "out") "/lib/bamtools")))))))
9794180d
RW		 296 (inputs `(("zlib" ,zlib)))
297 (home-page "https://github.com/pezmaster31/bamtools")
298 (synopsis "C++ API and command-line toolkit for working with BAM data")
299 (description
300 "BamTools provides both a C++ API and a command-line toolkit for handling
301BAM files.")
302 (license license:expat)))
303
bdc7be59
MB		 304(define-public bcftools
305 (package
306 (name "bcftools")
0620387a 307 (version "1.5")
bdc7be59
MB		 308 (source (origin
309 (method url-fetch)
310 (uri (string-append
311 "https://github.com/samtools/bcftools/releases/download/"
312 version "/bcftools-" version ".tar.bz2"))
313 (sha256
314 (base32
0620387a 315 "0093hkkvxmbwfaa7905s6185jymynvg42kq6sxv7fili11l5mxwz"))
3837108e 316 (patches (search-patches "bcftools-regidx-unsigned-char.patch"))
bdc7be59
MB		 317 (modules '((guix build utils)))
318 (snippet
319 ;; Delete bundled htslib.
0620387a 320 '(delete-file-recursively "htslib-1.5"))))
bdc7be59
MB		 321 (build-system gnu-build-system)
322 (arguments
323 `(#:test-target "test"
0620387a 324 #:configure-flags (list "--with-htslib=system")
bdc7be59
MB		 325 #:make-flags
326 (list
327 "USE_GPL=1"
0620387a 328 "LIBS=-lgsl -lgslcblas"
bdc7be59
MB		 329 (string-append "prefix=" (assoc-ref %outputs "out"))
330 (string-append "HTSDIR=" (assoc-ref %build-inputs "htslib") "/include")
98593f9f 331 (string-append "HTSLIB=" (assoc-ref %build-inputs "htslib") "/lib/libhts.so")
bdc7be59 332 (string-append "BGZIP=" (assoc-ref %build-inputs "htslib") "/bin/bgzip")
98593f9f
BW		 333 (string-append "TABIX=" (assoc-ref %build-inputs "htslib") "/bin/tabix")
334 (string-append "PACKAGE_VERSION=" ,version))
bdc7be59
MB		 335 #:phases
336 (modify-phases %standard-phases
bdc7be59
MB		 337 (add-before 'check 'patch-tests
338 (lambda _
339 (substitute* "test/test.pl"
340 (("/bin/bash") (which "bash")))
341 #t)))))
342 (native-inputs
343 `(("htslib" ,htslib)
344 ("perl" ,perl)))
345 (inputs
346 `(("gsl" ,gsl)
347 ("zlib" ,zlib)))
348 (home-page "https://samtools.github.io/bcftools/")
349 (synopsis "Utilities for variant calling and manipulating VCFs and BCFs")
350 (description
351 "BCFtools is a set of utilities that manipulate variant calls in the
352Variant Call Format (VCF) and its binary counterpart BCF. All commands work
353transparently with both VCFs and BCFs, both uncompressed and BGZF-compressed.")
354 ;; The sources are dual MIT/GPL, but becomes GPL-only when USE_GPL=1.
355 (license (list license:gpl3+ license:expat))))
356
8dd4ff11
RW		 357(define-public bedops
358 (package
359 (name "bedops")
1bbc3b1d 360 (version "2.4.14")
8dd4ff11
RW		 361 (source (origin
362 (method url-fetch)
363 (uri (string-append "https://github.com/bedops/bedops/archive/v"
364 version ".tar.gz"))
f586c877 365 (file-name (string-append name "-" version ".tar.gz"))
8dd4ff11
RW		 366 (sha256
367 (base32
1bbc3b1d 368 "1kqbac547wyqma81cyky9n7mkgikjpsfd3nnmcm6hpqwanqgh10v"))))
8dd4ff11
RW		 369 (build-system gnu-build-system)
370 (arguments
371 '(#:tests? #f
372 #:make-flags (list (string-append "BINDIR=" %output "/bin"))
373 #:phases
dc1d3cde
KK		 374 (modify-phases %standard-phases
375 (add-after 'unpack 'unpack-tarballs
376 (lambda _
377 ;; FIXME: Bedops includes tarballs of minimally patched upstream
378 ;; libraries jansson, zlib, and bzip2. We cannot just use stock
379 ;; libraries because at least one of the libraries (zlib) is
380 ;; patched to add a C++ function definition (deflateInit2cpp).
381 ;; Until the Bedops developers offer a way to link against system
382 ;; libraries we have to build the in-tree copies of these three
383 ;; libraries.
384
385 ;; See upstream discussion:
386 ;; https://github.com/bedops/bedops/issues/124
387
388 ;; Unpack the tarballs to benefit from shebang patching.
389 (with-directory-excursion "third-party"
390 (and (zero? (system* "tar" "xvf" "jansson-2.6.tar.bz2"))
391 (zero? (system* "tar" "xvf" "zlib-1.2.7.tar.bz2"))
392 (zero? (system* "tar" "xvf" "bzip2-1.0.6.tar.bz2"))))
393 ;; Disable unpacking of tarballs in Makefile.
394 (substitute* "system.mk/Makefile.linux"
395 (("^\tbzcat .*") "\t@echo \"not unpacking\"\n")
396 (("\\./configure") "CONFIG_SHELL=bash ./configure"))
397 (substitute* "third-party/zlib-1.2.7/Makefile.in"
398 (("^SHELL=.*$") "SHELL=bash\n"))
399 #t))
400 (delete 'configure))))
8dd4ff11
RW		 401 (home-page "https://github.com/bedops/bedops")
402 (synopsis "Tools for high-performance genomic feature operations")
403 (description
404 "BEDOPS is a suite of tools to address common questions raised in genomic
405studies---mostly with regard to overlap and proximity relationships between
406data sets. It aims to be scalable and flexible, facilitating the efficient
407and accurate analysis and management of large-scale genomic data.
408
409BEDOPS provides tools that perform highly efficient and scalable Boolean and
410other set operations, statistical calculations, archiving, conversion and
411other management of genomic data of arbitrary scale. Tasks can be easily
412split by chromosome for distributing whole-genome analyses across a
413computational cluster.")
414 (license license:gpl2+)))
415
81de5647
RW		 416(define-public bedtools
417 (package
418 (name "bedtools")
d285657e 419 (version "2.26.0")
81de5647
RW		 420 (source (origin
421 (method url-fetch)
422 (uri (string-append "https://github.com/arq5x/bedtools2/archive/v"
423 version ".tar.gz"))
f586c877 424 (file-name (string-append name "-" version ".tar.gz"))
81de5647
RW		 425 (sha256
426 (base32
d285657e 427 "0xvri5hnp2iim1cx6mcd5d9f102p5ql41x69rd6106x1c17pinqm"))))
81de5647
RW		 428 (build-system gnu-build-system)
429 (native-inputs `(("python" ,python-2)))
430 (inputs `(("samtools" ,samtools)
431 ("zlib" ,zlib)))
432 (arguments
433 '(#:test-target "test"
434 #:phases
6573ac82 435 (modify-phases %standard-phases
6573ac82
BW		 436 (delete 'configure)
437 (replace 'install
438 (lambda* (#:key outputs #:allow-other-keys)
439 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
440 (for-each (lambda (file)
441 (install-file file bin))
442 (find-files "bin" ".*")))
443 #t)))))
81de5647
RW		 444 (home-page "https://github.com/arq5x/bedtools2")
445 (synopsis "Tools for genome analysis and arithmetic")
446 (description
447 "Collectively, the bedtools utilities are a swiss-army knife of tools for
448a wide-range of genomics analysis tasks. The most widely-used tools enable
449genome arithmetic: that is, set theory on the genome. For example, bedtools
450allows one to intersect, merge, count, complement, and shuffle genomic
451intervals from multiple files in widely-used genomic file formats such as BAM,
452BED, GFF/GTF, VCF.")
453 (license license:gpl2)))
454
9a8f309c
RW		 455;; Later releases of bedtools produce files with more columns than
456;; what Ribotaper expects.
457(define-public bedtools-2.18
458 (package (inherit bedtools)
459 (name "bedtools")
460 (version "2.18.0")
461 (source (origin
462 (method url-fetch)
463 (uri (string-append "https://github.com/arq5x/bedtools2/"
464 "archive/v" version ".tar.gz"))
465 (file-name (string-append name "-" version ".tar.gz"))
466 (sha256
467 (base32
468 "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))))
469
17dc32a4
RW		 470(define-public ribotaper
471 (package
472 (name "ribotaper")
473 (version "1.3.1")
474 (source (origin
475 (method url-fetch)
476 (uri (string-append "https://ohlerlab.mdc-berlin.de/"
477 "files/RiboTaper/RiboTaper_Version_"
478 version ".tar.gz"))
479 (sha256
480 (base32
481 "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
482 (build-system gnu-build-system)
483 (inputs
484 `(("bedtools" ,bedtools-2.18)
485 ("samtools" ,samtools-0.1)
2d7c4ae3 486 ("r-minimal" ,r-minimal)
17dc32a4
RW		 487 ("r-foreach" ,r-foreach)
488 ("r-xnomial" ,r-xnomial)
489 ("r-domc" ,r-domc)
490 ("r-multitaper" ,r-multitaper)
491 ("r-seqinr" ,r-seqinr)))
492 (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
493 (synopsis "Define translated ORFs using ribosome profiling data")
494 (description
495 "Ribotaper is a method for defining translated @dfn{open reading
496frames} (ORFs) using ribosome profiling (ribo-seq) data. This package
497provides the Ribotaper pipeline.")
498 (license license:gpl3+)))
499
769fc6bb
RW		 500(define-public ribodiff
501 (package
502 (name "ribodiff")
503 (version "0.2.2")
504 (source
505 (origin
506 (method url-fetch)
507 (uri (string-append "https://github.com/ratschlab/RiboDiff/"
508 "archive/v" version ".tar.gz"))
509 (file-name (string-append name "-" version ".tar.gz"))
510 (sha256
511 (base32
512 "0wpbwmfv05wdjxv7ikm664f7s7p7cqr8jnw99zrda0q67rl50aaj"))))
513 (build-system python-build-system)
514 (arguments
515 `(#:python ,python-2
516 #:phases
517 (modify-phases %standard-phases
518 ;; Generate an installable executable script wrapper.
519 (add-after 'unpack 'patch-setup.py
520 (lambda _
521 (substitute* "setup.py"
522 (("^(.*)packages=.*" line prefix)
523 (string-append line "\n"
524 prefix "scripts=['scripts/TE.py'],\n")))
525 #t)))))
526 (inputs
527 `(("python-numpy" ,python2-numpy)
528 ("python-matplotlib" ,python2-matplotlib)
529 ("python-scipy" ,python2-scipy)
530 ("python-statsmodels" ,python2-statsmodels)))
2bb12f5a
MB		 531 (native-inputs
532 `(("python-mock" ,python2-mock)
533 ("python-nose" ,python2-nose)))
769fc6bb
RW		 534 (home-page "http://public.bmi.inf.ethz.ch/user/zhongy/RiboDiff/")
535 (synopsis "Detect translation efficiency changes from ribosome footprints")
536 (description "RiboDiff is a statistical tool that detects the protein
537translational efficiency change from Ribo-Seq (ribosome footprinting) and
538RNA-Seq data. It uses a generalized linear model to detect genes showing
539difference in translational profile taking mRNA abundance into account. It
540facilitates us to decipher the translational regulation that behave
541independently with transcriptional regulation.")
542 (license license:gpl3+)))
543
a0a71439
RJ		 544(define-public bioawk
545 (package
546 (name "bioawk")
547 (version "1.0")
548 (source (origin
549 (method url-fetch)
550 (uri (string-append "https://github.com/lh3/bioawk/archive/v"
551 version ".tar.gz"))
552 (file-name (string-append name "-" version ".tar.gz"))
553 (sha256
554 (base32 "1daizxsk17ahi9n58fj8vpgwyhzrzh54bzqhanjanp88kgrz7gjw"))))
555 (build-system gnu-build-system)
556 (inputs
557 `(("zlib" ,zlib)))
558 (native-inputs
559 `(("bison" ,bison)))
560 (arguments
561 `(#:tests? #f ; There are no tests to run.
562 ;; Bison must generate files, before other targets can build.
563 #:parallel-build? #f
564 #:phases
565 (modify-phases %standard-phases
566 (delete 'configure) ; There is no configure phase.
567 (replace 'install
568 (lambda* (#:key outputs #:allow-other-keys)
569 (let* ((out (assoc-ref outputs "out"))
570 (bin (string-append out "/bin"))
571 (man (string-append out "/share/man/man1")))
572 (mkdir-p man)
573 (copy-file "awk.1" (string-append man "/bioawk.1"))
574 (install-file "bioawk" bin)))))))
575 (home-page "https://github.com/lh3/bioawk")
576 (synopsis "AWK with bioinformatics extensions")
577 (description "Bioawk is an extension to Brian Kernighan's awk, adding the
578support of several common biological data formats, including optionally gzip'ed
579BED, GFF, SAM, VCF, FASTA/Q and TAB-delimited formats with column names. It
580also adds a few built-in functions and a command line option to use TAB as the
581input/output delimiter. When the new functionality is not used, bioawk is
582intended to behave exactly the same as the original BWK awk.")
583 (license license:x11)))
584
a2fb1492
RW		 585(define-public python2-pybedtools
586 (package
587 (name "python2-pybedtools")
588 (version "0.6.9")
589 (source (origin
590 (method url-fetch)
591 (uri (string-append
592 "https://pypi.python.org/packages/source/p/pybedtools/pybedtools-"
593 version ".tar.gz"))
594 (sha256
595 (base32
596 "1ldzdxw1p4y3g2ignmggsdypvqkcwqwzhdha4rbgpih048z5p4an"))))
597 (build-system python-build-system)
598 (arguments `(#:python ,python-2)) ; no Python 3 support
599 (inputs
f2516de2 600 `(("python-matplotlib" ,python2-matplotlib)))
a2fb1492
RW		 601 (propagated-inputs
602 `(("bedtools" ,bedtools)
603 ("samtools" ,samtools)))
604 (native-inputs
f2516de2
HG		 605 `(("python-cython" ,python2-cython)
606 ("python-pyyaml" ,python2-pyyaml)
f3b98f4f 607 ("python-nose" ,python2-nose)))
a2fb1492
RW		 608 (home-page "https://pythonhosted.org/pybedtools/")
609 (synopsis "Python wrapper for BEDtools programs")
610 (description
611 "pybedtools is a Python wrapper for Aaron Quinlan's BEDtools programs,
612which are widely used for genomic interval manipulation or \"genome algebra\".
613pybedtools extends BEDTools by offering feature-level manipulations from with
614Python.")
615 (license license:gpl2+)))
616
9e12eba8
BW		 617(define-public python-biom-format
618 (package
619 (name "python-biom-format")
abc08cba 620 (version "2.1.6")
9e12eba8
BW		 621 (source
622 (origin
623 (method url-fetch)
624 ;; Use GitHub as source because PyPI distribution does not contain
625 ;; test data: https://github.com/biocore/biom-format/issues/693
626 (uri (string-append "https://github.com/biocore/biom-format/archive/"
627 version ".tar.gz"))
628 (file-name (string-append name "-" version ".tar.gz"))
629 (sha256
630 (base32
abc08cba 631 "08cr7wpahk6zb31h4bs7jmzpvxcqv9s13xz40h6y2h656jvdvnpj"))))
9e12eba8 632 (build-system python-build-system)
de96ea28 633 (propagated-inputs
9e12eba8
BW		 634 `(("python-numpy" ,python-numpy)
635 ("python-scipy" ,python-scipy)
636 ("python-future" ,python-future)
637 ("python-click" ,python-click)
abc08cba
BW		 638 ("python-h5py" ,python-h5py)
639 ("python-pandas" ,python-pandas)))
da5ebd10
MB		 640 (native-inputs
641 `(("python-nose" ,python-nose)))
9e12eba8
BW		 642 (home-page "http://www.biom-format.org")
643 (synopsis "Biological Observation Matrix (BIOM) format utilities")
644 (description
645 "The BIOM file format is designed to be a general-use format for
646representing counts of observations e.g. operational taxonomic units, KEGG
647orthology groups or lipid types, in one or more biological samples
648e.g. microbiome samples, genomes, metagenomes.")
649 (license license:bsd-3)
650 (properties `((python2-variant . ,(delay python2-biom-format))))))
651
652(define-public python2-biom-format
653 (let ((base (package-with-python2 (strip-python2-variant python-biom-format))))
654 (package
655 (inherit base)
656 (arguments
657 `(#:phases
658 (modify-phases %standard-phases
659 ;; Do not require the unmaintained pyqi library.
660 (add-after 'unpack 'remove-pyqi
661 (lambda _
662 (substitute* "setup.py"
663 (("install_requires.append\\(\"pyqi\"\\)") "pass"))
664 #t)))
00e10c6e 665 ,@(package-arguments base))))))
9e12eba8 666
f7283db3
RW		 667(define-public bioperl-minimal
668 (let* ((inputs `(("perl-module-build" ,perl-module-build)
669 ("perl-data-stag" ,perl-data-stag)
670 ("perl-libwww" ,perl-libwww)
671 ("perl-uri" ,perl-uri)))
672 (transitive-inputs
673 (map (compose package-name cadr)
674 (delete-duplicates
675 (concatenate
676 (map (compose package-transitive-target-inputs cadr) inputs))))))
677 (package
678 (name "bioperl-minimal")
c70271ec 679 (version "1.7.0")
f7283db3
RW		 680 (source
681 (origin
682 (method url-fetch)
c70271ec
RW		 683 (uri (string-append "https://github.com/bioperl/bioperl-live/"
684 "archive/release-"
685 (string-map (lambda (c)
686 (if (char=? c #\.)
687 #\- c)) version)
688 ".tar.gz"))
f7283db3
RW		 689 (sha256
690 (base32
c70271ec 691 "12phgpxwgkqflkwfb9dcqg7a31dpjlfhar8wcgv0aj5ln4akfz06"))))
f7283db3
RW		 692 (build-system perl-build-system)
693 (arguments
694 `(#:phases
695 (modify-phases %standard-phases
696 (add-after
697 'install 'wrap-programs
698 (lambda* (#:key outputs #:allow-other-keys)
699 ;; Make sure all executables in "bin" find the required Perl
700 ;; modules at runtime. As the PERL5LIB variable contains also
701 ;; the paths of native inputs, we pick the transitive target
702 ;; inputs from %build-inputs.
703 (let* ((out (assoc-ref outputs "out"))
704 (bin (string-append out "/bin/"))
705 (path (string-join
706 (cons (string-append out "/lib/perl5/site_perl")
707 (map (lambda (name)
708 (assoc-ref %build-inputs name))
709 ',transitive-inputs))
710 ":")))
711 (for-each (lambda (file)
712 (wrap-program file
713 `("PERL5LIB" ":" prefix (,path))))
714 (find-files bin "\\.pl$"))
715 #t))))))
716 (inputs inputs)
717 (native-inputs
718 `(("perl-test-most" ,perl-test-most)))
719 (home-page "http://search.cpan.org/dist/BioPerl")
720 (synopsis "Bioinformatics toolkit")
721 (description
722 "BioPerl is the product of a community effort to produce Perl code which
723is useful in biology. Examples include Sequence objects, Alignment objects
724and database searching objects. These objects not only do what they are
725advertised to do in the documentation, but they also interact - Alignment
726objects are made from the Sequence objects, Sequence objects have access to
727Annotation and SeqFeature objects and databases, Blast objects can be
728converted to Alignment objects, and so on. This means that the objects
729provide a coordinated and extensible framework to do computational biology.")
2f3108ad 730 (license license:perl-license))))
f7283db3 731
85c37e29
RW		 732(define-public python-biopython
733 (package
734 (name "python-biopython")
af6ce610 735 (version "1.70")
85c37e29
RW		 736 (source (origin
737 (method url-fetch)
e815c094
BW		 738 ;; use PyPi rather than biopython.org to ease updating
739 (uri (pypi-uri "biopython" version))
85c37e29
RW		 740 (sha256
741 (base32
af6ce610 742 "0nz4n9d2y2dg849gn1z0vjlkwcpzzkzy3fij7x94a6ixy2c54z2a"))))
85c37e29 743 (build-system python-build-system)
4ce60305
BW		 744 (arguments
745 `(#:phases
746 (modify-phases %standard-phases
747 (add-before 'check 'set-home
748 ;; Some tests require a home directory to be set.
749 (lambda _ (setenv "HOME" "/tmp") #t)))))
f22efa01 750 (propagated-inputs
85c37e29 751 `(("python-numpy" ,python-numpy)))
85c37e29
RW		 752 (home-page "http://biopython.org/")
753 (synopsis "Tools for biological computation in Python")
754 (description
755 "Biopython is a set of tools for biological computation including parsers
756for bioinformatics files into Python data structures; interfaces to common
757bioinformatics programs; a standard sequence class and tools for performing
758common operations on them; code to perform data classification; code for
759dealing with alignments; code making it easy to split up parallelizable tasks
760into separate processes; and more.")
5c31f4aa 761 (license (license:non-copyleft "http://www.biopython.org/DIST/LICENSE"))))
85c37e29
RW		 762
763(define-public python2-biopython
5c31f4aa 764 (package-with-python2 python-biopython))
85c37e29 765
4b1a1528
BW		 766;; An outdated version of biopython is required for seqmagick, see
767;; https://github.com/fhcrc/seqmagick/issues/59
768;; When that issue has been resolved this package should be removed.
769(define python2-biopython-1.66
770 (package
771 (inherit python2-biopython)
772 (version "1.66")
773 (source (origin
774 (method url-fetch)
775 (uri (pypi-uri "biopython" version))
776 (sha256
777 (base32
778 "1gdv92593klimg22icf5j9by7xiq86jnwzkpz4abaa05ylkdf6hp"))))))
779
985d8411
BW		 780(define-public bpp-core
781 ;; The last release was in 2014 and the recommended way to install from source
782 ;; is to clone the git repository, so we do this.
783 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
784 (let ((commit "7d8bced0d1a87291ea8dd7046b7fb5ff9c35c582"))
785 (package
786 (name "bpp-core")
787 (version (string-append "2.2.0-1." (string-take commit 7)))
788 (source (origin
789 (method git-fetch)
790 (uri (git-reference
791 (url "http://biopp.univ-montp2.fr/git/bpp-core")
792 (commit commit)))
793 (file-name (string-append name "-" version "-checkout"))
794 (sha256
795 (base32
796 "10djsq5vlnkilv436gnmh4irpk49v29pa69r6xiryg32xmvn909j"))))
797 (build-system cmake-build-system)
798 (arguments
799 `(#:parallel-build? #f))
800 (inputs
801 `(("gcc" ,gcc-5))) ; Compilation of bpp-phyl fails with GCC 4.9 so we
802 ; compile all of the bpp packages with GCC 5.
803 (home-page "http://biopp.univ-montp2.fr")
804 (synopsis "C++ libraries for Bioinformatics")
805 (description
806 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
807analysis, phylogenetics, molecular evolution and population genetics. It is
808Object Oriented and is designed to be both easy to use and computer efficient.
809Bio++ intends to help programmers to write computer expensive programs, by
810providing them a set of re-usable tools.")
811 (license license:cecill-c))))
812
8b5f4d57
BW		 813(define-public bpp-phyl
814 ;; The last release was in 2014 and the recommended way to install from source
815 ;; is to clone the git repository, so we do this.
816 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
817 (let ((commit "0c07167b629f68b569bf274d1ad0c4af83276ae2"))
818 (package
819 (name "bpp-phyl")
820 (version (string-append "2.2.0-1." (string-take commit 7)))
821 (source (origin
822 (method git-fetch)
823 (uri (git-reference
824 (url "http://biopp.univ-montp2.fr/git/bpp-phyl")
825 (commit commit)))
826 (file-name (string-append name "-" version "-checkout"))
827 (sha256
828 (base32
829 "1ssjgchzwj3iai26kyly7gwkdv8sk59nqhkb1wpap3sf5m6kyllh"))))
830 (build-system cmake-build-system)
831 (arguments
832 `(#:parallel-build? #f
833 ;; If out-of-source, test data is not copied into the build directory
834 ;; so the tests fail.
835 #:out-of-source? #f))
836 (inputs
837 `(("bpp-core" ,bpp-core)
838 ("bpp-seq" ,bpp-seq)
839 ;; GCC 4.8 fails due to an 'internal compiler error', so we use a more
840 ;; modern GCC.
841 ("gcc" ,gcc-5)))
842 (home-page "http://biopp.univ-montp2.fr")
843 (synopsis "Bio++ phylogenetic Library")
844 (description
845 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
846analysis, phylogenetics, molecular evolution and population genetics. This
847library provides phylogenetics-related modules.")
848 (license license:cecill-c))))
849
159a7016
BW		 850(define-public bpp-popgen
851 ;; The last release was in 2014 and the recommended way to install from source
852 ;; is to clone the git repository, so we do this.
853 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
854 (let ((commit "e472bac9b1a148803895d747cd6d0c5904f85d9f"))
855 (package
856 (name "bpp-popgen")
857 (version (string-append "2.2.0-1." (string-take commit 7)))
858 (source (origin
859 (method git-fetch)
860 (uri (git-reference
861 (url "http://biopp.univ-montp2.fr/git/bpp-popgen")
862 (commit commit)))
863 (file-name (string-append name "-" version "-checkout"))
864 (sha256
865 (base32
866 "0yn82dzn1n5629nzja68xfrhi655709rjanyryb36vzkmymy6dw5"))))
867 (build-system cmake-build-system)
868 (arguments
869 `(#:parallel-build? #f
870 #:tests? #f)) ; There are no tests.
871 (inputs
872 `(("bpp-core" ,bpp-core)
873 ("bpp-seq" ,bpp-seq)
874 ("gcc" ,gcc-5)))
875 (home-page "http://biopp.univ-montp2.fr")
876 (synopsis "Bio++ population genetics library")
877 (description
878 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
879analysis, phylogenetics, molecular evolution and population genetics. This
880library provides population genetics-related modules.")
881 (license license:cecill-c))))
882
70f1bc05
BW		 883(define-public bpp-seq
884 ;; The last release was in 2014 and the recommended way to install from source
885 ;; is to clone the git repository, so we do this.
886 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
887 (let ((commit "6cfa07965ce152e5598a89df2fa80a75973bfa33"))
888 (package
889 (name "bpp-seq")
890 (version (string-append "2.2.0-1." (string-take commit 7)))
891 (source (origin
892 (method git-fetch)
893 (uri (git-reference
894 (url "http://biopp.univ-montp2.fr/git/bpp-seq")
895 (commit commit)))
896 (file-name (string-append name "-" version "-checkout"))
897 (sha256
898 (base32
899 "1nys5jq7jqvdg40d91wsmj3q2yzy4276cp7sp44n67p468f27zf2"))))
900 (build-system cmake-build-system)
901 (arguments
902 `(#:parallel-build? #f
903 ;; If out-of-source, test data is not copied into the build directory
904 ;; so the tests fail.
905 #:out-of-source? #f))
906 (inputs
907 `(("bpp-core" ,bpp-core)
908 ("gcc" ,gcc-5))) ; Use GCC 5 as per 'bpp-core'.
909 (home-page "http://biopp.univ-montp2.fr")
910 (synopsis "Bio++ sequence library")
911 (description
912 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
913analysis, phylogenetics, molecular evolution and population genetics. This
914library provides sequence-related modules.")
915 (license license:cecill-c))))
916
db7a3444
BW		 917(define-public bppsuite
918 ;; The last release was in 2014 and the recommended way to install from source
919 ;; is to clone the git repository, so we do this.
920 ;; http://biopp.univ-montp2.fr/wiki/index.php/Main_Page
921 (let ((commit "c516147f57aa50961121cd505bed52cd7603698b"))
922 (package
923 (name "bppsuite")
924 (version (string-append "2.2.0-1." (string-take commit 7)))
925 (source (origin
926 (method git-fetch)
927 (uri (git-reference
928 (url "http://biopp.univ-montp2.fr/git/bppsuite")
929 (commit commit)))
930 (file-name (string-append name "-" version "-checkout"))
931 (sha256
932 (base32
933 "1y87pxvw0jxjizhq2dr9g2r91md45k1p9ih2sl1yy1y3p934l2kb"))))
934 (build-system cmake-build-system)
935 (arguments
936 `(#:parallel-build? #f
937 #:tests? #f)) ; There are no tests.
938 (native-inputs
939 `(("groff" ,groff)
940 ("man-db" ,man-db)
941 ("texinfo" ,texinfo)))
942 (inputs
943 `(("bpp-core" ,bpp-core)
944 ("bpp-seq" ,bpp-seq)
945 ("bpp-phyl" ,bpp-phyl)
946 ("bpp-phyl" ,bpp-popgen)
947 ("gcc" ,gcc-5)))
948 (home-page "http://biopp.univ-montp2.fr")
949 (synopsis "Bioinformatics tools written with the Bio++ libraries")
950 (description
951 "Bio++ is a set of C++ libraries for Bioinformatics, including sequence
952analysis, phylogenetics, molecular evolution and population genetics. This
953package provides command line tools using the Bio++ library.")
954 (license license:cecill-c))))
955
82c370de
RW		 956(define-public blast+
957 (package
958 (name "blast+")
4732e6ee 959 (version "2.6.0")
82c370de
RW		 960 (source (origin
961 (method url-fetch)
962 (uri (string-append
963 "ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/"
964 version "/ncbi-blast-" version "+-src.tar.gz"))
965 (sha256
966 (base32
4732e6ee
BW		 967 "15n937pw5aqmyfjb6l387d18grqbb96l63d5xj4l7yyh0zbf2405"))
968 (patches (search-patches "blast+-fix-makefile.patch"))
82c370de
RW		 969 (modules '((guix build utils)))
970 (snippet
971 '(begin
4732e6ee 972 ;; Remove bundled bzip2, zlib and pcre.
82c370de
RW		 973 (delete-file-recursively "c++/src/util/compress/bzip2")
974 (delete-file-recursively "c++/src/util/compress/zlib")
4732e6ee 975 (delete-file-recursively "c++/src/util/regexp")
82c370de
RW		 976 (substitute* "c++/src/util/compress/Makefile.in"
977 (("bzip2 zlib api") "api"))
978 ;; Remove useless msbuild directory
979 (delete-file-recursively
980 "c++/src/build-system/project_tree_builder/msbuild")
981 #t))))
982 (build-system gnu-build-system)
983 (arguments
4732e6ee 984 `(;; There are two(!) tests for this massive library, and both fail with
82c370de 985 ;; "unparsable timing stats".
82c370de
RW		 986 ;; ERR [127] -- [serial/datatool] datatool.sh (unparsable timing stats)
987 ;; ERR [127] -- [serial/datatool] datatool_xml.sh (unparsable timing stats)
988 #:tests? #f
989 #:out-of-source? #t
990 #:parallel-build? #f ; not supported
991 #:phases
992 (modify-phases %standard-phases
993 (add-before
994 'configure 'set-HOME
995 ;; $HOME needs to be set at some point during the configure phase
996 (lambda _ (setenv "HOME" "/tmp") #t))
997 (add-after
998 'unpack 'enter-dir
999 (lambda _ (chdir "c++") #t))
1000 (add-after
1001 'enter-dir 'fix-build-system
1002 (lambda _
1003 (define (which* cmd)
1004 (cond ((string=? cmd "date")
1005 ;; make call to "date" deterministic
1006 "date -d @0")
1007 ((which cmd)
1008 => identity)
1009 (else
1010 (format (current-error-port)
1011 "WARNING: Unable to find absolute path for ~s~%"
1012 cmd)
1013 #f)))
1014
1015 ;; Rewrite hardcoded paths to various tools
1016 (substitute* (append '("src/build-system/configure.ac"
1017 "src/build-system/configure"
4732e6ee 1018 "src/build-system/helpers/run_with_lock.c"
82c370de
RW		 1019 "scripts/common/impl/if_diff.sh"
1020 "scripts/common/impl/run_with_lock.sh"
1021 "src/build-system/Makefile.configurables.real"
1022 "src/build-system/Makefile.in.top"
1023 "src/build-system/Makefile.meta.gmake=no"
1024 "src/build-system/Makefile.meta.in"
1025 "src/build-system/Makefile.meta_l"
1026 "src/build-system/Makefile.meta_p"
1027 "src/build-system/Makefile.meta_r"
1028 "src/build-system/Makefile.mk.in"
1029 "src/build-system/Makefile.requirements"
1030 "src/build-system/Makefile.rules_with_autodep.in")
1031 (find-files "scripts/common/check" "\\.sh$"))
1032 (("(/usr/bin/|/bin/)([a-z][-_.a-z]*)" all dir cmd)
1033 (or (which* cmd) all)))
1034
1035 (substitute* (find-files "src/build-system" "^config.*")
1036 (("LN_S=/bin/\\$LN_S") (string-append "LN_S=" (which "ln")))
1037 (("^PATH=.*") ""))
1038
1039 ;; rewrite "/var/tmp" in check script
1040 (substitute* "scripts/common/check/check_make_unix.sh"
1041 (("/var/tmp") "/tmp"))
1042
1043 ;; do not reset PATH
1044 (substitute* (find-files "scripts/common/impl/" "\\.sh$")
1045 (("^ *PATH=.*") "")
1046 (("action=/bin/") "action=")
1047 (("export PATH") ":"))
1048 #t))
1049 (replace
1050 'configure
1051 (lambda* (#:key inputs outputs #:allow-other-keys)
1052 (let ((out (assoc-ref outputs "out"))
1053 (lib (string-append (assoc-ref outputs "lib") "/lib"))
1054 (include (string-append (assoc-ref outputs "include")
1055 "/include/ncbi-tools++")))
1056 ;; The 'configure' script doesn't recognize things like
1057 ;; '--enable-fast-install'.
1058 (zero? (system* "./configure.orig"
1059 (string-append "--with-build-root=" (getcwd) "/build")
1060 (string-append "--prefix=" out)
1061 (string-append "--libdir=" lib)
1062 (string-append "--includedir=" include)
1063 (string-append "--with-bz2="
1064 (assoc-ref inputs "bzip2"))
1065 (string-append "--with-z="
1066 (assoc-ref inputs "zlib"))
4732e6ee
BW		 1067 (string-append "--with-pcre="
1068 (assoc-ref inputs "pcre"))
82c370de
RW		 1069 ;; Each library is built twice by default, once
1070 ;; with "-static" in its name, and again
1071 ;; without.
1072 "--without-static"
1073 "--with-dll"))))))))
4732e6ee
BW		 1074 (outputs '("out" ; 21 MB
1075 "lib" ; 226 MB
1076 "include")) ; 33 MB
82c370de
RW		 1077 (inputs
1078 `(("bzip2" ,bzip2)
4732e6ee
BW		 1079 ("zlib" ,zlib)
1080 ("pcre" ,pcre)
1081 ("perl" ,perl)
1082 ("python" ,python-wrapper)))
82c370de
RW		 1083 (native-inputs
1084 `(("cpio" ,cpio)))
1085 (home-page "http://blast.ncbi.nlm.nih.gov")
1086 (synopsis "Basic local alignment search tool")
1087 (description
1088 "BLAST is a popular method of performing a DNA or protein sequence
1089similarity search, using heuristics to produce results quickly. It also
1090calculates an “expect value” that estimates how many matches would have
1091occurred at a given score by chance, which can aid a user in judging how much
1092confidence to have in an alignment.")
1093 ;; Most of the sources are in the public domain, with the following
1094 ;; exceptions:
1095 ;; * Expat:
1096 ;; * ./c++/include/util/bitset/
1097 ;; * ./c++/src/html/ncbi_menu*.js
1098 ;; * Boost license:
1099 ;; * ./c++/include/util/impl/floating_point_comparison.hpp
1100 ;; * LGPL 2+:
1101 ;; * ./c++/include/dbapi/driver/odbc/unix_odbc/
1102 ;; * ASL 2.0:
1103 ;; * ./c++/src/corelib/teamcity_*
1104 (license (list license:public-domain
1105 license:expat
1106 license:boost1.0
1107 license:lgpl2.0+
1108 license:asl2.0))))
1109
6c2b26e2
RW		 1110(define-public bless
1111 (package
1112 (name "bless")
1113 (version "1p02")
1114 (source (origin
1115 (method url-fetch)
1116 (uri (string-append "mirror://sourceforge/bless-ec/bless.v"
1117 version ".tgz"))
1118 (sha256
1119 (base32
4d75e03a
RW		 1120 "0rm0gw2s18dqwzzpl3c2x1z05ni2v0xz5dmfk3d33j6g4cgrlrdd"))
1121 (modules '((guix build utils)))
6c2b26e2
RW		 1122 (snippet
1123 `(begin
1124 ;; Remove bundled boost, pigz, zlib, and .git directory
953c1223
RW		 1125 ;; FIXME: also remove bundled sources for murmurhash3 and
1126 ;; kmc once packaged.
6c2b26e2
RW		 1127 (delete-file-recursively "boost")
1128 (delete-file-recursively "pigz")
953c1223 1129 (delete-file-recursively "google-sparsehash")
6c2b26e2
RW		 1130 (delete-file-recursively "zlib")
1131 (delete-file-recursively ".git")
1132 #t))))
1133 (build-system gnu-build-system)
1134 (arguments
1135 '(#:tests? #f ;no "check" target
1136 #:make-flags
1137 (list (string-append "ZLIB="
1138 (assoc-ref %build-inputs "zlib")
1139 "/lib/libz.a")
1140 (string-append "LDFLAGS="
1141 (string-join '("-lboost_filesystem"
1142 "-lboost_system"
1143 "-lboost_iostreams"
1144 "-lz"
1145 "-fopenmp"
1146 "-std=c++11"))))
1147 #:phases
1148 (modify-phases %standard-phases
1149 (add-after 'unpack 'do-not-build-bundled-pigz
1150 (lambda* (#:key inputs outputs #:allow-other-keys)
1151 (substitute* "Makefile"
1152 (("cd pigz/pigz-2.3.3; make") ""))
1153 #t))
1154 (add-after 'unpack 'patch-paths-to-executables
1155 (lambda* (#:key inputs outputs #:allow-other-keys)
1156 (substitute* "parse_args.cpp"
1157 (("kmc_binary = .*")
1158 (string-append "kmc_binary = \""
1159 (assoc-ref outputs "out")
1160 "/bin/kmc\";"))
1161 (("pigz_binary = .*")
1162 (string-append "pigz_binary = \""
1163 (assoc-ref inputs "pigz")
1164 "/bin/pigz\";")))
1165 #t))
1166 (replace 'install
1167 (lambda* (#:key outputs #:allow-other-keys)
1168 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
1169 (for-each (lambda (file)
1170 (install-file file bin))
1171 '("bless" "kmc/bin/kmc"))
1172 #t)))
1173 (delete 'configure))))
1174 (native-inputs
1175 `(("perl" ,perl)))
1176 (inputs
1177 `(("openmpi" ,openmpi)
1178 ("boost" ,boost)
953c1223 1179 ("sparsehash" ,sparsehash)
6c2b26e2
RW		 1180 ("pigz" ,pigz)
1181 ("zlib" ,zlib)))
9641a899 1182 (supported-systems '("x86_64-linux"))
3b3b60d0 1183 (home-page "https://sourceforge.net/p/bless-ec/wiki/Home/")
6c2b26e2
RW		 1184 (synopsis "Bloom-filter-based error correction tool for NGS reads")
1185 (description
1186 "@dfn{Bloom-filter-based error correction solution for high-throughput
1187sequencing reads} (BLESS) uses a single minimum-sized bloom filter is a
1188correction tool for genomic reads produced by @dfn{Next-generation
1189sequencing} (NGS). BLESS produces accurate correction results with much less
1190memory compared with previous solutions and is also able to tolerate a higher
1191false-positive rate. BLESS can extend reads like DNA assemblers to correct
1192errors at the end of reads.")
1193 (license license:gpl3+)))
1194
2c7ee167
RW		 1195(define-public bowtie
1196 (package
1197 (name "bowtie")
d6e63cf3 1198 (version "2.3.2")
2c7ee167
RW		 1199 (source (origin
1200 (method url-fetch)
1201 (uri (string-append "https://github.com/BenLangmead/bowtie2/archive/v"
1202 version ".tar.gz"))
f586c877 1203 (file-name (string-append name "-" version ".tar.gz"))
2c7ee167
RW		 1204 (sha256
1205 (base32
d6e63cf3 1206 "0hwa5r9qbglppb7sz5z79rlmmddr3n51n468jb3wh8rwjgn3yr90"))
2c7ee167
RW		 1207 (modules '((guix build utils)))
1208 (snippet
1209 '(substitute* "Makefile"
2c7ee167
RW		 1210 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
1211 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
0047d26a 1212 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\"")))))
2c7ee167 1213 (build-system gnu-build-system)
d6e63cf3
BW		 1214 (inputs
1215 `(("perl" ,perl)
1216 ("perl-clone" ,perl-clone)
1217 ("perl-test-deep" ,perl-test-deep)
1218 ("perl-test-simple" ,perl-test-simple)
1219 ("python" ,python-2)
1220 ("tbb" ,tbb)
1221 ("zlib" ,zlib)))
2c7ee167 1222 (arguments
0047d26a
RW		 1223 '(#:make-flags
1224 (list "allall"
1225 "WITH_TBB=1"
1226 (string-append "prefix=" (assoc-ref %outputs "out")))
2c7ee167 1227 #:phases
06e37236
BW		 1228 (modify-phases %standard-phases
1229 (delete 'configure)
1230 (replace 'check
1231 (lambda* (#:key outputs #:allow-other-keys)
1232 (zero? (system* "perl"
1233 "scripts/test/simple_tests.pl"
1234 "--bowtie2=./bowtie2"
1235 "--bowtie2-build=./bowtie2-build")))))))
2c7ee167
RW		 1236 (home-page "http://bowtie-bio.sourceforge.net/bowtie2/index.shtml")
1237 (synopsis "Fast and sensitive nucleotide sequence read aligner")
1238 (description
1239 "Bowtie 2 is a fast and memory-efficient tool for aligning sequencing
1240reads to long reference sequences. It is particularly good at aligning reads
1241of about 50 up to 100s or 1,000s of characters, and particularly good at
1242aligning to relatively long (e.g. mammalian) genomes. Bowtie 2 indexes the
1243genome with an FM Index to keep its memory footprint small: for the human
1244genome, its memory footprint is typically around 3.2 GB. Bowtie 2 supports
1245gapped, local, and paired-end alignment modes.")
241e1221 1246 (supported-systems '("x86_64-linux"))
2c7ee167
RW		 1247 (license license:gpl3+)))
1248
94ce537e
RW		 1249(define-public tophat
1250 (package
1251 (name "tophat")
1252 (version "2.1.0")
1253 (source (origin
1254 (method url-fetch)
1255 (uri (string-append
1256 "http://ccb.jhu.edu/software/tophat/downloads/tophat-"
1257 version ".tar.gz"))
1258 (sha256
1259 (base32
1260 "168zlzykq622zbgkh90a90f1bdgsxkscq2zxzbj8brq80hbjpyp7"))
fc1adab1 1261 (patches (search-patches "tophat-build-with-later-seqan.patch"))
94ce537e
RW		 1262 (modules '((guix build utils)))
1263 (snippet
1264 '(begin
1265 ;; Remove bundled SeqAn and samtools
1266 (delete-file-recursively "src/SeqAn-1.3")
1267 (delete-file-recursively "src/samtools-0.1.18")
1268 #t))))
1269 (build-system gnu-build-system)
1270 (arguments
1271 '(#:parallel-build? #f ; not supported
1272 #:phases
1273 (modify-phases %standard-phases
1274 (add-after 'unpack 'use-system-samtools
1275 (lambda* (#:key inputs #:allow-other-keys)
1276 (substitute* "src/Makefile.in"
1277 (("(noinst_LIBRARIES = )\\$\\(SAMLIB\\)" _ prefix) prefix)
1278 (("\\$\\(SAMPROG\\): \\$\\(SAMLIB\\)") "")
1279 (("SAMPROG = samtools_0\\.1\\.18") "")
1280 (("\\$\\(samtools_0_1_18_SOURCES\\)") "")
1281 (("am__EXEEXT_1 = samtools_0\\.1\\.18\\$\\(EXEEXT\\)") ""))
1282 (substitute* '("src/common.cpp"
1283 "src/tophat.py")
1284 (("samtools_0.1.18") (which "samtools")))
1285 (substitute* '("src/common.h"
1286 "src/bam2fastx.cpp")
1287 (("#include \"bam.h\"") "#include <samtools/bam.h>")
1288 (("#include \"sam.h\"") "#include <samtools/sam.h>"))
1289 (substitute* '("src/bwt_map.h"
1290 "src/map2gtf.h"
1291 "src/align_status.h")
1292 (("#include <bam.h>") "#include <samtools/bam.h>")
1293 (("#include <sam.h>") "#include <samtools/sam.h>"))
1294 #t)))))
1295 (inputs
1296 `(("boost" ,boost)
1297 ("bowtie" ,bowtie)
1298 ("samtools" ,samtools-0.1)
1299 ("ncurses" ,ncurses)
1300 ("python" ,python-2)
1301 ("perl" ,perl)
1302 ("zlib" ,zlib)
1303 ("seqan" ,seqan)))
1304 (home-page "http://ccb.jhu.edu/software/tophat/index.shtml")
1305 (synopsis "Spliced read mapper for RNA-Seq data")
1306 (description
1307 "TopHat is a fast splice junction mapper for nucleotide sequence
1308reads produced by the RNA-Seq method. It aligns RNA-Seq reads to
1309mammalian-sized genomes using the ultra high-throughput short read
1310aligner Bowtie, and then analyzes the mapping results to identify
1311splice junctions between exons.")
1312 ;; TopHat is released under the Boost Software License, Version 1.0
1313 ;; See https://github.com/infphilo/tophat/issues/11#issuecomment-121589893
1314 (license license:boost1.0)))
1315
9a8336d8
RW		 1316(define-public bwa
1317 (package
1318 (name "bwa")
6f141eff 1319 (version "0.7.17")
9a8336d8
RW		 1320 (source (origin
1321 (method url-fetch)
ae6e00f6
BW		 1322 (uri (string-append
1323 "https://github.com/lh3/bwa/releases/download/v"
1324 version "/bwa-" version ".tar.bz2"))
9a8336d8
RW		 1325 (sha256
1326 (base32
6f141eff 1327 "1zfhv2zg9v1icdlq4p9ssc8k01mca5d1bd87w71py2swfi74s6yy"))))
9a8336d8
RW		 1328 (build-system gnu-build-system)
1329 (arguments
1330 '(#:tests? #f ;no "check" target
1331 #:phases
dc1d3cde
KK		 1332 (modify-phases %standard-phases
1333 (replace 'install
1334 (lambda* (#:key outputs #:allow-other-keys)
1335 (let ((bin (string-append
1336 (assoc-ref outputs "out") "/bin"))
1337 (doc (string-append
1338 (assoc-ref outputs "out") "/share/doc/bwa"))
1339 (man (string-append
1340 (assoc-ref outputs "out") "/share/man/man1")))
1341 (install-file "bwa" bin)
1342 (install-file "README.md" doc)
1343 (install-file "bwa.1" man))
1344 #t))
1345 ;; no "configure" script
1346 (delete 'configure))))
9a8336d8 1347 (inputs `(("zlib" ,zlib)))
db94f8c7
RW		 1348 ;; Non-portable SSE instructions are used so building fails on platforms
1349 ;; other than x86_64.
1350 (supported-systems '("x86_64-linux"))
9a8336d8
RW		 1351 (home-page "http://bio-bwa.sourceforge.net/")
1352 (synopsis "Burrows-Wheeler sequence aligner")
1353 (description
1354 "BWA is a software package for mapping low-divergent sequences against a
1355large reference genome, such as the human genome. It consists of three
1356algorithms: BWA-backtrack, BWA-SW and BWA-MEM. The first algorithm is
1357designed for Illumina sequence reads up to 100bp, while the rest two for
1358longer sequences ranged from 70bp to 1Mbp. BWA-MEM and BWA-SW share similar
1359features such as long-read support and split alignment, but BWA-MEM, which is
1360the latest, is generally recommended for high-quality queries as it is faster
1361and more accurate. BWA-MEM also has better performance than BWA-backtrack for
136270-100bp Illumina reads.")
1363 (license license:gpl3+)))
1364
d29150b5
RW		 1365(define-public bwa-pssm
1366 (package (inherit bwa)
1367 (name "bwa-pssm")
1368 (version "0.5.11")
1369 (source (origin
1370 (method url-fetch)
1371 (uri (string-append "https://github.com/pkerpedjiev/bwa-pssm/"
1372 "archive/" version ".tar.gz"))
1373 (file-name (string-append name "-" version ".tar.gz"))
1374 (sha256
1375 (base32
1376 "02p7mpbs4mlxmn84g2x4ghak638vbj4lqix2ipx5g84pz9bhdavg"))))
1377 (build-system gnu-build-system)
1378 (inputs
1379 `(("gdsl" ,gdsl)
1380 ("zlib" ,zlib)
1381 ("perl" ,perl)))
1382 (home-page "http://bwa-pssm.binf.ku.dk/")
1383 (synopsis "Burrows-Wheeler transform-based probabilistic short read mapper")
1384 (description
1385 "BWA-PSSM is a probabilistic short genomic sequence read aligner based on
1386the use of @dfn{position specific scoring matrices} (PSSM). Like many of the
1387existing aligners it is fast and sensitive. Unlike most other aligners,
1388however, it is also adaptible in the sense that one can direct the alignment
1389based on known biases within the data set. It is coded as a modification of
1390the original BWA alignment program and shares the genome index structure as
1391well as many of the command line options.")
1392 (license license:gpl3+)))
1393
ad641d53
RW		 1394(define-public python2-bx-python
1395 (package
1396 (name "python2-bx-python")
c1dfe8c3 1397 (version "0.7.3")
ad641d53
RW		 1398 (source (origin
1399 (method url-fetch)
c1dfe8c3 1400 (uri (pypi-uri "bx-python" version))
ad641d53
RW		 1401 (sha256
1402 (base32
c1dfe8c3 1403 "15z2w3bvnc0n4qmb9bd6d8ylc2h2nj883x2w9iixf4x3vki9b22i"))
ad641d53
RW		 1404 (modules '((guix build utils)))
1405 (snippet
1406 '(substitute* "setup.py"
1407 ;; remove dependency on outdated "distribute" module
1408 (("^from distribute_setup import use_setuptools") "")
1409 (("^use_setuptools\\(\\)") "")))))
1410 (build-system python-build-system)
1411 (arguments
1412 `(#:tests? #f ;tests fail because test data are not included
1413 #:python ,python-2))
1414 (inputs
1415 `(("python-numpy" ,python2-numpy)
1416 ("zlib" ,zlib)))
1417 (native-inputs
f3b98f4f 1418 `(("python-nose" ,python2-nose)))
ad641d53
RW		 1419 (home-page "http://bitbucket.org/james_taylor/bx-python/")
1420 (synopsis "Tools for manipulating biological data")
1421 (description
1422 "bx-python provides tools for manipulating biological data, particularly
1423multiple sequence alignments.")
1424 (license license:expat)))
1425
55a9a8c2
RW		 1426(define-public python-pysam
1427 (package
1428 (name "python-pysam")
60739965 1429 (version "0.11.2.2")
d454640c
RW		 1430 (source (origin
1431 (method url-fetch)
f536dce5
MB		 1432 ;; Test data is missing on PyPi.
1433 (uri (string-append
1434 "https://github.com/pysam-developers/pysam/archive/v"
1435 version ".tar.gz"))
1436 (file-name (string-append name "-" version ".tar.gz"))
d454640c
RW		 1437 (sha256
1438 (base32
60739965 1439 "1cfqdxsqs3xhacns9n0271ck6wkc76px66ddjm91wfw2jxxfklvc"))
dff26b23
MB		 1440 (modules '((guix build utils)))
1441 (snippet
1442 ;; Drop bundled htslib. TODO: Also remove samtools and bcftools.
1443 '(delete-file-recursively "htslib"))))
55a9a8c2
RW		 1444 (build-system python-build-system)
1445 (arguments
71dbf592
RW		 1446 `(#:modules ((ice-9 ftw)
1447 (srfi srfi-26)
1448 (guix build python-build-system)
1449 (guix build utils))
1450 #:phases
397d463a
MB		 1451 (modify-phases %standard-phases
1452 (add-before 'build 'set-flags
dff26b23
MB		 1453 (lambda* (#:key inputs #:allow-other-keys)
1454 (setenv "HTSLIB_MODE" "external")
1455 (setenv "HTSLIB_LIBRARY_DIR"
1456 (string-append (assoc-ref inputs "htslib") "/lib"))
1457 (setenv "HTSLIB_INCLUDE_DIR"
1458 (string-append (assoc-ref inputs "htslib") "/include"))
397d463a
MB		 1459 (setenv "LDFLAGS" "-lncurses")
1460 (setenv "CFLAGS" "-D_CURSES_LIB=1")
f536dce5 1461 #t))
71dbf592 1462 (replace 'check
f536dce5 1463 (lambda* (#:key inputs outputs #:allow-other-keys)
71dbf592 1464 ;; Add first subdirectory of "build" directory to PYTHONPATH.
f536dce5
MB		 1465 (setenv "PYTHONPATH"
1466 (string-append
1467 (getenv "PYTHONPATH")
71dbf592
RW		 1468 ":" (getcwd) "/build/"
1469 (car (scandir "build"
e1f02f92 1470 (negate (cut string-prefix? "." <>))))))
f536dce5 1471 ;; Step out of source dir so python does not import from CWD.
71dbf592
RW		 1472 (with-directory-excursion "tests"
1473 (setenv "HOME" "/tmp")
1474 (and (zero? (system* "make" "-C" "pysam_data"))
1475 (zero? (system* "make" "-C" "cbcf_data"))
b2955e22
RW		 1476 ;; Running nosetests without explicitly asking for a
1477 ;; single process leads to a crash. Running with multiple
1478 ;; processes fails because the tests are not designed to
1479 ;; run in parallel.
31c374e0
RW		 1480
1481 ;; FIXME: tests keep timing out on some systems.
1482 ;; (zero? (system* "nosetests" "-v"
1483 ;; "--processes" "1"))
1484 )))))))
dff26b23
MB		 1485 (propagated-inputs
1486 `(("htslib" ,htslib))) ; Included from installed header files.
55a9a8c2 1487 (inputs
649e9b3b 1488 `(("ncurses" ,ncurses)
55a9a8c2 1489 ("zlib" ,zlib)))
649e9b3b
RW		 1490 (native-inputs
1491 `(("python-cython" ,python-cython)
f536dce5
MB		 1492 ;; Dependencies below are are for tests only.
1493 ("samtools" ,samtools)
1494 ("bcftools" ,bcftools)
1495 ("python-nose" ,python-nose)))
55a9a8c2
RW		 1496 (home-page "https://github.com/pysam-developers/pysam")
1497 (synopsis "Python bindings to the SAMtools C API")
1498 (description
1499 "Pysam is a Python module for reading and manipulating files in the
1500SAM/BAM format. Pysam is a lightweight wrapper of the SAMtools C API. It
1501also includes an interface for tabix.")
1502 (license license:expat)))
1503
1504(define-public python2-pysam
1505 (package-with-python2 python-pysam))
1506
4db9433a
RW		 1507(define-public python-twobitreader
1508 (package
1509 (name "python-twobitreader")
044ac8d2 1510 (version "3.1.4")
4db9433a
RW		 1511 (source (origin
1512 (method url-fetch)
1513 (uri (pypi-uri "twobitreader" version))
1514 (sha256
1515 (base32
044ac8d2 1516 "1q8wnj2kga9nz1lwc4w7qv52smfm536hp6mc8w6s53lhyj0mpi22"))))
4db9433a 1517 (build-system python-build-system)
900fb8d0
LF		 1518 (arguments
1519 '(;; Tests are not distributed in the PyPi release.
1520 ;; TODO Try building from the Git repo or asking the upstream maintainer
1521 ;; to distribute the tests on PyPi.
1522 #:tests? #f))
4db9433a
RW		 1523 (native-inputs
1524 `(("python-sphinx" ,python-sphinx)))
1525 (home-page "https://github.com/benjschiller/twobitreader")
1526 (synopsis "Python library for reading .2bit files")
1527 (description
1528 "twobitreader is a Python library for reading .2bit files as used by the
1529UCSC genome browser.")
1530 (license license:artistic2.0)))
1531
1532(define-public python2-twobitreader
5c31f4aa 1533 (package-with-python2 python-twobitreader))
4db9433a 1534
f94bf198
RW		 1535(define-public python-plastid
1536 (package
1537 (name "python-plastid")
897ab082 1538 (version "0.4.8")
f94bf198
RW		 1539 (source (origin
1540 (method url-fetch)
1541 (uri (pypi-uri "plastid" version))
1542 (sha256
1543 (base32
897ab082 1544 "0l24dd3q66if8yj042m4s0g95n6acn7im1imqd3p6h8ns43kxhj8"))))
f94bf198
RW		 1545 (build-system python-build-system)
1546 (arguments
1547 ;; Some test files are not included.
1548 `(#:tests? #f))
1549 (propagated-inputs
1550 `(("python-numpy" ,python-numpy)
1551 ("python-scipy" ,python-scipy)
1552 ("python-pandas" ,python-pandas)
1553 ("python-pysam" ,python-pysam)
1554 ("python-matplotlib" ,python-matplotlib)
1555 ("python-biopython" ,python-biopython)
99caa6f7
BW		 1556 ("python-twobitreader" ,python-twobitreader)
1557 ("python-termcolor" ,python-termcolor)))
f94bf198
RW		 1558 (native-inputs
1559 `(("python-cython" ,python-cython)
1560 ("python-nose" ,python-nose)))
1561 (home-page "https://github.com/joshuagryphon/plastid")
1562 (synopsis "Python library for genomic analysis")
1563 (description
1564 "plastid is a Python library for genomic analysis – in particular,
1565high-throughput sequencing data – with an emphasis on simplicity.")
1566 (license license:bsd-3)))
1567
1568(define-public python2-plastid
5c31f4aa 1569 (package-with-python2 python-plastid))
f94bf198 1570
6c1305f9
RW		 1571(define-public cd-hit
1572 (package
1573 (name "cd-hit")
ba773f65 1574 (version "4.6.8")
6c1305f9
RW		 1575 (source (origin
1576 (method url-fetch)
1577 (uri (string-append "https://github.com/weizhongli/cdhit"
1578 "/releases/download/V" version
ba773f65
BW		 1579 "/cd-hit-v" version
1580 "-2017-0621-source.tar.gz"))
6c1305f9
RW		 1581 (sha256
1582 (base32
ba773f65 1583 "1386dg2npx8p62wmv08mjzsd2z3waknb9j1gg3gkvblcy57hymnn"))))
6c1305f9
RW		 1584 (build-system gnu-build-system)
1585 (arguments
1586 `(#:tests? #f ; there are no tests
1587 #:make-flags
1588 ;; Executables are copied directly to the PREFIX.
1589 (list (string-append "PREFIX=" (assoc-ref %outputs "out") "/bin"))
1590 #:phases
1591 (modify-phases %standard-phases
1592 ;; No "configure" script
1593 (delete 'configure)
1594 ;; Remove sources of non-determinism
1595 (add-after 'unpack 'be-timeless
1596 (lambda _
1597 (substitute* "cdhit-utility.c++"
1598 ((" \\(built on \" __DATE__ \"\\)") ""))
1599 (substitute* "cdhit-common.c++"
1600 (("__DATE__") "\"0\"")
1601 (("\", %s, \" __TIME__ \"\\\\n\", date") ""))
1602 #t))
ba773f65 1603 ;; The "install" target does not create the target directory.
6c1305f9
RW		 1604 (add-before 'install 'create-target-dir
1605 (lambda* (#:key outputs #:allow-other-keys)
1606 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
1607 #t)))))
1608 (inputs
1609 `(("perl" ,perl)))
1610 (home-page "http://weizhongli-lab.org/cd-hit/")
1611 (synopsis "Cluster and compare protein or nucleotide sequences")
1612 (description
1613 "CD-HIT is a program for clustering and comparing protein or nucleotide
1614sequences. CD-HIT is designed to be fast and handle extremely large
1615databases.")
1616 ;; The manual says: "It can be copied under the GNU General Public License
1617 ;; version 2 (GPLv2)."
1618 (license license:gpl2)))
1619
810cff85
RW		 1620(define-public clipper
1621 (package
1622 (name "clipper")
433530a5 1623 (version "1.1")
810cff85
RW		 1624 (source (origin
1625 (method url-fetch)
1626 (uri (string-append
1627 "https://github.com/YeoLab/clipper/archive/"
1628 version ".tar.gz"))
9ab5ea44 1629 (file-name (string-append name "-" version ".tar.gz"))
810cff85
RW		 1630 (sha256
1631 (base32
433530a5 1632 "0pflmsvhbf8izbgwhbhj1i7349sw1f55qpqj8ljmapp16hb0p0qi"))
810cff85
RW		 1633 (modules '((guix build utils)))
1634 (snippet
433530a5
RW		 1635 '(begin
1636 ;; remove unnecessary setup dependency
1637 (substitute* "setup.py"
1638 (("setup_requires = .*") ""))
1639 (for-each delete-file
1640 '("clipper/src/peaks.so"
1641 "clipper/src/readsToWiggle.so"))
1642 (delete-file-recursively "dist/")
1643 #t))))
810cff85
RW		 1644 (build-system python-build-system)
1645 (arguments `(#:python ,python-2)) ; only Python 2 is supported
1646 (inputs
92971d68 1647 `(("htseq" ,python2-htseq)
810cff85
RW		 1648 ("python-pybedtools" ,python2-pybedtools)
1649 ("python-cython" ,python2-cython)
1650 ("python-scikit-learn" ,python2-scikit-learn)
1651 ("python-matplotlib" ,python2-matplotlib)
433530a5 1652 ("python-pandas" ,python2-pandas)
810cff85
RW		 1653 ("python-pysam" ,python2-pysam)
1654 ("python-numpy" ,python2-numpy)
1655 ("python-scipy" ,python2-scipy)))
1656 (native-inputs
f3b98f4f 1657 `(("python-mock" ,python2-mock) ; for tests
d281be18 1658 ("python-nose" ,python2-nose) ; for tests
f3b98f4f 1659 ("python-pytz" ,python2-pytz))) ; for tests
810cff85
RW		 1660 (home-page "https://github.com/YeoLab/clipper")
1661 (synopsis "CLIP peak enrichment recognition")
1662 (description
1663 "CLIPper is a tool to define peaks in CLIP-seq datasets.")
1664 (license license:gpl2)))
1665
6a35566d
RS		 1666(define-public codingquarry
1667 (package
1668 (name "codingquarry")
1669 (version "2.0")
1670 (source (origin
1671 (method url-fetch)
1672 (uri (string-append
1673 "mirror://sourceforge/codingquarry/CodingQuarry_v"
1674 version ".tar.gz"))
1675 (sha256
1676 (base32
1677 "0115hkjflsnfzn36xppwf9h9avfxlavr43djqmshkkzbgjzsz60i"))))
1678 (build-system gnu-build-system)
1679 (arguments
1680 '(#:tests? #f ; no "check" target
1681 #:phases
1682 (modify-phases %standard-phases
1683 (delete 'configure)
1684 (replace 'install
1685 (lambda* (#:key outputs #:allow-other-keys)
1686 (let* ((out (assoc-ref outputs "out"))
1687 (bin (string-append out "/bin"))
1688 (doc (string-append out "/share/doc/codingquarry")))
1689 (install-file "INSTRUCTIONS.pdf" doc)
1690 (copy-recursively "QuarryFiles"
1691 (string-append out "/QuarryFiles"))
1692 (install-file "CodingQuarry" bin)
1693 (install-file "CufflinksGTF_to_CodingQuarryGFF3.py" bin)))))))
1694 (inputs `(("openmpi" ,openmpi)))
1695 (native-search-paths
1696 (list (search-path-specification
1697 (variable "QUARRY_PATH")
1698 (files '("QuarryFiles")))))
1699 (native-inputs `(("python" ,python-2))) ; Only Python 2 is supported
1700 (synopsis "Fungal gene predictor")
1701 (description "CodingQuarry is a highly accurate, self-training GHMM fungal
1702gene predictor designed to work with assembled, aligned RNA-seq transcripts.")
1703 (home-page "https://sourceforge.net/projects/codingquarry/")
1704 (license license:gpl3+)))
1705
36742f43
RW		 1706(define-public couger
1707 (package
1708 (name "couger")
1709 (version "1.8.2")
1710 (source (origin
1711 (method url-fetch)
1712 (uri (string-append
1713 "http://couger.oit.duke.edu/static/assets/COUGER"
1714 version ".zip"))
1715 (sha256
1716 (base32
1717 "04p2b14nmhzxw5h72mpzdhalv21bx4w9b87z0wpw0xzxpysyncmq"))))
1718 (build-system gnu-build-system)
1719 (arguments
1720 `(#:tests? #f
1721 #:phases
1722 (modify-phases %standard-phases
1723 (delete 'configure)
1724 (delete 'build)
1725 (replace
1726 'install
1727 (lambda* (#:key outputs #:allow-other-keys)
f3860753
TGR		 1728 (let* ((out (assoc-ref outputs "out"))
1729 (bin (string-append out "/bin")))
36742f43 1730 (copy-recursively "src" (string-append out "/src"))
f3860753 1731 (mkdir bin)
36742f43
RW		 1732 ;; Add "src" directory to module lookup path.
1733 (substitute* "couger"
1734 (("from argparse")
1735 (string-append "import sys\nsys.path.append(\""
1736 out "\")\nfrom argparse")))
f3860753 1737 (install-file "couger" bin))
36742f43
RW		 1738 #t))
1739 (add-after
1740 'install 'wrap-program
1741 (lambda* (#:key inputs outputs #:allow-other-keys)
1742 ;; Make sure 'couger' runs with the correct PYTHONPATH.
1743 (let* ((out (assoc-ref outputs "out"))
1744 (path (getenv "PYTHONPATH")))
1745 (wrap-program (string-append out "/bin/couger")
1746 `("PYTHONPATH" ":" prefix (,path))))
1747 #t)))))
1748 (inputs
1749 `(("python" ,python-2)
1750 ("python2-pillow" ,python2-pillow)
1751 ("python2-numpy" ,python2-numpy)
1752 ("python2-scipy" ,python2-scipy)
1753 ("python2-matplotlib" ,python2-matplotlib)))
1754 (propagated-inputs
2d7c4ae3 1755 `(("r-minimal" ,r-minimal)
36742f43
RW		 1756 ("libsvm" ,libsvm)
1757 ("randomjungle" ,randomjungle)))
1758 (native-inputs
1759 `(("unzip" ,unzip)))
1760 (home-page "http://couger.oit.duke.edu")
1761 (synopsis "Identify co-factors in sets of genomic regions")
1762 (description
1763 "COUGER can be applied to any two sets of genomic regions bound by
1764paralogous TFs (e.g., regions derived from ChIP-seq experiments) to identify
1765putative co-factors that provide specificity to each TF. The framework
1766determines the genomic targets uniquely-bound by each TF, and identifies a
1767small set of co-factors that best explain the in vivo binding differences
1768between the two TFs.
1769
1770COUGER uses classification algorithms (support vector machines and random
1771forests) with features that reflect the DNA binding specificities of putative
1772co-factors. The features are generated either from high-throughput TF-DNA
1773binding data (from protein binding microarray experiments), or from large
1774collections of DNA motifs.")
1775 (license license:gpl3+)))
1776
bfe3c685
RW		 1777(define-public clustal-omega
1778 (package
1779 (name "clustal-omega")
b3936f35 1780 (version "1.2.4")
bfe3c685
RW		 1781 (source (origin
1782 (method url-fetch)
b3936f35
RW		 1783 (uri (string-append "http://www.clustal.org/omega/clustal-omega-"
1784 version ".tar.gz"))
bfe3c685
RW		 1785 (sha256
1786 (base32
b3936f35 1787 "1vm30mzncwdv881vrcwg11vzvrsmwy4wg80j5i0lcfk6dlld50w6"))))
bfe3c685
RW		 1788 (build-system gnu-build-system)
1789 (inputs
1790 `(("argtable" ,argtable)))
1791 (home-page "http://www.clustal.org/omega/")
1792 (synopsis "Multiple sequence aligner for protein and DNA/RNA")
1793 (description
1794 "Clustal-Omega is a general purpose multiple sequence alignment (MSA)
1795program for protein and DNA/RNA. It produces high quality MSAs and is capable
1796of handling data-sets of hundreds of thousands of sequences in reasonable
1797time.")
1798 (license license:gpl2+)))
1799
191c7101
RW		 1800(define-public crossmap
1801 (package
1802 (name "crossmap")
61d5fd03 1803 (version "0.2.1")
191c7101
RW		 1804 (source (origin
1805 (method url-fetch)
1806 (uri (string-append "mirror://sourceforge/crossmap/CrossMap-"
1807 version ".tar.gz"))
1808 (sha256
1809 (base32
61d5fd03
RW		 1810 "07y179f63d7qnzdvkqcziwk9bs3k4zhp81q392fp1hwszjdvy22f"))
1811 ;; This patch has been sent upstream already and is available
1812 ;; for download from Sourceforge, but it has not been merged.
fc1adab1 1813 (patches (search-patches "crossmap-allow-system-pysam.patch"))
191c7101
RW		 1814 (modules '((guix build utils)))
1815 ;; remove bundled copy of pysam
1816 (snippet
1817 '(delete-file-recursively "lib/pysam"))))
1818 (build-system python-build-system)
1819 (arguments
1820 `(#:python ,python-2
1821 #:phases
dc1d3cde
KK		 1822 (modify-phases %standard-phases
1823 (add-after 'unpack 'set-env
1824 (lambda _ (setenv "CROSSMAP_USE_SYSTEM_PYSAM" "1") #t)))))
191c7101
RW		 1825 (inputs
1826 `(("python-numpy" ,python2-numpy)
1827 ("python-pysam" ,python2-pysam)
1828 ("zlib" ,zlib)))
1829 (native-inputs
1830 `(("python-cython" ,python2-cython)
f3b98f4f 1831 ("python-nose" ,python2-nose)))
191c7101
RW		 1832 (home-page "http://crossmap.sourceforge.net/")
1833 (synopsis "Convert genome coordinates between assemblies")
1834 (description
1835 "CrossMap is a program for conversion of genome coordinates or annotation
1836files between different genome assemblies. It supports most commonly used
1837file formats including SAM/BAM, Wiggle/BigWig, BED, GFF/GTF, VCF.")
1838 (license license:gpl2+)))
1839
8e913213
RW		 1840(define-public cutadapt
1841 (package
1842 (name "cutadapt")
1f94bff2 1843 (version "1.12")
8e913213
RW		 1844 (source (origin
1845 (method url-fetch)
1846 (uri (string-append
1847 "https://github.com/marcelm/cutadapt/archive/v"
1848 version ".tar.gz"))
1849 (file-name (string-append name "-" version ".tar.gz"))
1850 (sha256
1851 (base32
1f94bff2 1852 "19smhh6444ikn4jlmyhvffw4m5aw7yg07rqsk7arg8dkwyga1i4v"))))
8e913213
RW		 1853 (build-system python-build-system)
1854 (arguments
33d5b246
TGR		 1855 `(#:phases
1856 (modify-phases %standard-phases
1857 ;; The tests must be run after installation.
1858 (delete 'check)
1859 (add-after 'install 'check
1860 (lambda* (#:key inputs outputs #:allow-other-keys)
1861 (setenv "PYTHONPATH"
1862 (string-append
1863 (getenv "PYTHONPATH")
1864 ":" (assoc-ref outputs "out")
1865 "/lib/python"
1866 (string-take (string-take-right
1867 (assoc-ref inputs "python") 5) 3)
1868 "/site-packages"))
1869 (zero? (system* "nosetests" "-P" "tests")))))))
1f94bff2
TGR		 1870 (inputs
1871 `(("python-xopen" ,python-xopen)))
8e913213
RW		 1872 (native-inputs
1873 `(("python-cython" ,python-cython)
f3b98f4f 1874 ("python-nose" ,python-nose)))
0c6c9c00 1875 (home-page "https://cutadapt.readthedocs.io/en/stable/")
8e913213
RW		 1876 (synopsis "Remove adapter sequences from nucleotide sequencing reads")
1877 (description
1878 "Cutadapt finds and removes adapter sequences, primers, poly-A tails and
1879other types of unwanted sequence from high-throughput sequencing reads.")
1880 (license license:expat)))
1881
1baee943
RW		 1882(define-public libbigwig
1883 (package
1884 (name "libbigwig")
1885 (version "0.1.4")
1886 (source (origin
1887 (method url-fetch)
1888 (uri (string-append "https://github.com/dpryan79/libBigWig/"
1889 "archive/" version ".tar.gz"))
1890 (file-name (string-append name "-" version ".tar.gz"))
1891 (sha256
1892 (base32
1893 "098rjh35pi4a9q83n8wiwvyzykjqj6l8q189p1xgfw4ghywdlvw1"))))
1894 (build-system gnu-build-system)
1895 (arguments
1896 `(#:test-target "test"
1897 #:make-flags
1898 (list "CC=gcc"
1899 (string-append "prefix=" (assoc-ref %outputs "out")))
1900 #:phases
1901 (modify-phases %standard-phases
1902 (delete 'configure)
1903 (add-before 'check 'disable-curl-test
1904 (lambda _
1905 (substitute* "Makefile"
1906 (("./test/testRemote.*") ""))
1907 #t))
1908 ;; This has been fixed with the upstream commit 4ff6959cd8a0, but
1909 ;; there has not yet been a release containing this change.
1910 (add-before 'install 'create-target-dirs
1911 (lambda* (#:key outputs #:allow-other-keys)
1912 (let ((out (assoc-ref outputs "out")))
1913 (mkdir-p (string-append out "/lib"))
1914 (mkdir-p (string-append out "/include"))
1915 #t))))))
1916 (inputs
1917 `(("zlib" ,zlib)
1918 ("curl" ,curl)))
1919 (native-inputs
1920 `(("doxygen" ,doxygen)))
1921 (home-page "https://github.com/dpryan79/libBigWig")
1922 (synopsis "C library for handling bigWig files")
1923 (description
1924 "This package provides a C library for parsing local and remote BigWig
1925files.")
1926 (license license:expat)))
1927
69e0e03c
RW		 1928(define-public python-pybigwig
1929 (package
1930 (name "python-pybigwig")
1931 (version "0.2.5")
1932 (source (origin
1933 (method url-fetch)
1934 (uri (pypi-uri "pyBigWig" version))
1935 (sha256
1936 (base32
1937 "0yrpdxg3y0sny25x4w22lv1k47jzccqjmg7j4bp0hywklvp0hg7d"))
1938 (modules '((guix build utils)))
1939 (snippet
1940 '(begin
1941 ;; Delete bundled libBigWig sources
1942 (delete-file-recursively "libBigWig")))))
1943 (build-system python-build-system)
1944 (arguments
1945 `(#:phases
1946 (modify-phases %standard-phases
1947 (add-after 'unpack 'link-with-libBigWig
1948 (lambda* (#:key inputs #:allow-other-keys)
1949 (substitute* "setup.py"
1950 (("libs=\\[") "libs=[\"BigWig\", "))
1951 #t)))))
1952 (inputs
1953 `(("libbigwig" ,libbigwig)
1954 ("zlib" ,zlib)
1955 ("curl" ,curl)))
1956 (home-page "https://github.com/dpryan79/pyBigWig")
1957 (synopsis "Access bigWig files in Python using libBigWig")
1958 (description
1959 "This package provides Python bindings to the libBigWig library for
1960accessing bigWig files.")
1961 (license license:expat)))
1962
1963(define-public python2-pybigwig
5c31f4aa 1964 (package-with-python2 python-pybigwig))
69e0e03c 1965
ec2a67de
BW		 1966(define-public python-dendropy
1967 (package
1968 (name "python-dendropy")
25d84d31 1969 (version "4.2.0")
ec2a67de
BW		 1970 (source
1971 (origin
1972 (method url-fetch)
1973 (uri (pypi-uri "DendroPy" version))
1974 (sha256
1975 (base32
1885bb0c
RW		 1976 "15c7s3d5gf19ljsxvq5advaa752wfi7pwrdjyhzmg85hccyvp47p"))
1977 (patches (search-patches "python-dendropy-fix-tests.patch"))))
ec2a67de
BW		 1978 (build-system python-build-system)
1979 (home-page "http://packages.python.org/DendroPy/")
1980 (synopsis "Library for phylogenetics and phylogenetic computing")
1981 (description
1982 "DendroPy is a library for phylogenetics and phylogenetic computing: reading,
1983writing, simulation, processing and manipulation of phylogenetic
1984trees (phylogenies) and characters.")
1985 (license license:bsd-3)
1986 (properties `((python2-variant . ,(delay python2-dendropy))))))
1987
1988(define-public python2-dendropy
1989 (let ((base (package-with-python2 (strip-python2-variant python-dendropy))))
1990 (package
1991 (inherit base)
9602e3cc
BW		 1992 (arguments
1993 `(#:python ,python-2
1994 #:phases
1995 (modify-phases %standard-phases
1996 (replace 'check
1997 ;; There is currently a test failure that only happens on some
1998 ;; systems, and only using "setup.py test"
1999 (lambda _ (zero? (system* "nosetests")))))))
f3b98f4f 2000 (native-inputs `(("python2-nose" ,python2-nose)
ec2a67de
BW		 2001 ,@(package-native-inputs base))))))
2002
eb2200f3
RW		 2003(define-public python-py2bit
2004 (package
2005 (name "python-py2bit")
2006 (version "0.2.1")
2007 (source
2008 (origin
2009 (method url-fetch)
2010 (uri (pypi-uri "py2bit" version))
2011 (sha256
2012 (base32
2013 "1cdf4qlmgwsh1f4k0wdv2sr8x9qn4366p0k3614vbd0fpqiarxrl"))))
2014 (build-system python-build-system)
2015 (home-page "https://github.com/dpryan79/py2bit")
2016 (synopsis "Access 2bit files using lib2bit")
2017 (description
2018 "This package provides Python bindings for lib2bit to access 2bit files
2019with Python.")
2020 (license license:expat)))
ec2a67de 2021
1921b1de
RW		 2022(define-public deeptools
2023 (package
2024 (name "deeptools")
fed72008 2025 (version "2.5.1")
1921b1de
RW		 2026 (source (origin
2027 (method url-fetch)
3acb8c85
RW		 2028 (uri (string-append "https://github.com/fidelram/deepTools/"
2029 "archive/" version ".tar.gz"))
1921b1de
RW		 2030 (file-name (string-append name "-" version ".tar.gz"))
2031 (sha256
2032 (base32
fed72008 2033 "1q8i12l2gvk4n2s8lhyzwhh9g4qbc8lrk5l7maz00yvd5g6z5540"))))
1921b1de 2034 (build-system python-build-system)
14bda1ff 2035 (inputs
fed72008
RW		 2036 `(("python-scipy" ,python-scipy)
2037 ("python-numpy" ,python-numpy)
2038 ("python-numpydoc" ,python-numpydoc)
2039 ("python-matplotlib" ,python-matplotlib)
2040 ("python-pysam" ,python-pysam)
2041 ("python-py2bit" ,python-py2bit)
2042 ("python-pybigwig" ,python-pybigwig)))
1921b1de 2043 (native-inputs
fed72008
RW		 2044 `(("python-mock" ,python-mock) ;for tests
2045 ("python-nose" ,python-nose) ;for tests
2046 ("python-pytz" ,python-pytz))) ;for tests
1921b1de
RW		 2047 (home-page "https://github.com/fidelram/deepTools")
2048 (synopsis "Tools for normalizing and visualizing deep-sequencing data")
2049 (description
2050 "DeepTools addresses the challenge of handling the large amounts of data
2051that are now routinely generated from DNA sequencing centers. To do so,
2052deepTools contains useful modules to process the mapped reads data to create
2053coverage files in standard bedGraph and bigWig file formats. By doing so,
2054deepTools allows the creation of normalized coverage files or the comparison
2055between two files (for example, treatment and control). Finally, using such
2056normalized and standardized files, multiple visualizations can be created to
2057identify enrichments with functional annotations of the genome.")
2058 (license license:gpl3+)))
2059
684bf7c7
BW		 2060(define-public diamond
2061 (package
2062 (name "diamond")
171ba248 2063 (version "0.9.12")
684bf7c7
BW		 2064 (source (origin
2065 (method url-fetch)
2066 (uri (string-append
2067 "https://github.com/bbuchfink/diamond/archive/v"
2068 version ".tar.gz"))
2069 (file-name (string-append name "-" version ".tar.gz"))
2070 (sha256
2071 (base32
171ba248 2072 "1zrnr59pqdg56wxfdx3xg4h4d8fa14cxq5kfmc3q88hxrj42rqgs"))))
122395f9 2073 (build-system cmake-build-system)
684bf7c7 2074 (arguments
7c544991
BW		 2075 '(#:tests? #f ; no "check" target
2076 #:phases
2077 (modify-phases %standard-phases
2078 (add-after 'unpack 'remove-native-compilation
2079 (lambda _
2080 (substitute* "CMakeLists.txt" (("-march=native") ""))
2081 #t)))))
684bf7c7 2082 (inputs
122395f9 2083 `(("zlib" ,zlib)))
684bf7c7
BW		 2084 (home-page "https://github.com/bbuchfink/diamond")
2085 (synopsis "Accelerated BLAST compatible local sequence aligner")
2086 (description
2087 "DIAMOND is a BLAST-compatible local aligner for mapping protein and
2088translated DNA query sequences against a protein reference database (BLASTP
2089and BLASTX alignment mode). The speedup over BLAST is up to 20,000 on short
2090reads at a typical sensitivity of 90-99% relative to BLAST depending on the
2091data and settings.")
ef81341f 2092 (license license:agpl3+)))
684bf7c7 2093
97b9da68
RW		 2094(define-public discrover
2095 (package
2096 (name "discrover")
2097 (version "1.6.0")
2098 (source
2099 (origin
2100 (method url-fetch)
2101 (uri (string-append "https://github.com/maaskola/discrover/archive/"
2102 version ".tar.gz"))
2103 (file-name (string-append name "-" version ".tar.gz"))
2104 (sha256
2105 (base32
2106 "0rah9ja4m0rl5mldd6vag9rwrivw1zrqxssfq8qx64m7961fp68k"))))
2107 (build-system cmake-build-system)
fa702e1a
RW		 2108 (arguments
2109 `(#:tests? #f ; there are no tests
2110 #:phases
2111 (modify-phases %standard-phases
2112 (add-after 'unpack 'add-missing-includes
2113 (lambda _
2114 (substitute* "src/executioninformation.hpp"
2115 (("#define EXECUTIONINFORMATION_HPP" line)
2116 (string-append line "\n#include <random>")))
2117 (substitute* "src/plasma/fasta.hpp"
2118 (("#define FASTA_HPP" line)
2119 (string-append line "\n#include <random>")))
2120 #t)))))
97b9da68
RW		 2121 (inputs
2122 `(("boost" ,boost)
2123 ("cairo" ,cairo)))
2124 (native-inputs
2125 `(("texlive" ,texlive)
2126 ("imagemagick" ,imagemagick)))
2127 (home-page "http://dorina.mdc-berlin.de/public/rajewsky/discrover/")
2128 (synopsis "Discover discriminative nucleotide sequence motifs")
2129 (description "Discrover is a motif discovery method to find binding sites
2130of nucleic acid binding proteins.")
2131 (license license:gpl3+)))
2132
6619f9c7
RW		 2133(define-public eigensoft
2134 (let ((revision "1")
2135 (commit "b14d1e202e21e532536ff8004f0419cd5e259dc7"))
2136 (package
2137 (name "eigensoft")
2138 (version (string-append "6.1.2-"
2139 revision "."
2140 (string-take commit 9)))
2141 (source
2142 (origin
2143 (method git-fetch)
2144 (uri (git-reference
2145 (url "https://github.com/DReichLab/EIG.git")
2146 (commit commit)))
2147 (file-name (string-append "eigensoft-" commit "-checkout"))
2148 (sha256
2149 (base32
2150 "0f5m6k2j5c16xc3xbywcs989xyc26ncy1zfzp9j9n55n9r4xcaiq"))
2151 (modules '((guix build utils)))
2152 ;; Remove pre-built binaries.
2153 (snippet '(begin
2154 (delete-file-recursively "bin")
2155 (mkdir "bin")
2156 #t))))
2157 (build-system gnu-build-system)
2158 (arguments
2159 `(#:tests? #f ; There are no tests.
2160 #:make-flags '("CC=gcc")
2161 #:phases
2162 (modify-phases %standard-phases
2163 ;; There is no configure phase, but the Makefile is in a
2164 ;; sub-directory.
2165 (replace 'configure
2166 (lambda _
2167 (chdir "src")
2168 ;; The link flags are incomplete.
2169 (substitute* "Makefile"
2170 (("-lgsl") "-lgsl -lm -llapack -llapacke -lpthread"))
2171 #t))
2172 ;; The provided install target only copies executables to
2173 ;; the "bin" directory in the build root.
2174 (add-after 'install 'actually-install
2175 (lambda* (#:key outputs #:allow-other-keys)
2176 (let* ((out (assoc-ref outputs "out"))
2177 (bin (string-append out "/bin")))
6619f9c7
RW		 2178 (for-each (lambda (file)
2179 (install-file file bin))
2180 (find-files "../bin" ".*"))
2181 #t))))))
2182 (inputs
2183 `(("gsl" ,gsl)
2184 ("lapack" ,lapack)
6619f9c7
RW		 2185 ("openblas" ,openblas)
2186 ("perl" ,perl)
2187 ("gfortran" ,gfortran "lib")))
2188 (home-page "https://github.com/DReichLab/EIG")
2189 (synopsis "Tools for population genetics")
2190 (description "The EIGENSOFT package provides tools for population
2191genetics and stratification correction. EIGENSOFT implements methods commonly
2192used in population genetics analyses such as PCA, computation of Tracy-Widom
2193statistics, and finding related individuals in structured populations. It
2194comes with a built-in plotting script and supports multiple file formats and
2195quantitative phenotypes.")
2196 ;; The license of the eigensoft tools is Expat, but since it's
2197 ;; linking with the GNU Scientific Library (GSL) the effective
2198 ;; license is the GPL.
2199 (license license:gpl3+))))
2200
365c8153
RW		 2201(define-public edirect
2202 (package
2203 (name "edirect")
83b84fa8 2204 (version "4.10")
365c8153
RW		 2205 (source (origin
2206 (method url-fetch)
83b84fa8
RW		 2207 (uri (string-append "ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/"
2208 "versions/2016-05-03/edirect.tar.gz"))
365c8153
RW		 2209 (sha256
2210 (base32
83b84fa8 2211 "15zsprak5yh8c1yrz4r1knmb5s8qcmdid4xdhkh3lqcv64l60hli"))))
365c8153
RW		 2212 (build-system perl-build-system)
2213 (arguments
2214 `(#:tests? #f ;no "check" target
2215 #:phases
2216 (modify-phases %standard-phases
2217 (delete 'configure)
2218 (delete 'build)
2219 (replace 'install
2220 (lambda* (#:key outputs #:allow-other-keys)
2221 (let ((target (string-append (assoc-ref outputs "out")
2222 "/bin")))
2223 (mkdir-p target)
f3860753 2224 (install-file "edirect.pl" target)
365c8153
RW		 2225 #t)))
2226 (add-after
2227 'install 'wrap-program
2228 (lambda* (#:key inputs outputs #:allow-other-keys)
2229 ;; Make sure 'edirect.pl' finds all perl inputs at runtime.
2230 (let* ((out (assoc-ref outputs "out"))
2231 (path (getenv "PERL5LIB")))
2232 (wrap-program (string-append out "/bin/edirect.pl")
2233 `("PERL5LIB" ":" prefix (,path)))))))))
2234 (inputs
2235 `(("perl-html-parser" ,perl-html-parser)
2236 ("perl-encode-locale" ,perl-encode-locale)
2237 ("perl-file-listing" ,perl-file-listing)
2238 ("perl-html-tagset" ,perl-html-tagset)
2239 ("perl-html-tree" ,perl-html-tree)
2240 ("perl-http-cookies" ,perl-http-cookies)
2241 ("perl-http-date" ,perl-http-date)
2242 ("perl-http-message" ,perl-http-message)
2243 ("perl-http-negotiate" ,perl-http-negotiate)
2244 ("perl-lwp-mediatypes" ,perl-lwp-mediatypes)
2245 ("perl-lwp-protocol-https" ,perl-lwp-protocol-https)
2246 ("perl-net-http" ,perl-net-http)
2247 ("perl-uri" ,perl-uri)
2248 ("perl-www-robotrules" ,perl-www-robotrules)
2249 ("perl" ,perl)))
3d51ec91 2250 (home-page "http://www.ncbi.nlm.nih.gov/books/NBK179288/")
365c8153
RW		 2251 (synopsis "Tools for accessing the NCBI's set of databases")
2252 (description
2253 "Entrez Direct (EDirect) is a method for accessing the National Center
2254for Biotechnology Information's (NCBI) set of interconnected
2255databases (publication, sequence, structure, gene, variation, expression,
2256etc.) from a terminal. Functions take search terms from command-line
2257arguments. Individual operations are combined to build multi-step queries.
2258Record retrieval and formatting normally complete the process.
2259
2260EDirect also provides an argument-driven function that simplifies the
2261extraction of data from document summaries or other results that are returned
2262in structured XML format. This can eliminate the need for writing custom
2263software to answer ad hoc questions.")
2264 (license license:public-domain)))
2265
b16728b0
BW		 2266(define-public exonerate
2267 (package
2268 (name "exonerate")
2269 (version "2.4.0")
2270 (source
2271 (origin
2272 (method url-fetch)
2273 (uri
2274 (string-append
2275 "http://ftp.ebi.ac.uk/pub/software/vertebrategenomics/exonerate/"
2276 "exonerate-" version ".tar.gz"))
2277 (sha256
2278 (base32
2279 "0hj0m9xygiqsdxvbg79wq579kbrx1mdrabi2bzqz2zn9qwfjcjgq"))))
2280 (build-system gnu-build-system)
2281 (arguments
2282 `(#:parallel-build? #f)) ; Building in parallel fails on some machines.
2283 (native-inputs
2284 `(("pkg-config" ,pkg-config)))
2285 (inputs
2286 `(("glib" ,glib)))
2287 (home-page
2288 "https://www.ebi.ac.uk/about/vertebrate-genomics/software/exonerate")
2289 (synopsis "Generic tool for biological sequence alignment")
2290 (description
2291 "Exonerate is a generic tool for pairwise sequence comparison. It allows
2292the alignment of sequences using a many alignment models, either exhaustive
2293dynamic programming or a variety of heuristics.")
2294 (license license:gpl3)))
2295
e4e5a4d8
RW		 2296(define-public express
2297 (package
2298 (name "express")
2299 (version "1.5.1")
2300 (source (origin
2301 (method url-fetch)
2302 (uri
2303 (string-append
2304 "http://bio.math.berkeley.edu/eXpress/downloads/express-"
2305 version "/express-" version "-src.tgz"))
2306 (sha256
2307 (base32
2308 "03rczxd0gjp2l1jxcmjfmf5j94j77zqyxa6x063zsc585nj40n0c"))))
2309 (build-system cmake-build-system)
2310 (arguments
2311 `(#:tests? #f ;no "check" target
2312 #:phases
dc1d3cde
KK		 2313 (modify-phases %standard-phases
2314 (add-after 'unpack 'use-shared-boost-libs-and-set-bamtools-paths
2315 (lambda* (#:key inputs #:allow-other-keys)
2316 (substitute* "CMakeLists.txt"
2317 (("set\\(Boost_USE_STATIC_LIBS ON\\)")
2318 "set(Boost_USE_STATIC_LIBS OFF)")
2319 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/bamtools/include")
2320 (string-append (assoc-ref inputs "bamtools") "/include/bamtools")))
2321 (substitute* "src/CMakeLists.txt"
2322 (("\\$\\{CMAKE_CURRENT_SOURCE_DIR\\}/\\.\\./bamtools/lib")
2323 (string-append (assoc-ref inputs "bamtools") "/lib/bamtools")))
2324 #t)))))
e4e5a4d8
RW		 2325 (inputs
2326 `(("boost" ,boost)
2327 ("bamtools" ,bamtools)
2328 ("protobuf" ,protobuf)
2329 ("zlib" ,zlib)))
2330 (home-page "http://bio.math.berkeley.edu/eXpress")
2331 (synopsis "Streaming quantification for high-throughput genomic sequencing")
2332 (description
2333 "eXpress is a streaming tool for quantifying the abundances of a set of
2334target sequences from sampled subsequences. Example applications include
2335transcript-level RNA-Seq quantification, allele-specific/haplotype expression
2336analysis (from RNA-Seq), transcription factor binding quantification in
2337ChIP-Seq, and analysis of metagenomic data.")
2338 (license license:artistic2.0)))
2339
f3674b1c
BW		 2340(define-public express-beta-diversity
2341 (package
2342 (name "express-beta-diversity")
2343 (version "1.0.7")
2344 (source (origin
2345 (method url-fetch)
2346 (uri
2347 (string-append
2348 "https://github.com/dparks1134/ExpressBetaDiversity/archive/v"
2349 version ".tar.gz"))
2350 (file-name (string-append name "-" version ".tar.gz"))
2351 (sha256
2352 (base32
2353 "1djvdlmqvjf6h0zq7w36y8cl5cli6rgj86x65znl48agnwmzxfxr"))))
2354 (build-system gnu-build-system)
2355 (arguments
2356 `(#:phases
2357 (modify-phases %standard-phases
2358 (delete 'configure)
2359 (add-before 'build 'enter-source (lambda _ (chdir "source") #t))
2360 (replace 'check
2361 (lambda _ (zero? (system* "../bin/ExpressBetaDiversity"
2362 "-u"))))
2363 (add-after 'check 'exit-source (lambda _ (chdir "..") #t))
2364 (replace 'install
2365 (lambda* (#:key outputs #:allow-other-keys)
2366 (let ((bin (string-append (assoc-ref outputs "out")
2367 "/bin")))
2368 (mkdir-p bin)
f3860753
TGR		 2369 (install-file "scripts/convertToEBD.py" bin)
2370 (install-file "bin/ExpressBetaDiversity" bin)
f3674b1c
BW		 2371 #t))))))
2372 (inputs
2373 `(("python" ,python-2)))
2374 (home-page "http://kiwi.cs.dal.ca/Software/ExpressBetaDiversity")
2375 (synopsis "Taxon- and phylogenetic-based beta diversity measures")
2376 (description
2377 "Express Beta Diversity (EBD) calculates ecological beta diversity
2378(dissimilarity) measures between biological communities. EBD implements a
2379variety of diversity measures including those that make use of phylogenetic
2380similarity of community members.")
2381 (license license:gpl3+)))
2382
12b04cbe
BW		 2383(define-public fasttree
2384 (package
2385 (name "fasttree")
88682c9a 2386 (version "2.1.10")
12b04cbe
BW		 2387 (source (origin
2388 (method url-fetch)
2389 (uri (string-append
2390 "http://www.microbesonline.org/fasttree/FastTree-"
2391 version ".c"))
2392 (sha256
2393 (base32
88682c9a 2394 "0vcjdvy1j4m702vmak4svbfkrpcw63k7wymfksjp9a982zy8kjsl"))))
12b04cbe
BW		 2395 (build-system gnu-build-system)
2396 (arguments
2397 `(#:tests? #f ; no "check" target
2398 #:phases
2399 (modify-phases %standard-phases
2400 (delete 'unpack)
2401 (delete 'configure)
2402 (replace 'build
e03a5153
BW		 2403 (lambda* (#:key source #:allow-other-keys)
2404 (and (zero? (system* "gcc"
2405 "-O3"
2406 "-finline-functions"
2407 "-funroll-loops"
2408 "-Wall"
2409 "-o"
2410 "FastTree"
2411 source
2412 "-lm"))
2413 (zero? (system* "gcc"
2414 "-DOPENMP"
2415 "-fopenmp"
2416 "-O3"
2417 "-finline-functions"
2418 "-funroll-loops"
2419 "-Wall"
2420 "-o"
2421 "FastTreeMP"
2422 source
2423 "-lm")))))
12b04cbe 2424 (replace 'install
e03a5153
BW		 2425 (lambda* (#:key outputs #:allow-other-keys)
2426 (let ((bin (string-append (assoc-ref outputs "out")
2427 "/bin")))
2428 (mkdir-p bin)
f3860753
TGR		 2429 (install-file "FastTree" bin)
2430 (install-file "FastTreeMP" bin)
e03a5153 2431 #t))))))
12b04cbe
BW		 2432 (home-page "http://www.microbesonline.org/fasttree")
2433 (synopsis "Infers approximately-maximum-likelihood phylogenetic trees")
2434 (description
2435 "FastTree can handle alignments with up to a million of sequences in a
2436reasonable amount of time and memory. For large alignments, FastTree is
2437100-1,000 times faster than PhyML 3.0 or RAxML 7.")
2438 (license license:gpl2+)))
2439
2127cedb
RW		 2440(define-public fastx-toolkit
2441 (package
2442 (name "fastx-toolkit")
2443 (version "0.0.14")
2444 (source (origin
2445 (method url-fetch)
2446 (uri
2447 (string-append
2448 "https://github.com/agordon/fastx_toolkit/releases/download/"
2449 version "/fastx_toolkit-" version ".tar.bz2"))
2450 (sha256
2451 (base32
2452 "01jqzw386873sr0pjp1wr4rn8fsga2vxs1qfmicvx1pjr72007wy"))))
2453 (build-system gnu-build-system)
2454 (inputs
2455 `(("libgtextutils" ,libgtextutils)))
2456 (native-inputs
2457 `(("pkg-config" ,pkg-config)))
2458 (home-page "http://hannonlab.cshl.edu/fastx_toolkit/")
2459 (synopsis "Tools for FASTA/FASTQ file preprocessing")
2460 (description
2461 "The FASTX-Toolkit is a collection of command line tools for Short-Reads
2462FASTA/FASTQ files preprocessing.
2463
2464Next-Generation sequencing machines usually produce FASTA or FASTQ files,
2465containing multiple short-reads sequences. The main processing of such
2466FASTA/FASTQ files is mapping the sequences to reference genomes. However, it
2467is sometimes more productive to preprocess the files before mapping the
2468sequences to the genome---manipulating the sequences to produce better mapping
2469results. The FASTX-Toolkit tools perform some of these preprocessing tasks.")
2470 (license license:agpl3+)))
2471
d7678942
RW		 2472(define-public flexbar
2473 (package
2474 (name "flexbar")
2475 (version "2.5")
2476 (source (origin
2477 (method url-fetch)
2478 (uri
2479 (string-append "mirror://sourceforge/flexbar/"
2480 version "/flexbar_v" version "_src.tgz"))
2481 (sha256
2482 (base32
2483 "13jaykc3y1x8y5nn9j8ljnb79s5y51kyxz46hdmvvjj6qhyympmf"))))
2484 (build-system cmake-build-system)
2485 (arguments
4ca009c0 2486 `(#:configure-flags (list
d7678942
RW		 2487 (string-append "-DFLEXBAR_BINARY_DIR="
2488 (assoc-ref %outputs "out")
2489 "/bin/"))
2490 #:phases
dc1d3cde
KK		 2491 (modify-phases %standard-phases
2492 (replace 'check
2493 (lambda* (#:key outputs #:allow-other-keys)
2494 (setenv "PATH" (string-append
2495 (assoc-ref outputs "out") "/bin:"
2496 (getenv "PATH")))
2497 (chdir "../flexbar_v2.5_src/test")
2498 (zero? (system* "bash" "flexbar_validate.sh"))))
2499 (delete 'install))))
d7678942
RW		 2500 (inputs
2501 `(("tbb" ,tbb)
2502 ("zlib" ,zlib)))
2503 (native-inputs
2504 `(("pkg-config" ,pkg-config)
2505 ("seqan" ,seqan)))
2506 (home-page "http://flexbar.sourceforge.net")
2507 (synopsis "Barcode and adapter removal tool for sequencing platforms")
2508 (description
2509 "Flexbar preprocesses high-throughput nucleotide sequencing data
2510efficiently. It demultiplexes barcoded runs and removes adapter sequences.
2511Moreover, trimming and filtering features are provided. Flexbar increases
2512read mapping rates and improves genome and transcriptome assemblies. It
2513supports next-generation sequencing data in fasta/q and csfasta/q format from
2514Illumina, Roche 454, and the SOLiD platform.")
2515 (license license:gpl3)))
2516
19f4554c
BW		 2517(define-public fraggenescan
2518 (package
2519 (name "fraggenescan")
74297231 2520 (version "1.30")
19f4554c
BW		 2521 (source
2522 (origin
2523 (method url-fetch)
2524 (uri
2525 (string-append "mirror://sourceforge/fraggenescan/"
2526 "FragGeneScan" version ".tar.gz"))
2527 (sha256
74297231 2528 (base32 "158dcnwczgcyhwm4qlx19sanrwgdpzf6bn2y57mbpx55lkgz1mzj"))))
19f4554c
BW		 2529 (build-system gnu-build-system)
2530 (arguments
2531 `(#:phases
2532 (modify-phases %standard-phases
2533 (delete 'configure)
2534 (add-before 'build 'patch-paths
2535 (lambda* (#:key outputs #:allow-other-keys)
2536 (let* ((out (string-append (assoc-ref outputs "out")))
2537 (share (string-append out "/share/fraggenescan/")))
2538 (substitute* "run_FragGeneScan.pl"
2539 (("system\\(\"rm")
2540 (string-append "system(\"" (which "rm")))
2541 (("system\\(\"mv")
2542 (string-append "system(\"" (which "mv")))
74297231 2543 (("\\\"awk") (string-append "\"" (which "awk")))
19f4554c
BW		 2544 ;; This script and other programs expect the training files
2545 ;; to be in the non-standard location bin/train/XXX. Change
2546 ;; this to be share/fraggenescan/train/XXX instead.
2547 (("^\\$train.file = \\$dir.*")
2548 (string-append "$train_file = \""
2549 share
2550 "train/\".$FGS_train_file;")))
2551 (substitute* "run_hmm.c"
2552 (("^ strcat\\(train_dir, \\\"train/\\\"\\);")
74297231 2553 (string-append " strcpy(train_dir, \"" share "/train/\");"))))
19f4554c
BW		 2554 #t))
2555 (replace 'build
2556 (lambda _ (and (zero? (system* "make" "clean"))
2557 (zero? (system* "make" "fgs")))))
2558 (replace 'install
2559 (lambda* (#:key outputs #:allow-other-keys)
2560 (let* ((out (string-append (assoc-ref outputs "out")))
2561 (bin (string-append out "/bin/"))
2562 (share (string-append out "/share/fraggenescan/train")))
2563 (install-file "run_FragGeneScan.pl" bin)
2564 (install-file "FragGeneScan" bin)
19f4554c
BW		 2565 (copy-recursively "train" share))))
2566 (delete 'check)
2567 (add-after 'install 'post-install-check
2568 ;; In lieu of 'make check', run one of the examples and check the
2569 ;; output files gets created.
2570 (lambda* (#:key outputs #:allow-other-keys)
2571 (let* ((out (string-append (assoc-ref outputs "out")))
74297231
BW		 2572 (bin (string-append out "/bin/"))
2573 (frag (string-append bin "run_FragGeneScan.pl")))
2574 (and (zero? (system* frag ; Test complete genome.
19f4554c
BW		 2575 "-genome=./example/NC_000913.fna"
2576 "-out=./test2"
2577 "-complete=1"
2578 "-train=complete"))
2579 (file-exists? "test2.faa")
2580 (file-exists? "test2.ffn")
2581 (file-exists? "test2.gff")
74297231
BW		 2582 (file-exists? "test2.out")
2583 (zero? (system* ; Test incomplete sequences.
2584 frag
2585 "-genome=./example/NC_000913-fgs.ffn"
2586 "-out=out"
2587 "-complete=0"
2588 "-train=454_30")))))))))
19f4554c
BW		 2589 (inputs
2590 `(("perl" ,perl)
2591 ("python" ,python-2))) ;not compatible with python 3.
2592 (home-page "https://sourceforge.net/projects/fraggenescan/")
2593 (synopsis "Finds potentially fragmented genes in short reads")
2594 (description
2595 "FragGeneScan is a program for predicting bacterial and archaeal genes in
2596short and error-prone DNA sequencing reads. It can also be applied to predict
2597genes in incomplete assemblies or complete genomes.")
2598 ;; GPL3+ according to private correspondense with the authors.
2599 (license license:gpl3+)))
2600
81f3e0c1
BW		 2601(define-public fxtract
2602 (let ((util-commit "776ca85a18a47492af3794745efcb4a905113115"))
2603 (package
2604 (name "fxtract")
2605 (version "2.3")
2606 (source
2607 (origin
2608 (method url-fetch)
2609 (uri (string-append
2610 "https://github.com/ctSkennerton/fxtract/archive/"
2611 version ".tar.gz"))
2612 (file-name (string-append "ctstennerton-util-"
2613 (string-take util-commit 7)
2614 "-checkout"))
2615 (sha256
2616 (base32
2617 "0275cfdhis8517hm01is62062swmi06fxzifq7mr3knbbxjlaiwj"))))
2618 (build-system gnu-build-system)
2619 (arguments
2620 `(#:make-flags (list
2621 (string-append "PREFIX=" (assoc-ref %outputs "out"))
2622 "CC=gcc")
2623 #:test-target "fxtract_test"
2624 #:phases
2625 (modify-phases %standard-phases
2626 (delete 'configure)
2627 (add-before 'build 'copy-util
2628 (lambda* (#:key inputs #:allow-other-keys)
2629 (rmdir "util")
2630 (copy-recursively (assoc-ref inputs "ctskennerton-util") "util")
2631 #t))
2632 ;; Do not use make install as this requires additional dependencies.
2633 (replace 'install
2634 (lambda* (#:key outputs #:allow-other-keys)
2635 (let* ((out (assoc-ref outputs "out"))
2636 (bin (string-append out"/bin")))
2637 (install-file "fxtract" bin)
2638 #t))))))
2639 (inputs
2640 `(("pcre" ,pcre)
2641 ("zlib" ,zlib)))
2642 (native-inputs
2643 ;; ctskennerton-util is licensed under GPL2.
2644 `(("ctskennerton-util"
2645 ,(origin
2646 (method git-fetch)
2647 (uri (git-reference
2648 (url "https://github.com/ctSkennerton/util.git")
2649 (commit util-commit)))
2650 (file-name (string-append
2651 "ctstennerton-util-" util-commit "-checkout"))
2652 (sha256
2653 (base32
2654 "0cls1hd4vgj3f36fpzzg4xc77d6f3hpc60cbpfmn2gdr7ykzzad7"))))))
2655 (home-page "https://github.com/ctSkennerton/fxtract")
2656 (synopsis "Extract sequences from FASTA and FASTQ files")
2657 (description
2658 "Fxtract extracts sequences from a protein or nucleotide fastx (FASTA
2659or FASTQ) file given a subsequence. It uses a simple substring search for
2660basic tasks but can change to using POSIX regular expressions, PCRE, hash
2661lookups or multi-pattern searching as required. By default fxtract looks in
2662the sequence of each record but can also be told to look in the header,
2663comment or quality sections.")
afde1a26
BW		 2664 ;; 'util' requires SSE instructions.
2665 (supported-systems '("x86_64-linux"))
81f3e0c1
BW		 2666 (license license:expat))))
2667
2b18ad05
PP		 2668(define-public gemma
2669 (package
2670 (name "gemma")
2671 (version "0.96")
2672 (source (origin
2673 (method url-fetch)
2674 (uri (string-append "https://github.com/xiangzhou/GEMMA/archive/v"
2675 version ".tar.gz"))
2676 (file-name (string-append name "-" version ".tar.gz"))
2677 (sha256
2678 (base32
ce7e361f
EF		 2679 "055ynn16gd12pf78n4vr2a9jlwsbwzajpdnf2y2yilg1krfff222"))
2680 (patches (search-patches "gemma-intel-compat.patch"))))
2b18ad05
PP		 2681 (inputs
2682 `(("gsl" ,gsl)
2683 ("lapack" ,lapack)
2684 ("zlib" ,zlib)))
2685 (build-system gnu-build-system)
2686 (arguments
2c9232ae 2687 `(#:make-flags
ce7e361f
EF		 2688 '(,@(match (%current-system)
2689 ("x86_64-linux"
2690 '("FORCE_DYNAMIC=1"))
2691 ("i686-linux"
2692 '("FORCE_DYNAMIC=1" "FORCE_32BIT=1"))
2693 (_
2694 '("FORCE_DYNAMIC=1" "NO_INTEL_COMPAT=1"))))
2b18ad05
PP		 2695 #:phases
2696 (modify-phases %standard-phases
2697 (delete 'configure)
2698 (add-before 'build 'bin-mkdir
07bf6929
EF		 2699 (lambda _
2700 (mkdir-p "bin")
2701 #t))
2b18ad05 2702 (replace 'install
07bf6929
EF		 2703 (lambda* (#:key outputs #:allow-other-keys)
2704 (let ((out (assoc-ref outputs "out")))
2705 (install-file "bin/gemma"
2706 (string-append
2707 out "/bin")))
2708 #t)))
2b18ad05
PP		 2709 #:tests? #f)) ; no tests included yet
2710 (home-page "https://github.com/xiangzhou/GEMMA")
2711 (synopsis "Tool for genome-wide efficient mixed model association")
2712 (description
2713 "Genome-wide Efficient Mixed Model Association (GEMMA) provides a
2714standard linear mixed model resolver with application in genome-wide
2715association studies (GWAS).")
2716 (license license:gpl3)))
2717
5854f685
RW		 2718(define-public grit
2719 (package
2720 (name "grit")
2721 (version "2.0.2")
2722 (source (origin
2723 (method url-fetch)
2724 (uri (string-append
2725 "https://github.com/nboley/grit/archive/"
2726 version ".tar.gz"))
2727 (file-name (string-append name "-" version ".tar.gz"))
2728 (sha256
2729 (base32
2730 "157in84dj70wimbind3x7sy1whs3h57qfgcnj2s6lrd38fbrb7mj"))))
2731 (build-system python-build-system)
2732 (arguments
2733 `(#:python ,python-2
2734 #:phases
dc1d3cde
KK		 2735 (modify-phases %standard-phases
2736 (add-after 'unpack 'generate-from-cython-sources
2737 (lambda* (#:key inputs outputs #:allow-other-keys)
2738 ;; Delete these C files to force fresh generation from pyx sources.
2739 (delete-file "grit/sparsify_support_fns.c")
2740 (delete-file "grit/call_peaks_support_fns.c")
2741 (substitute* "setup.py"
2742 (("Cython.Setup") "Cython.Build")
2743 ;; Add numpy include path to fix compilation
2744 (("pyx\", \\]")
2745 (string-append "pyx\", ], include_dirs = ['"
2746 (assoc-ref inputs "python-numpy")
2747 "/lib/python2.7/site-packages/numpy/core/include/"
2748 "']")))
2749 #t)))))
5854f685
RW		 2750 (inputs
2751 `(("python-scipy" ,python2-scipy)
2752 ("python-numpy" ,python2-numpy)
2753 ("python-pysam" ,python2-pysam)
2754 ("python-networkx" ,python2-networkx)))
2755 (native-inputs
f3b98f4f 2756 `(("python-cython" ,python2-cython)))
5854f685
RW		 2757 (home-page "http://grit-bio.org")
2758 (synopsis "Tool for integrative analysis of RNA-seq type assays")
2759 (description
2760 "GRIT is designed to use RNA-seq, TES, and TSS data to build and quantify
2761full length transcript models. When none of these data sources are available,
2762GRIT can be run by providing a candidate set of TES or TSS sites. In
2763addition, GRIT can merge in reference junctions and gene boundaries. GRIT can
2764also be run in quantification mode, where it uses a provided GTF file and just
2765estimates transcript expression.")
2766 (license license:gpl3+)))
2767
346a829a
RW		 2768(define-public hisat
2769 (package
2770 (name "hisat")
2771 (version "0.1.4")
2772 (source (origin
2773 (method url-fetch)
2774 (uri (string-append
2775 "http://ccb.jhu.edu/software/hisat/downloads/hisat-"
2776 version "-beta-source.zip"))
2777 (sha256
2778 (base32
2779 "1k381ydranqxp09yf2y7w1d0chz5d59vb6jchi89hbb0prq19lk5"))))
2780 (build-system gnu-build-system)
2781 (arguments
e58d01fa
RW		 2782 `(#:tests? #f ;no check target
2783 #:make-flags '("allall"
2784 ;; Disable unsupported `popcnt' instructions on
2785 ;; architectures other than x86_64
2786 ,@(if (string-prefix? "x86_64"
2787 (or (%current-target-system)
2788 (%current-system)))
2789 '()
2790 '("POPCNT_CAPABILITY=0")))
346a829a 2791 #:phases
dc1d3cde
KK		 2792 (modify-phases %standard-phases
2793 (add-after 'unpack 'patch-sources
2794 (lambda _
2795 ;; XXX Cannot use snippet because zip files are not supported
2796 (substitute* "Makefile"
2797 (("^CC = .*$") "CC = gcc")
2798 (("^CPP = .*$") "CPP = g++")
2799 ;; replace BUILD_HOST and BUILD_TIME for deterministic build
2800 (("-DBUILD_HOST=.*") "-DBUILD_HOST=\"\\\"guix\\\"\"")
2801 (("-DBUILD_TIME=.*") "-DBUILD_TIME=\"\\\"0\\\"\""))
2802 (substitute* '("hisat-build" "hisat-inspect")
2803 (("/usr/bin/env") (which "env")))
2804 #t))
2805 (replace 'install
2806 (lambda* (#:key outputs #:allow-other-keys)
2807 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
2808 (for-each (lambda (file)
2809 (install-file file bin))
2810 (find-files
2811 "."
2812 "hisat(-(build|align|inspect)(-(s|l)(-debug)*)*)*$")))
2813 #t))
2814 (delete 'configure))))
346a829a
RW		 2815 (native-inputs
2816 `(("unzip" ,unzip)))
2817 (inputs
2818 `(("perl" ,perl)
2819 ("python" ,python)
2820 ("zlib" ,zlib)))
60af3d82
RW		 2821 ;; Non-portable SSE instructions are used so building fails on platforms
2822 ;; other than x86_64.
2823 (supported-systems '("x86_64-linux"))
346a829a
RW		 2824 (home-page "http://ccb.jhu.edu/software/hisat/index.shtml")
2825 (synopsis "Hierarchical indexing for spliced alignment of transcripts")
2826 (description
2827 "HISAT is a fast and sensitive spliced alignment program for mapping
2828RNA-seq reads. In addition to one global FM index that represents a whole
2829genome, HISAT uses a large set of small FM indexes that collectively cover the
2830whole genome. These small indexes (called local indexes) combined with
2831several alignment strategies enable effective alignment of RNA-seq reads, in
2832particular, reads spanning multiple exons.")
2833 (license license:gpl3+)))
2834
e84efc50
RW		 2835(define-public hisat2
2836 (package
2837 (name "hisat2")
2838 (version "2.0.5")
2839 (source
2840 (origin
2841 (method url-fetch)
2842 ;; FIXME: a better source URL is
2843 ;; (string-append "ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2"
2844 ;; "/downloads/hisat2-" version "-source.zip")
2845 ;; with hash "0lywnr8kijwsc2aw10dwxic0n0yvip6fl3rjlvc8zzwahamy4x7g"
2846 ;; but it is currently unavailable.
2847 (uri "https://github.com/infphilo/hisat2/archive/cba6e8cb.tar.gz")
2848 (file-name (string-append name "-" version ".tar.gz"))
2849 (sha256
2850 (base32
2851 "1mf2hdsyv7cd97xm9mp9a4qws02yrj95y6w6f6cdwnq0klp81r50"))))
2852 (build-system gnu-build-system)
2853 (arguments
2854 `(#:tests? #f ; no check target
2855 #:make-flags (list "CC=gcc" "CXX=g++" "allall")
2856 #:modules ((guix build gnu-build-system)
2857 (guix build utils)
2858 (srfi srfi-26))
2859 #:phases
2860 (modify-phases %standard-phases
2861 (add-after 'unpack 'make-deterministic
2862 (lambda _
2863 (substitute* "Makefile"
2864 (("`date`") "0"))
2865 #t))
2866 (delete 'configure)
2867 (replace 'install
2868 (lambda* (#:key outputs #:allow-other-keys)
2869 (let* ((out (assoc-ref outputs "out"))
2870 (bin (string-append out "/bin/"))
2871 (doc (string-append out "/share/doc/hisat2/")))
2872 (for-each
2873 (cut install-file <> bin)
2874 (find-files "."
2875 "hisat2(-(build|align|inspect)(-(s|l)(-debug)*)*)*$"))
2876 (mkdir-p doc)
2877 (install-file "doc/manual.inc.html" doc))
2878 #t)))))
2879 (native-inputs
2880 `(("unzip" ,unzip) ; needed for archive from ftp
2881 ("perl" ,perl)
2882 ("pandoc" ,ghc-pandoc))) ; for documentation
2883 (home-page "http://ccb.jhu.edu/software/hisat2/index.shtml")
2884 (synopsis "Graph-based alignment of genomic sequencing reads")
2885 (description "HISAT2 is a fast and sensitive alignment program for mapping
2886next-generation sequencing reads (both DNA and RNA) to a population of human
2887genomes (as well as to a single reference genome). In addition to using one
2888global @dfn{graph FM} (GFM) index that represents a population of human
2889genomes, HISAT2 uses a large set of small GFM indexes that collectively cover
2890the whole genome. These small indexes, combined with several alignment
2891strategies, enable rapid and accurate alignment of sequencing reads. This new
2892indexing scheme is called a @dfn{Hierarchical Graph FM index} (HGFM).")
2893 ;; HISAT2 contains files from Bowtie2, which is released under
2894 ;; GPLv2 or later. The HISAT2 source files are released under
2895 ;; GPLv3 or later.
2896 (license license:gpl3+)))
2897
c684629f
BW		 2898(define-public hmmer
2899 (package
2900 (name "hmmer")
2901 (version "3.1b2")
79f09fa2
BW		 2902 (source
2903 (origin
2904 (method url-fetch)
2905 (uri (string-append
2906 "http://eddylab.org/software/hmmer"
2907 (version-prefix version 1) "/"
2908 version "/hmmer-" version ".tar.gz"))
2909 (sha256
2910 (base32
2911 "0djmgc0pfli0jilfx8hql1axhwhqxqb8rxg2r5rg07aw73sfs5nx"))
2912 (patches (search-patches "hmmer-remove-cpu-specificity.patch"))))
c684629f 2913 (build-system gnu-build-system)
b3546174 2914 (native-inputs `(("perl" ,perl)))
a83e6046 2915 (home-page "http://hmmer.org/")
c684629f
BW		 2916 (synopsis "Biosequence analysis using profile hidden Markov models")
2917 (description
2918 "HMMER is used for searching sequence databases for homologs of protein
2919sequences, and for making protein sequence alignments. It implements methods
2920using probabilistic models called profile hidden Markov models (profile
2921HMMs).")
2922 (license (list license:gpl3+
2923 ;; The bundled library 'easel' is distributed
2924 ;; under The Janelia Farm Software License.
2925 (license:non-copyleft
2926 "file://easel/LICENSE"
2927 "See easel/LICENSE in the distribution.")))))
2928
85652f59
RW		 2929(define-public htseq
2930 (package
2931 (name "htseq")
92971d68 2932 (version "0.9.1")
85652f59
RW		 2933 (source (origin
2934 (method url-fetch)
75e6639f 2935 (uri (pypi-uri "HTSeq" version))
85652f59
RW		 2936 (sha256
2937 (base32
92971d68 2938 "11flgb1381xdhk43bzbfm3vhnszkpqg6jk76rpa5xd1zbrvvlnxg"))))
85652f59 2939 (build-system python-build-system)
92971d68
BW		 2940 (native-inputs
2941 `(("python-cython" ,python-cython)))
0536727e
RW		 2942 ;; Numpy needs to be propagated when htseq is used as a Python library.
2943 (propagated-inputs
92971d68 2944 `(("python-numpy" ,python-numpy)))
578b05d9 2945 (inputs
92971d68
BW		 2946 `(("python-pysam" ,python-pysam)
2947 ("python-matplotlib" ,python-matplotlib)))
85652f59
RW		 2948 (home-page "http://www-huber.embl.de/users/anders/HTSeq/")
2949 (synopsis "Analysing high-throughput sequencing data with Python")
2950 (description
2951 "HTSeq is a Python package that provides infrastructure to process data
2952from high-throughput sequencing assays.")
2953 (license license:gpl3+)))
2954
92971d68
BW		 2955(define-public python2-htseq
2956 (package-with-python2 htseq))
2957
1ad15c16 2958(define-public java-htsjdk
15a3c3d4 2959 (package
1ad15c16 2960 (name "java-htsjdk")
15a3c3d4
RW		 2961 (version "1.129")
2962 (source (origin
2963 (method url-fetch)
2964 (uri (string-append
2965 "https://github.com/samtools/htsjdk/archive/"
2966 version ".tar.gz"))
2967 (file-name (string-append name "-" version ".tar.gz"))
2968 (sha256
2969 (base32
2970 "0asdk9b8jx2ij7yd6apg9qx03li8q7z3ml0qy2r2qczkra79y6fw"))
2971 (modules '((guix build utils)))
2972 ;; remove build dependency on git
2973 (snippet '(substitute* "build.xml"
2974 (("failifexecutionfails=\"true\"")
2975 "failifexecutionfails=\"false\"")))))
10b4a969 2976 (build-system ant-build-system)
15a3c3d4 2977 (arguments
10b4a969
RW		 2978 `(#:tests? #f ; test require Internet access
2979 #:make-flags
2980 (list (string-append "-Ddist=" (assoc-ref %outputs "out")
2981 "/share/java/htsjdk/"))
2982 #:build-target "all"
2983 #:phases
2984 (modify-phases %standard-phases
2985 ;; The build phase also installs the jars
2986 (delete 'install))))
15a3c3d4
RW		 2987 (home-page "http://samtools.github.io/htsjdk/")
2988 (synopsis "Java API for high-throughput sequencing data (HTS) formats")
2989 (description
2990 "HTSJDK is an implementation of a unified Java library for accessing
2991common file formats, such as SAM and VCF, used for high-throughput
2992sequencing (HTS) data. There are also an number of useful utilities for
2993manipulating HTS data.")
2994 (license license:expat)))
2995
e7c09730
RW		 2996(define-public htslib
2997 (package
2998 (name "htslib")
4fb6d128 2999 (version "1.5")
e7c09730
RW		 3000 (source (origin
3001 (method url-fetch)
3002 (uri (string-append
3003 "https://github.com/samtools/htslib/releases/download/"
3004 version "/htslib-" version ".tar.bz2"))
3005 (sha256
3006 (base32
4fb6d128 3007 "0bcjmnbwp2bib1z1bkrp95w9v2syzdwdfqww10mkb1hxlmg52ax0"))))
e7c09730
RW		 3008 (build-system gnu-build-system)
3009 (arguments
3010 `(#:phases
3011 (modify-phases %standard-phases
3012 (add-after
3013 'unpack 'patch-tests
3014 (lambda _
3015 (substitute* "test/test.pl"
3016 (("/bin/bash") (which "bash")))
3017 #t)))))
3018 (inputs
a9e4a1e6
BW		 3019 `(("openssl" ,openssl)
3020 ("curl" ,curl)
3021 ("zlib" ,zlib)))
e7c09730
RW		 3022 (native-inputs
3023 `(("perl" ,perl)))
3024 (home-page "http://www.htslib.org")
3025 (synopsis "C library for reading/writing high-throughput sequencing data")
3026 (description
3027 "HTSlib is a C library for reading/writing high-throughput sequencing
3028data. It also provides the bgzip, htsfile, and tabix utilities.")
3029 ;; Files under cram/ are released under the modified BSD license;
3030 ;; the rest is released under the Expat license
3031 (license (list license:expat license:bsd-3))))
3032
bca2c576
BW		 3033;; This package should be removed once no packages rely upon it.
3034(define htslib-1.3
3035 (package
3036 (inherit htslib)
3037 (version "1.3.1")
3038 (source (origin
3039 (method url-fetch)
3040 (uri (string-append
3041 "https://github.com/samtools/htslib/releases/download/"
3042 version "/htslib-" version ".tar.bz2"))
3043 (sha256
3044 (base32
3045 "1rja282fwdc25ql6izkhdyh8ppw8x2fs0w0js78zgkmqjlikmma9"))))))
3046
c4325f62
RW		 3047(define-public idr
3048 (package
3049 (name "idr")
3050 (version "2.0.0")
3051 (source (origin
3052 (method url-fetch)
3053 (uri (string-append
3054 "https://github.com/nboley/idr/archive/"
3055 version ".tar.gz"))
3056 (file-name (string-append name "-" version ".tar.gz"))
3057 (sha256
3058 (base32
3059 "1k3x44biak00aiv3hpm1yd6nn4hhp7n0qnbs3zh2q9sw7qr1qj5r"))))
3060 (build-system python-build-system)
3061 (arguments
14386fc7 3062 `(#:tests? #f)) ; FIXME: "ImportError: No module named 'utility'"
b7a820fc 3063 (propagated-inputs
c4325f62 3064 `(("python-scipy" ,python-scipy)
b7a820fc 3065 ("python-sympy" ,python-sympy)
c4325f62
RW		 3066 ("python-numpy" ,python-numpy)
3067 ("python-matplotlib" ,python-matplotlib)))
3068 (native-inputs
f3b98f4f 3069 `(("python-cython" ,python-cython)))
c4325f62
RW		 3070 (home-page "https://github.com/nboley/idr")
3071 (synopsis "Tool to measure the irreproducible discovery rate (IDR)")
3072 (description
3073 "The IDR (Irreproducible Discovery Rate) framework is a unified approach
3074to measure the reproducibility of findings identified from replicate
3075experiments and provide highly stable thresholds based on reproducibility.")
3076 (license license:gpl3+)))
3077
43c565d2
RW		 3078(define-public jellyfish
3079 (package
3080 (name "jellyfish")
3081 (version "2.2.4")
3082 (source (origin
3083 (method url-fetch)
3084 (uri (string-append "https://github.com/gmarcais/Jellyfish/"
3085 "releases/download/v" version
3086 "/jellyfish-" version ".tar.gz"))
3087 (sha256
3088 (base32
3089 "0a6xnynqy2ibfbfz86b9g2m2dgm7f1469pmymkpam333gi3p26nk"))))
3090 (build-system gnu-build-system)
3091 (outputs '("out" ;for library
3092 "ruby" ;for Ruby bindings
3093 "python")) ;for Python bindings
3094 (arguments
3095 `(#:configure-flags
3096 (list (string-append "--enable-ruby-binding="
3097 (assoc-ref %outputs "ruby"))
3098 (string-append "--enable-python-binding="
3099 (assoc-ref %outputs "python")))
3100 #:phases
3101 (modify-phases %standard-phases
3102 (add-before 'check 'set-SHELL-variable
3103 (lambda _
3104 ;; generator_manager.hpp either uses /bin/sh or $SHELL
3105 ;; to run tests.
3106 (setenv "SHELL" (which "bash"))
3107 #t)))))
3108 (native-inputs
3109 `(("bc" ,bc)
3110 ("time" ,time)
3111 ("ruby" ,ruby)
3112 ("python" ,python-2)))
3113 (synopsis "Tool for fast counting of k-mers in DNA")
3114 (description
3115 "Jellyfish is a tool for fast, memory-efficient counting of k-mers in
3116DNA. A k-mer is a substring of length k, and counting the occurrences of all
3117such substrings is a central step in many analyses of DNA sequence. Jellyfish
3118is a command-line program that reads FASTA and multi-FASTA files containing
3119DNA sequences. It outputs its k-mer counts in a binary format, which can be
3120translated into a human-readable text format using the @code{jellyfish dump}
3121command, or queried for specific k-mers with @code{jellyfish query}.")
3122 (home-page "http://www.genome.umd.edu/jellyfish.html")
6e8faf77
EF		 3123 ;; From their website: JELLYFISH runs on 64-bit Intel-compatible processors
3124 (supported-systems '("x86_64-linux"))
43c565d2
RW		 3125 ;; The combined work is published under the GPLv3 or later. Individual
3126 ;; files such as lib/jsoncpp.cpp are released under the Expat license.
3127 (license (list license:gpl3+ license:expat))))
3128
94ff3157
BW		 3129(define-public khmer
3130 (package
3131 (name "khmer")
3132 (version "2.0")
3133 (source
3134 (origin
3135 (method url-fetch)
3136 (uri (pypi-uri "khmer" version))
3137 (sha256
3138 (base32
3139 "0wb05shqh77v00256qlm68vbbx3kl76fyzihszbz5nhanl4ni33a"))
3140 (patches (search-patches "khmer-use-libraries.patch"))))
3141 (build-system python-build-system)
3142 (arguments
3143 `(#:phases
3144 (modify-phases %standard-phases
3145 (add-after 'unpack 'set-paths
3146 (lambda* (#:key inputs outputs #:allow-other-keys)
3147 ;; Delete bundled libraries.
3148 (delete-file-recursively "third-party/zlib")
3149 (delete-file-recursively "third-party/bzip2")
3150 ;; Replace bundled seqan.
3151 (let* ((seqan-all "third-party/seqan")
3152 (seqan-include (string-append
3153 seqan-all "/core/include")))
3154 (delete-file-recursively seqan-all)
3155 (copy-recursively (string-append (assoc-ref inputs "seqan")
3156 "/include/seqan")
3157 (string-append seqan-include "/seqan")))
3158 ;; We do not replace the bundled MurmurHash as the canonical
3159 ;; repository for this code 'SMHasher' is unsuitable for
3160 ;; providing a library. See
3161 ;; https://lists.gnu.org/archive/html/guix-devel/2016-06/msg00977.html
3162 #t))
3163 (add-after 'unpack 'set-cc
3164 (lambda _
3165 (setenv "CC" "gcc")
3166 #t))
3167 ;; It is simpler to test after installation.
3168 (delete 'check)
3169 (add-after 'install 'post-install-check
3170 (lambda* (#:key inputs outputs #:allow-other-keys)
3171 (let ((out (assoc-ref outputs "out")))
3172 (setenv "PATH"
3173 (string-append
3174 (getenv "PATH")
3175 ":"
3176 (assoc-ref outputs "out")
3177 "/bin"))
3178 (setenv "PYTHONPATH"
3179 (string-append
3180 (getenv "PYTHONPATH")
3181 ":"
3182 out
3183 "/lib/python"
3184 (string-take (string-take-right
3185 (assoc-ref inputs "python") 5) 3)
3186 "/site-packages"))
3187 (with-directory-excursion "build"
3188 (zero? (system* "nosetests" "khmer" "--attr"
3189 "!known_failing")))))))))
3190 (native-inputs
3191 `(("seqan" ,seqan)
3192 ("python-nose" ,python-nose)))
3193 (inputs
3194 `(("zlib" ,zlib)
3195 ("bzip2" ,bzip2)
3196 ("python-screed" ,python-screed)
cf1d5814
BW		 3197 ("python-bz2file" ,python-bz2file)
3198 ;; Tests fail when gcc-5 is used for compilation. Use gcc-4.9 at least
3199 ;; until the next version of khmer (likely 2.1) is released.
3200 ("gcc" ,gcc-4.9)))
94ff3157
BW		 3201 (home-page "https://khmer.readthedocs.org/")
3202 (synopsis "K-mer counting, filtering and graph traversal library")
3203 (description "The khmer software is a set of command-line tools for
3204working with DNA shotgun sequencing data from genomes, transcriptomes,
3205metagenomes and single cells. Khmer can make de novo assemblies faster, and
3206sometimes better. Khmer can also identify and fix problems with shotgun
3207data.")
8157af2e
EF		 3208 ;; When building on i686, armhf and mips64el, we get the following error:
3209 ;; error: ['khmer', 'khmer.tests', 'oxli'] require 64-bit operating system
3210 (supported-systems '("x86_64-linux"))
94ff3157
BW		 3211 (license license:bsd-3)))
3212
b9a601d9
RJ		 3213(define-public kaiju
3214 (package
3215 (name "kaiju")
3216 (version "1.5.0")
3217 (source (origin
3218 (method url-fetch)
3219 (uri (string-append
3220 "https://github.com/bioinformatics-centre/kaiju/archive/v"
3221 version ".tar.gz"))
3222 (file-name (string-append name "-" version ".tar.gz"))
3223 (sha256
3224 (base32
3225 "0afbfalfw9y39bkwnqjrh9bghs118ws1pzj5h8l0nblgn3mbjdks"))))
3226 (build-system gnu-build-system)
3227 (arguments
3228 `(#:tests? #f ; There are no tests.
3229 #:phases
3230 (modify-phases %standard-phases
3231 (delete 'configure)
3232 (add-before 'build 'move-to-src-dir
3233 (lambda _ (chdir "src") #t))
3234 (replace 'install
3235 (lambda* (#:key inputs outputs #:allow-other-keys)
3236 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
3237 (mkdir-p bin)
3238 (chdir "..")
3239 (copy-recursively "bin" bin)
3240 (copy-recursively "util" bin))
3241 #t)))))
3242 (inputs
3243 `(("perl" ,perl)))
3244 (home-page "http://kaiju.binf.ku.dk/")
3245 (synopsis "Fast and sensitive taxonomic classification for metagenomics")
3246 (description "Kaiju is a program for sensitive taxonomic classification
3247of high-throughput sequencing reads from metagenomic whole genome sequencing
3248experiments.")
3249 (license license:gpl3+)))
3250
d57e6d0f
RW		 3251(define-public macs
3252 (package
3253 (name "macs")
0d0bcaa0 3254 (version "2.1.0.20151222")
d57e6d0f
RW		 3255 (source (origin
3256 (method url-fetch)
43ec07f1 3257 (uri (pypi-uri "MACS2" version))
d57e6d0f
RW		 3258 (sha256
3259 (base32
0d0bcaa0 3260 "1r2hcz6irhcq7lwbafjks98jbn34hv05avgbdjnp6w6mlfjkf8x5"))))
d57e6d0f
RW		 3261 (build-system python-build-system)
3262 (arguments
3263 `(#:python ,python-2 ; only compatible with Python 2.7
3264 #:tests? #f)) ; no test target
3265 (inputs
3266 `(("python-numpy" ,python2-numpy)))
7bf837fd 3267 (home-page "https://github.com/taoliu/MACS/")
d57e6d0f
RW		 3268 (synopsis "Model based analysis for ChIP-Seq data")
3269 (description
3270 "MACS is an implementation of a ChIP-Seq analysis algorithm for
3271identifying transcript factor binding sites named Model-based Analysis of
3272ChIP-Seq (MACS). MACS captures the influence of genome complexity to evaluate
3273the significance of enriched ChIP regions and it improves the spatial
3274resolution of binding sites through combining the information of both
3275sequencing tag position and orientation.")
3276 (license license:bsd-3)))
3277
41ddebdd
BW		 3278(define-public mafft
3279 (package
3280 (name "mafft")
88eb119f 3281 (version "7.310")
41ddebdd
BW		 3282 (source (origin
3283 (method url-fetch)
3284 (uri (string-append
3285 "http://mafft.cbrc.jp/alignment/software/mafft-" version
3286 "-without-extensions-src.tgz"))
3287 (file-name (string-append name "-" version ".tgz"))
3288 (sha256
3289 (base32
88eb119f 3290 "0gbsaz6z2qa307kd7wfb06c3y4ikmv1hsdvlns11f6zq4w1z9pwc"))))
41ddebdd
BW		 3291 (build-system gnu-build-system)
3292 (arguments
3293 `(#:tests? #f ; no automated tests, though there are tests in the read me
3294 #:make-flags (let ((out (assoc-ref %outputs "out")))
3295 (list (string-append "PREFIX=" out)
3296 (string-append "BINDIR="
3297 (string-append out "/bin"))))
3298 #:phases
3299 (modify-phases %standard-phases
3300 (add-after 'unpack 'enter-dir
101e8f71 3301 (lambda _ (chdir "core") #t))
41ddebdd 3302 (add-after 'enter-dir 'patch-makefile
101e8f71
BW		 3303 (lambda _
3304 ;; on advice from the MAFFT authors, there is no need to
3305 ;; distribute mafft-profile, mafft-distance, or
3306 ;; mafft-homologs.rb as they are too "specialised".
3307 (substitute* "Makefile"
3308 ;; remove mafft-homologs.rb from SCRIPTS
3309 (("^SCRIPTS = mafft mafft-homologs.rb")
3310 "SCRIPTS = mafft")
3311 ;; remove mafft-homologs from MANPAGES
3312 (("^MANPAGES = mafft.1 mafft-homologs.1")
3313 "MANPAGES = mafft.1")
3314 ;; remove mafft-distance from PROGS
3315 (("^PROGS = dvtditr dndfast7 dndblast sextet5 mafft-distance")
3316 "PROGS = dvtditr dndfast7 dndblast sextet5")
3317 ;; remove mafft-profile from PROGS
3318 (("splittbfast disttbfast tbfast mafft-profile 2cl mccaskillwrap")
3319 "splittbfast disttbfast tbfast f2cl mccaskillwrap")
3320 (("^rm -f mafft-profile mafft-profile.exe") "#")
3321 (("^rm -f mafft-distance mafft-distance.exe") ")#")
3322 ;; do not install MAN pages in libexec folder
3323 (("^\t\\$\\(INSTALL\\) -m 644 \\$\\(MANPAGES\\) \
41ddebdd 3324\\$\\(DESTDIR\\)\\$\\(LIBDIR\\)") "#"))
101e8f71 3325 #t))
02f35bb5
BW		 3326 (add-after 'enter-dir 'patch-paths
3327 (lambda* (#:key inputs #:allow-other-keys)
3328 (substitute* '("pairash.c"
3329 "mafft.tmpl")
3330 (("perl") (which "perl"))
3331 (("([\"`| ])awk" _ prefix)
3332 (string-append prefix (which "awk")))
3333 (("grep") (which "grep")))
3334 #t))
101e8f71
BW		 3335 (delete 'configure)
3336 (add-after 'install 'wrap-programs
3337 (lambda* (#:key outputs #:allow-other-keys)
3338 (let* ((out (assoc-ref outputs "out"))
3339 (bin (string-append out "/bin"))
3340 (path (string-append
3341 (assoc-ref %build-inputs "coreutils") "/bin:")))
3342 (for-each (lambda (file)
3343 (wrap-program file
3344 `("PATH" ":" prefix (,path))))
3345 (find-files bin)))
3346 #t)))))
41ddebdd 3347 (inputs
02f35bb5 3348 `(("perl" ,perl)
71461f88 3349 ("ruby" ,ruby)
02f35bb5 3350 ("gawk" ,gawk)
101e8f71
BW		 3351 ("grep" ,grep)
3352 ("coreutils" ,coreutils)))
41ddebdd
BW		 3353 (home-page "http://mafft.cbrc.jp/alignment/software/")
3354 (synopsis "Multiple sequence alignment program")
3355 (description
3356 "MAFFT offers a range of multiple alignment methods for nucleotide and
3357protein sequences. For instance, it offers L-INS-i (accurate; for alignment
3358of <~200 sequences) and FFT-NS-2 (fast; for alignment of <~30,000
3359sequences).")
3360 (license (license:non-copyleft
3361 "http://mafft.cbrc.jp/alignment/software/license.txt"
3362 "BSD-3 with different formatting"))))
8fd790eb 3363
84be3b99
MB		 3364(define-public mash
3365 (package
3366 (name "mash")
3367 (version "1.1.1")
3368 (source (origin
3369 (method url-fetch)
3370 (uri (string-append
3371 "https://github.com/marbl/mash/archive/v"
3372 version ".tar.gz"))
3373 (file-name (string-append name "-" version ".tar.gz"))
3374 (sha256
3375 (base32
3376 "08znbvqq5xknfhmpp3wcj574zvi4p7i8zifi67c9qw9a6ikp42fj"))
3377 (modules '((guix build utils)))
3378 (snippet
3379 ;; Delete bundled kseq.
3380 ;; TODO: Also delete bundled murmurhash and open bloom filter.
3381 '(delete-file "src/mash/kseq.h"))))
3382 (build-system gnu-build-system)
3383 (arguments
3384 `(#:tests? #f ; No tests.
3385 #:configure-flags
3386 (list
3387 (string-append "--with-capnp=" (assoc-ref %build-inputs "capnproto"))
3388 (string-append "--with-gsl=" (assoc-ref %build-inputs "gsl")))
3389 #:make-flags (list "CC=gcc")
3390 #:phases
3391 (modify-phases %standard-phases
3392 (add-after 'unpack 'fix-includes
3393 (lambda _
3394 (substitute* '("src/mash/Sketch.cpp" "src/mash/CommandFind.cpp")
3395 (("^#include \"kseq\\.h\"")
3396 "#include \"htslib/kseq.h\""))
3397 #t))
d10092b8 3398 (add-after 'fix-includes 'autoconf
84be3b99
MB		 3399 (lambda _ (zero? (system* "autoconf")))))))
3400 (native-inputs
3401 `(("autoconf" ,autoconf)
3402 ;; Capnproto and htslib are statically embedded in the final
3403 ;; application. Therefore we also list their licenses, below.
3404 ("capnproto" ,capnproto)
3405 ("htslib" ,htslib)))
3406 (inputs
3407 `(("gsl" ,gsl)
3408 ("zlib" ,zlib)))
3409 (supported-systems '("x86_64-linux"))
3410 (home-page "https://mash.readthedocs.io")
3411 (synopsis "Fast genome and metagenome distance estimation using MinHash")
3412 (description "Mash is a fast sequence distance estimator that uses the
3413MinHash algorithm and is designed to work with genomes and metagenomes in the
3414form of assemblies or reads.")
3415 (license (list license:bsd-3 ; Mash
3416 license:expat ; HTSlib and capnproto
3417 license:public-domain ; MurmurHash 3
3418 license:cpl1.0)))) ; Open Bloom Filter
3419
8fd790eb 3420(define-public metabat
f3f68a44
BW		 3421 (package
3422 (name "metabat")
25bd1fc1 3423 (version "2.12.1")
f3f68a44
BW		 3424 (source
3425 (origin
25bd1fc1
BW		 3426 (method url-fetch)
3427 (uri (string-append "https://bitbucket.org/berkeleylab/metabat/get/v"
3428 version ".tar.gz"))
3429 (file-name (string-append name "-" version ".tar.gz"))
f3f68a44
BW		 3430 (sha256
3431 (base32
25bd1fc1
BW		 3432 "1hmvdalz3zj5sqqklg0l4npjdv37cv2hsdi1al9iby2ndxjs1b73"))
3433 (patches (search-patches "metabat-fix-compilation.patch"))))
8fd790eb
BW		 3434 (build-system gnu-build-system)
3435 (arguments
3436 `(#:phases
3437 (modify-phases %standard-phases
3438 (add-after 'unpack 'fix-includes
45469ebe
BW		 3439 (lambda _
3440 (substitute* "src/BamUtils.h"
3441 (("^#include \"bam/bam\\.h\"")
3442 "#include \"samtools/bam.h\"")
3443 (("^#include \"bam/sam\\.h\"")
3444 "#include \"samtools/sam.h\""))
3445 (substitute* "src/KseqReader.h"
3446 (("^#include \"bam/kseq\\.h\"")
3447 "#include \"htslib/kseq.h\""))
3448 #t))
8fd790eb 3449 (add-after 'unpack 'fix-scons
f3f68a44
BW		 3450 (lambda* (#:key inputs #:allow-other-keys)
3451 (substitute* "SConstruct"
3452 (("^htslib_dir += 'samtools'")
3453 (string-append "htslib_dir = '"
3454 (assoc-ref inputs "htslib")
3455 "'"))
3456 (("^samtools_dir = 'samtools'")
3457 (string-append "samtools_dir = '"
3458 (assoc-ref inputs "samtools")
3459 "'"))
3460 (("^findStaticOrShared\\('bam', hts_lib")
3461 (string-append "findStaticOrShared('bam', '"
3462 (assoc-ref inputs "samtools")
3463 "/lib'"))
3464 ;; Do not distribute README.
3465 (("^env\\.Install\\(idir_prefix, 'README\\.md'\\)") ""))
3466 #t))
8fd790eb
BW		 3467 (delete 'configure)
3468 (replace 'build
f3f68a44
BW		 3469 (lambda* (#:key inputs outputs #:allow-other-keys)
3470 (mkdir (assoc-ref outputs "out"))
3471 (zero? (system* "scons"
3472 (string-append
3473 "PREFIX="
3474 (assoc-ref outputs "out"))
3475 (string-append
3476 "BOOST_ROOT="
3477 (assoc-ref inputs "boost"))
3478 "install"))))
45469ebe 3479 ;; Check and install are carried out during build phase.
8fd790eb
BW		 3480 (delete 'check)
3481 (delete 'install))))
3482 (inputs
3483 `(("zlib" ,zlib)
3484 ("perl" ,perl)
3485 ("samtools" ,samtools)
3486 ("htslib" ,htslib)
3487 ("boost" ,boost)))
3488 (native-inputs
3489 `(("scons" ,scons)))
3490 (home-page "https://bitbucket.org/berkeleylab/metabat")
3491 (synopsis
3492 "Reconstruction of single genomes from complex microbial communities")
3493 (description
3494 "Grouping large genomic fragments assembled from shotgun metagenomic
3495sequences to deconvolute complex microbial communities, or metagenome binning,
3496enables the study of individual organisms and their interactions. MetaBAT is
3497an automated metagenome binning software, which integrates empirical
3498probabilistic distances of genome abundance and tetranucleotide frequency.")
d931a4bb
EF		 3499 ;; The source code contains inline assembly.
3500 (supported-systems '("x86_64-linux" "i686-linux"))
f3f68a44
BW		 3501 (license (license:non-copyleft "file://license.txt"
3502 "See license.txt in the distribution."))))
8fd790eb 3503
318c0aee
MB		 3504(define-public minced
3505 (package
3506 (name "minced")
3507 (version "0.2.0")
3508 (source (origin
3509 (method url-fetch)
3510 (uri (string-append
3511 "https://github.com/ctSkennerton/minced/archive/"
3512 version ".tar.gz"))
3513 (file-name (string-append name "-" version ".tar.gz"))
3514 (sha256
3515 (base32
3516 "0wxmlsapxfpxfd3ps9636h7i2xy6la8i42mwh0j2lsky63h63jp1"))))
3517 (build-system gnu-build-system)
3518 (arguments
3519 `(#:test-target "test"
3520 #:phases
3521 (modify-phases %standard-phases
3522 (delete 'configure)
3523 (add-before 'check 'fix-test
3524 (lambda _
3525 ;; Fix test for latest version.
3526 (substitute* "t/Aquifex_aeolicus_VF5.expected"
3527 (("minced:0.1.6") "minced:0.2.0"))
3528 #t))
3529 (replace 'install ; No install target.
3530 (lambda* (#:key inputs outputs #:allow-other-keys)
3531 (let* ((out (assoc-ref outputs "out"))
3532 (bin (string-append out "/bin"))
3533 (wrapper (string-append bin "/minced")))
3534 ;; Minced comes with a wrapper script that tries to figure out where
3535 ;; it is located before running the JAR. Since these paths are known
3536 ;; to us, we build our own wrapper to avoid coreutils dependency.
3537 (install-file "minced.jar" bin)
3538 (with-output-to-file wrapper
3539 (lambda _
3540 (display
3541 (string-append
3542 "#!" (assoc-ref inputs "bash") "/bin/sh\n\n"
3543 (assoc-ref inputs "jre") "/bin/java -jar "
3544 bin "/minced.jar \"$@\"\n"))))
3545 (chmod wrapper #o555)))))))
3546 (native-inputs
3547 `(("jdk" ,icedtea "jdk")))
3548 (inputs
3549 `(("bash" ,bash)
3550 ("jre" ,icedtea "out")))
3551 (home-page "https://github.com/ctSkennerton/minced")
3552 (synopsis "Mining CRISPRs in Environmental Datasets")
3553 (description
3554 "MinCED is a program to find Clustered Regularly Interspaced Short
3555Palindromic Repeats (CRISPRs) in DNA sequences. It can be used for
3556unassembled metagenomic reads, but is mainly designed for full genomes and
3557assembled metagenomic sequence.")
3558 (license license:gpl3+)))
3559
ddd82e0e
RW		 3560(define-public miso
3561 (package
3562 (name "miso")
ce4dfde0 3563 (version "0.5.4")
ddd82e0e
RW		 3564 (source (origin
3565 (method url-fetch)
3566 (uri (string-append
86517de6 3567 "https://pypi.python.org/packages/source/m/misopy/misopy-"
ddd82e0e
RW		 3568 version ".tar.gz"))
3569 (sha256
3570 (base32
ce4dfde0 3571 "1z3x0vd8ma7pdrnywj7i3kgwl89sdkwrrn62zl7r5calqaq2hyip"))
ddd82e0e 3572 (modules '((guix build utils)))
ddd82e0e
RW		 3573 (snippet
3574 '(substitute* "setup.py"
0ec8b206
RW		 3575 ;; Use setuptools, or else the executables are not
3576 ;; installed.
3577 (("distutils.core") "setuptools")
3578 ;; use "gcc" instead of "cc" for compilation
ddd82e0e
RW		 3579 (("^defines")
3580 "cc.set_executables(
3581compiler='gcc',
3582compiler_so='gcc',
3583linker_exe='gcc',
3584linker_so='gcc -shared'); defines")))))
3585 (build-system python-build-system)
3586 (arguments
3587 `(#:python ,python-2 ; only Python 2 is supported
3588 #:tests? #f)) ; no "test" target
3589 (inputs
3590 `(("samtools" ,samtools)
3591 ("python-numpy" ,python2-numpy)
3592 ("python-pysam" ,python2-pysam)
3593 ("python-scipy" ,python2-scipy)
3594 ("python-matplotlib" ,python2-matplotlib)))
3595 (native-inputs
f3b98f4f
HG		 3596 `(("python-mock" ,python2-mock) ;for tests
3597 ("python-pytz" ,python2-pytz))) ;for tests
ddd82e0e
RW		 3598 (home-page "http://genes.mit.edu/burgelab/miso/index.html")
3599 (synopsis "Mixture of Isoforms model for RNA-Seq isoform quantitation")
3600 (description
3601 "MISO (Mixture-of-Isoforms) is a probabilistic framework that quantitates
3602the expression level of alternatively spliced genes from RNA-Seq data, and
3603identifies differentially regulated isoforms or exons across samples. By
3604modeling the generative process by which reads are produced from isoforms in
3605RNA-Seq, the MISO model uses Bayesian inference to compute the probability
3606that a read originated from a particular isoform.")
3607 (license license:gpl2)))
3608
324efb88
BW		 3609(define-public muscle
3610 (package
3611 (name "muscle")
3612 (version "3.8.1551")
3613 (source (origin
3614 (method url-fetch/tarbomb)
324efb88
BW		 3615 (uri (string-append
3616 "http://www.drive5.com/muscle/muscle_src_"
3617 version ".tar.gz"))
3618 (sha256
3619 (base32
3620 "0bj8kj7sdizy3987zx6w7axihk40fk8rn76mpbqqjcnd64i5a367"))))
3621 (build-system gnu-build-system)
3622 (arguments
3623 `(#:make-flags (list "LDLIBS = -lm")
3624 #:phases
3625 (modify-phases %standard-phases
3626 (delete 'configure)
3627 (replace 'check
3628 ;; There are no tests, so just test if it runs.
3629 (lambda _ (zero? (system* "./muscle" "-version"))))
3630 (replace 'install
3631 (lambda* (#:key outputs #:allow-other-keys)
3632 (let* ((out (assoc-ref outputs "out"))
3633 (bin (string-append out "/bin")))
3634 (install-file "muscle" bin)))))))
3635 (home-page "http://www.drive5.com/muscle")
3636 (synopsis "Multiple sequence alignment program")
3637 (description
3638 "MUSCLE aims to be a fast and accurate multiple sequence alignment
3639program for nucleotide and protein sequences.")
3640 ;; License information found in 'muscle -h' and usage.cpp.
3641 (license license:public-domain)))
3642
99268755
BW		 3643(define-public newick-utils
3644 ;; There are no recent releases so we package from git.
3645 (let ((commit "da121155a977197cab9fbb15953ca1b40b11eb87"))
3646 (package
3647 (name "newick-utils")
3648 (version (string-append "1.6-1." (string-take commit 8)))
3649 (source (origin
3650 (method git-fetch)
3651 (uri (git-reference
3652 (url "https://github.com/tjunier/newick_utils.git")
3653 (commit commit)))
3654 (file-name (string-append name "-" version "-checkout"))
3655 (sha256
3656 (base32
3657 "1hkw21rq1mwf7xp0rmbb2gqc0i6p11108m69i7mr7xcjl268pxnb"))))
3658 (build-system gnu-build-system)
3659 (arguments
3660 `(#:phases
3661 (modify-phases %standard-phases
3662 (add-after 'unpack 'autoconf
3663 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
3664 (inputs
3665 ;; XXX: TODO: Enable Lua and Guile bindings.
3666 ;; https://github.com/tjunier/newick_utils/issues/13
3667 `(("libxml2" ,libxml2)
3668 ("flex" ,flex)
3669 ("bison" ,bison)))
3670 (native-inputs
3671 `(("autoconf" ,autoconf)
3672 ("automake" ,automake)
3673 ("libtool" ,libtool)))
3674 (synopsis "Programs for working with newick format phylogenetic trees")
3675 (description
3676 "Newick-utils is a suite of utilities for processing phylogenetic trees
3677in Newick format. Functions include re-rooting, extracting subtrees,
3678trimming, pruning, condensing, drawing (ASCII graphics or SVG).")
3679 (home-page "https://github.com/tjunier/newick_utils")
3680 (license license:bsd-3))))
3681
1e44cf8b
BW		 3682(define-public orfm
3683 (package
3684 (name "orfm")
dfc83ead 3685 (version "0.7.1")
1e44cf8b
BW		 3686 (source (origin
3687 (method url-fetch)
3688 (uri (string-append
3689 "https://github.com/wwood/OrfM/releases/download/v"
3690 version "/orfm-" version ".tar.gz"))
3691 (sha256
3692 (base32
dfc83ead 3693 "16iigyr2gd8x0imzkk1dr3k5xsds9bpmwg31ayvjg0f4pir9rwqr"))))
1e44cf8b
BW		 3694 (build-system gnu-build-system)
3695 (inputs `(("zlib" ,zlib)))
6b6f7d6a
BW		 3696 (native-inputs
3697 `(("ruby-bio-commandeer" ,ruby-bio-commandeer)
3698 ("ruby-rspec" ,ruby-rspec)
3699 ("ruby" ,ruby)))
1e44cf8b
BW		 3700 (synopsis "Simple and not slow open reading frame (ORF) caller")
3701 (description
6b6f7d6a 3702 "An ORF caller finds stretches of DNA that, when translated, are not
1e44cf8b
BW		 3703interrupted by stop codons. OrfM finds and prints these ORFs.")
3704 (home-page "https://github.com/wwood/OrfM")
3705 (license license:lgpl3+)))
3706
c033f5d6
BW		 3707(define-public pplacer
3708 (let ((commit "g807f6f3"))
3709 (package
3710 (name "pplacer")
3711 ;; The commit should be updated with each version change.
3712 (version "1.1.alpha19")
3713 (source
3714 (origin
3715 (method url-fetch)
3716 (uri (string-append "https://github.com/matsen/pplacer/archive/v"
3717 version ".tar.gz"))
3718 (file-name (string-append name "-" version ".tar.gz"))
3719 (sha256
3720 (base32 "0z1lnd2s8sh6kpzg106wzbh2szw7h0hvq8syd5a6wv4rmyyz6x0f"))))
3721 (build-system ocaml-build-system)
3722 (arguments
3723 `(#:ocaml ,ocaml-4.01
3724 #:findlib ,ocaml4.01-findlib
3725 #:modules ((guix build ocaml-build-system)
3726 (guix build utils)
3727 (ice-9 ftw))
3728 #:phases
3729 (modify-phases %standard-phases
3730 (delete 'configure)
3731 (add-after 'unpack 'replace-bundled-cddlib
3732 (lambda* (#:key inputs #:allow-other-keys)
3733 (let* ((cddlib-src (assoc-ref inputs "cddlib-src"))
3734 (local-dir "cddlib_guix"))
3735 (mkdir local-dir)
3736 (with-directory-excursion local-dir
3737 (system* "tar" "xvf" cddlib-src))
3738 (let ((cddlib-src-folder
3739 (string-append local-dir "/"
3740 (list-ref (scandir local-dir) 2)
3741 "/lib-src")))
3742 (for-each
3743 (lambda (file)
3744 (copy-file file
3745 (string-append "cdd_src/" (basename file))))
3746 (find-files cddlib-src-folder ".*[ch]$")))
3747 #t)))
3748 (add-after 'unpack 'fix-makefile
3749 (lambda _
3750 ;; Remove system calls to 'git'.
3751 (substitute* "Makefile"
3752 (("^DESCRIPT:=pplacer-.*")
3753 (string-append
3754 "DESCRIPT:=pplacer-$(shell uname)-v" ,version "\n")))
3755 (substitute* "myocamlbuild.ml"
3756 (("git describe --tags --long .*\\\" with")
3757 (string-append
3758 "echo -n v" ,version "-" ,commit "\" with")))
3759 #t))
3760 (replace 'install
3761 (lambda* (#:key outputs #:allow-other-keys)
3762 (let* ((out (assoc-ref outputs "out"))
3763 (bin (string-append out "/bin")))
3764 (copy-recursively "bin" bin))
3765 #t)))))
3766 (native-inputs
3767 `(("zlib" ,zlib)
3768 ("gsl" ,gsl)
3769 ("ocaml-ounit" ,ocaml4.01-ounit)
3770 ("ocaml-batteries" ,ocaml4.01-batteries)
3771 ("ocaml-camlzip" ,ocaml4.01-camlzip)
3772 ("ocaml-csv" ,ocaml4.01-csv)
3773 ("ocaml-sqlite3" ,ocaml4.01-sqlite3)
3774 ("ocaml-xmlm" ,ocaml4.01-xmlm)
3775 ("ocaml-mcl" ,ocaml4.01-mcl)
3776 ("ocaml-gsl" ,ocaml4.01-gsl)
3777 ("cddlib-src" ,(package-source cddlib))))
3778 (propagated-inputs
3779 `(("pplacer-scripts" ,pplacer-scripts)))
3780 (synopsis "Phylogenetic placement of biological sequences")
3781 (description
3782 "Pplacer places query sequences on a fixed reference phylogenetic tree
3783to maximize phylogenetic likelihood or posterior probability according to a
3784reference alignment. Pplacer is designed to be fast, to give useful
3785information about uncertainty, and to offer advanced visualization and
3786downstream analysis.")
3787 (home-page "http://matsen.fhcrc.org/pplacer")
3788 (license license:gpl3))))
3789
3790;; This package is installed alongside 'pplacer'. It is a separate package so
3791;; that it can use the python-build-system for the scripts that are
3792;; distributed alongside the main OCaml binaries.
3793(define pplacer-scripts
3794 (package
3795 (inherit pplacer)
3796 (name "pplacer-scripts")
3797 (build-system python-build-system)
3798 (arguments
3799 `(#:python ,python-2
3800 #:phases
3801 (modify-phases %standard-phases
3802 (add-after 'unpack 'enter-scripts-dir
3803 (lambda _ (chdir "scripts")))
3804 (replace 'check
3805 (lambda _
3806 (zero? (system* "python" "-m" "unittest" "discover" "-v"))))
3807 (add-after 'install 'wrap-executables
3808 (lambda* (#:key inputs outputs #:allow-other-keys)
3809 (let* ((out (assoc-ref outputs "out"))
3810 (bin (string-append out "/bin")))
3811 (let ((path (string-append
3812 (assoc-ref inputs "hmmer") "/bin:"
3813 (assoc-ref inputs "infernal") "/bin")))
3814 (display path)
3815 (wrap-program (string-append bin "/refpkg_align.py")
3816 `("PATH" ":" prefix (,path))))
3817 (let ((path (string-append
3818 (assoc-ref inputs "hmmer") "/bin")))
3819 (wrap-program (string-append bin "/hrefpkg_query.py")
3820 `("PATH" ":" prefix (,path)))))
3821 #t)))))
3822 (inputs
3823 `(("infernal" ,infernal)
3824 ("hmmer" ,hmmer)))
3825 (propagated-inputs
3826 `(("python-biopython" ,python2-biopython)
3827 ("taxtastic" ,taxtastic)))
3828 (synopsis "Pplacer Python scripts")))
3829
19ee9201
RW		 3830(define-public python2-pbcore
3831 (package
3832 (name "python2-pbcore")
e301bfc8 3833 (version "1.2.10")
19ee9201
RW		 3834 (source (origin
3835 (method url-fetch)
ddb83129 3836 (uri (pypi-uri "pbcore" version))
19ee9201
RW		 3837 (sha256
3838 (base32
e301bfc8 3839 "1kjmv891d6qbpp4shhhvkl02ff4q5xlpnls2513sm2cjcrs52f1i"))))
19ee9201
RW		 3840 (build-system python-build-system)
3841 (arguments `(#:python ,python-2)) ; pbcore requires Python 2.7
de5bc890 3842 (propagated-inputs
19ee9201
RW		 3843 `(("python-cython" ,python2-cython)
3844 ("python-numpy" ,python2-numpy)
3845 ("python-pysam" ,python2-pysam)
3846 ("python-h5py" ,python2-h5py)))
3847 (native-inputs
de5bc890
HG		 3848 `(("python-nose" ,python2-nose)
3849 ("python-sphinx" ,python2-sphinx)
3850 ("python-pyxb" ,python2-pyxb)))
19ee9201
RW		 3851 (home-page "http://pacificbiosciences.github.io/pbcore/")
3852 (synopsis "Library for reading and writing PacBio data files")
3853 (description
3854 "The pbcore package provides Python APIs for interacting with PacBio data
3855files and writing bioinformatics applications.")
3856 (license license:bsd-3)))
3857
c61fe02c
RW		 3858(define-public python2-warpedlmm
3859 (package
3860 (name "python2-warpedlmm")
3861 (version "0.21")
3862 (source
3863 (origin
3864 (method url-fetch)
3865 (uri (string-append
3866 "https://pypi.python.org/packages/source/W/WarpedLMM/WarpedLMM-"
3867 version ".zip"))
3868 (sha256
3869 (base32
3870 "1agfz6zqa8nc6cw47yh0s3y14gkpa9wqazwcj7mwwj3ffnw39p3j"))))
3871 (build-system python-build-system)
3872 (arguments
54c85e12 3873 `(#:python ,python-2)) ; requires Python 2.7
c61fe02c
RW		 3874 (propagated-inputs
3875 `(("python-scipy" ,python2-scipy)
3876 ("python-numpy" ,python2-numpy)
3877 ("python-matplotlib" ,python2-matplotlib)
3878 ("python-fastlmm" ,python2-fastlmm)
3879 ("python-pandas" ,python2-pandas)
3880 ("python-pysnptools" ,python2-pysnptools)))
3881 (native-inputs
f3b98f4f 3882 `(("python-mock" ,python2-mock)
c61fe02c
RW		 3883 ("python-nose" ,python2-nose)
3884 ("unzip" ,unzip)))
3885 (home-page "https://github.com/PMBio/warpedLMM")
3886 (synopsis "Implementation of warped linear mixed models")
3887 (description
3888 "WarpedLMM is a Python implementation of the warped linear mixed model,
3889which automatically learns an optimal warping function (or transformation) for
3890the phenotype as it models the data.")
3891 (license license:asl2.0)))
3892
2c16316e 3893(define-public pbtranscript-tofu
698bd297 3894 (let ((commit "8f5467fe6a4472bcfb4226c8720993c8507adfe4"))
2c16316e
RW		 3895 (package
3896 (name "pbtranscript-tofu")
698bd297 3897 (version (string-append "2.2.3." (string-take commit 7)))
2c16316e
RW		 3898 (source (origin
3899 (method git-fetch)
3900 (uri (git-reference
3901 (url "https://github.com/PacificBiosciences/cDNA_primer.git")
3902 (commit commit)))
9a067efd 3903 (file-name (string-append name "-" version "-checkout"))
2c16316e
RW		 3904 (sha256
3905 (base32
9a067efd
RW		 3906 "1lgnpi35ihay42qx0b6yl3kkgra723i413j33kvs0kvs61h82w0f"))
3907 (modules '((guix build utils)))
3908 (snippet
3909 '(begin
3910 ;; remove bundled Cython sources
3911 (delete-file "pbtranscript-tofu/pbtranscript/Cython-0.20.1.tar.gz")
3912 #t))))
2c16316e
RW		 3913 (build-system python-build-system)
3914 (arguments
3915 `(#:python ,python-2
cdc2bb50
MB		 3916 ;; FIXME: Tests fail with "No such file or directory:
3917 ;; pbtools/pbtranscript/modified_bx_intervals/intersection_unique.so"
3918 #:tests? #f
2c16316e 3919 #:phases
9a067efd
RW		 3920 (modify-phases %standard-phases
3921 (add-after 'unpack 'enter-directory
3922 (lambda _
3923 (chdir "pbtranscript-tofu/pbtranscript/")
3924 #t))
3925 ;; With setuptools version 18.0 and later this setup.py hack causes
3926 ;; a build error, so we disable it.
3927 (add-after 'enter-directory 'patch-setuppy
3928 (lambda _
3929 (substitute* "setup.py"
3930 (("if 'setuptools.extension' in sys.modules:")
3931 "if False:"))
3932 #t)))))
2c16316e 3933 (inputs
9a067efd 3934 `(("python-numpy" ,python2-numpy)
2c16316e 3935 ("python-bx-python" ,python2-bx-python)
c5372108
RW		 3936 ("python-networkx" ,python2-networkx)
3937 ("python-scipy" ,python2-scipy)
9a067efd
RW		 3938 ("python-pbcore" ,python2-pbcore)
3939 ("python-h5py" ,python2-h5py)))
2c16316e 3940 (native-inputs
9a067efd 3941 `(("python-cython" ,python2-cython)
f3b98f4f 3942 ("python-nose" ,python2-nose)))
2c16316e
RW		 3943 (home-page "https://github.com/PacificBiosciences/cDNA_primer")
3944 (synopsis "Analyze transcriptome data generated with the Iso-Seq protocol")
3945 (description
3946 "pbtranscript-tofu contains scripts to analyze transcriptome data
3947generated using the PacBio Iso-Seq protocol.")
3948 (license license:bsd-3))))
3949
024130d2
BW		 3950(define-public prank
3951 (package
3952 (name "prank")
3953 (version "150803")
3954 (source (origin
3955 (method url-fetch)
3956 (uri (string-append
3957 "http://wasabiapp.org/download/prank/prank.source."
3958 version ".tgz"))
3959 (sha256
3960 (base32
3961 "0am4z94fs3w2n5xpfls9zda61vq7qqz4q2i7b9hlsxz5q4j3kfm4"))))
3962 (build-system gnu-build-system)
3963 (arguments
3964 `(#:phases
3965 (modify-phases %standard-phases
3966 (add-after 'unpack 'enter-src-dir
3967 (lambda _
3968 (chdir "src")
3969 #t))
62d00095
EF		 3970 (add-after 'unpack 'remove-m64-flag
3971 ;; Prank will build with the correct 'bit-ness' without this flag
3972 ;; and this allows building on 32-bit machines.
3973 (lambda _ (substitute* "src/Makefile"
3974 (("-m64") ""))
3975 #t))
024130d2
BW		 3976 (delete 'configure)
3977 (replace 'install
3978 (lambda* (#:key outputs #:allow-other-keys)
3979 (let* ((out (assoc-ref outputs "out"))
3980 (bin (string-append out "/bin"))
3981 (man (string-append out "/share/man/man1"))
3982 (path (string-append
3983 (assoc-ref %build-inputs "mafft") "/bin:"
3984 (assoc-ref %build-inputs "exonerate") "/bin:"
3985 (assoc-ref %build-inputs "bppsuite") "/bin")))
3986 (install-file "prank" bin)
3987 (wrap-program (string-append bin "/prank")
3988 `("PATH" ":" prefix (,path)))
3989 (install-file "prank.1" man))
3990 #t)))))
3991 (inputs
3992 `(("mafft" ,mafft)
3993 ("exonerate" ,exonerate)
3994 ("bppsuite" ,bppsuite)))
3995 (home-page "http://wasabiapp.org/software/prank/")
3996 (synopsis "Probabilistic multiple sequence alignment program")
3997 (description
3998 "PRANK is a probabilistic multiple sequence alignment program for DNA,
3999codon and amino-acid sequences. It is based on a novel algorithm that treats
4000insertions correctly and avoids over-estimation of the number of deletion
4001events. In addition, PRANK borrows ideas from maximum likelihood methods used
4002in phylogenetics and correctly takes into account the evolutionary distances
4003between sequences. Lastly, PRANK allows for defining a potential structure
4004for sequences to be aligned and then, simultaneously with the alignment,
4005predicts the locations of structural units in the sequences.")
4006 (license license:gpl2+)))
4007
31a9d653
BW		 4008(define-public proteinortho
4009 (package
4010 (name "proteinortho")
8c864901 4011 (version "5.16b")
31a9d653
BW		 4012 (source
4013 (origin
4014 (method url-fetch)
4015 (uri
4016 (string-append
4017 "http://www.bioinf.uni-leipzig.de/Software/proteinortho/proteinortho_v"
4018 version "_src.tar.gz"))
4019 (sha256
4020 (base32
8c864901 4021 "1wl0dawpssqwfjvr651r4wlww8hhjin8nba6xh71ks7sbypx886j"))))
31a9d653
BW		 4022 (build-system gnu-build-system)
4023 (arguments
4024 `(#:test-target "test"
4025 #:phases
4026 (modify-phases %standard-phases
4027 (replace 'configure
4028 ;; There is no configure script, so we modify the Makefile directly.
4029 (lambda* (#:key outputs #:allow-other-keys)
4030 (substitute* "Makefile"
4031 (("INSTALLDIR=.*")
4032 (string-append
4033 "INSTALLDIR=" (assoc-ref outputs "out") "/bin\n")))
4034 #t))
4035 (add-before 'install 'make-install-directory
4036 ;; The install directory is not created during 'make install'.
4037 (lambda* (#:key outputs #:allow-other-keys)
4038 (mkdir-p (string-append (assoc-ref outputs "out") "/bin"))
4039 #t))
4040 (add-after 'install 'wrap-programs
4041 (lambda* (#:key inputs outputs #:allow-other-keys)
4042 (let* ((path (getenv "PATH"))
4043 (out (assoc-ref outputs "out"))
4044 (binary (string-append out "/bin/proteinortho5.pl")))
4045 (wrap-program binary `("PATH" ":" prefix (,path))))
4046 #t)))))
4047 (inputs
4048 `(("perl" ,perl)
4049 ("python" ,python-2)
4050 ("blast+" ,blast+)))
4051 (home-page "http://www.bioinf.uni-leipzig.de/Software/proteinortho")
4052 (synopsis "Detect orthologous genes across species")
4053 (description
4054 "Proteinortho is a tool to detect orthologous genes across different
4055species. For doing so, it compares similarities of given gene sequences and
4056clusters them to find significant groups. The algorithm was designed to handle
4057large-scale data and can be applied to hundreds of species at once.")
4058 (license license:gpl2+)))
4059
846e3409
RW		 4060(define-public pyicoteo
4061 (package
4062 (name "pyicoteo")
4063 (version "2.0.7")
4064 (source
4065 (origin
4066 (method url-fetch)
4067 (uri (string-append "https://bitbucket.org/regulatorygenomicsupf/"
4068 "pyicoteo/get/v" version ".tar.bz2"))
4069 (file-name (string-append name "-" version ".tar.bz2"))
4070 (sha256
4071 (base32
4072 "0d6087f29xp8wxwlj111c3sylli98n0l8ry58c51ixzq0zfm50wa"))))
4073 (build-system python-build-system)
4074 (arguments
4075 `(#:python ,python-2 ; does not work with Python 3
4076 #:tests? #f)) ; there are no tests
4077 (inputs
4078 `(("python2-matplotlib" ,python2-matplotlib)))
4079 (home-page "https://bitbucket.org/regulatorygenomicsupf/pyicoteo")
4080 (synopsis "Analyze high-throughput genetic sequencing data")
4081 (description
4082 "Pyicoteo is a suite of tools for the analysis of high-throughput genetic
4083sequencing data. It works with genomic coordinates. There are currently six
4084different command-line tools:
4085
4086@enumerate
4087@item pyicoregion: for generating exploratory regions automatically;
4088@item pyicoenrich: for differential enrichment between two conditions;
4089@item pyicoclip: for calling CLIP-Seq peaks without a control;
4090@item pyicos: for genomic coordinates manipulation;
4091@item pyicoller: for peak calling on punctuated ChIP-Seq;
4092@item pyicount: to count how many reads from N experiment files overlap in a
4093 region file;
4094@item pyicotrocol: to combine operations from pyicoteo.
4095@end enumerate\n")
4096 (license license:gpl3+)))
4097
af860475
BW		 4098(define-public prodigal
4099 (package
4100 (name "prodigal")
e70f7a23 4101 (version "2.6.3")
af860475
BW		 4102 (source (origin
4103 (method url-fetch)
4104 (uri (string-append
4105 "https://github.com/hyattpd/Prodigal/archive/v"
4106 version ".tar.gz"))
4107 (file-name (string-append name "-" version ".tar.gz"))
4108 (sha256
4109 (base32
e70f7a23 4110 "17srxkqd3jc77xk15pfbgg1a9xahqg7337w95mrsia7mpza4l2c9"))))
af860475
BW		 4111 (build-system gnu-build-system)
4112 (arguments
4113 `(#:tests? #f ;no check target
4114 #:make-flags (list (string-append "INSTALLDIR="
4115 (assoc-ref %outputs "out")
4116 "/bin"))
4117 #:phases
4118 (modify-phases %standard-phases
4119 (delete 'configure))))
4120 (home-page "http://prodigal.ornl.gov")
4121 (synopsis "Protein-coding gene prediction for Archaea and Bacteria")
4122 (description
4123 "Prodigal runs smoothly on finished genomes, draft genomes, and
4124metagenomes, providing gene predictions in GFF3, Genbank, or Sequin table
4125format. It runs quickly, in an unsupervised fashion, handles gaps, handles
4126partial genes, and identifies translation initiation sites.")
4127 (license license:gpl3+)))
608dd932 4128
ceb62d54
BW		 4129(define-public roary
4130 (package
4131 (name "roary")
75696130 4132 (version "3.8.2")
ceb62d54
BW		 4133 (source
4134 (origin
4135 (method url-fetch)
4136 (uri (string-append
4137 "mirror://cpan/authors/id/A/AJ/AJPAGE/Bio-Roary-"
4138 version ".tar.gz"))
4139 (sha256
4140 (base32
75696130 4141 "03dfr2cd5fp80bcr65923zpdzrasvcxl7c2vgh8373v25a1yfap7"))))
ceb62d54
BW		 4142 (build-system perl-build-system)
4143 (arguments
4144 `(#:phases
4145 (modify-phases %standard-phases
4146 (delete 'configure)
4147 (delete 'build)
4148 (replace 'check
4149 (lambda _
4150 ;; The tests are not run by default, so we run each test file
4151 ;; directly.
4152 (setenv "PATH" (string-append (getcwd) "/bin" ":"
4153 (getenv "PATH")))
4154 (setenv "PERL5LIB" (string-append (getcwd) "/lib" ":"
4155 (getenv "PERL5LIB")))
4156 (zero? (length (filter (lambda (file)
4157 (display file)(display "\n")
4158 (not (zero? (system* "perl" file))))
4159 (find-files "t" ".*\\.t$"))))))
4160 (replace 'install
4161 ;; There is no 'install' target in the Makefile.
4162 (lambda* (#:key outputs #:allow-other-keys)
4163 (let* ((out (assoc-ref outputs "out"))
4164 (bin (string-append out "/bin"))
4165 (perl (string-append out "/lib/perl5/site_perl"))
4166 (roary-plots "contrib/roary_plots"))
4167 (mkdir-p bin)
4168 (mkdir-p perl)
4169 (copy-recursively "bin" bin)
4170 (copy-recursively "lib" perl)
4171 #t)))
4172 (add-after 'install 'wrap-programs
4173 (lambda* (#:key inputs outputs #:allow-other-keys)
4174 (let* ((out (assoc-ref outputs "out"))
4175 (perl5lib (getenv "PERL5LIB"))
4176 (path (getenv "PATH")))
4177 (for-each (lambda (prog)
4178 (let ((binary (string-append out "/" prog)))
4179 (wrap-program binary
4180 `("PERL5LIB" ":" prefix
4181 (,(string-append perl5lib ":" out
4182 "/lib/perl5/site_perl"))))
4183 (wrap-program binary
4184 `("PATH" ":" prefix
4185 (,(string-append path ":" out "/bin"))))))
4186 (find-files "bin" ".*[^R]$"))
4187 (let ((file
4188 (string-append out "/bin/roary-create_pan_genome_plots.R"))
4189 (r-site-lib (getenv "R_LIBS_SITE"))
4190 (coreutils-path
4191 (string-append (assoc-ref inputs "coreutils") "/bin")))
4192 (wrap-program file
4193 `("R_LIBS_SITE" ":" prefix
4194 (,(string-append r-site-lib ":" out "/site-library/"))))
4195 (wrap-program file
4196 `("PATH" ":" prefix
4197 (,(string-append coreutils-path ":" out "/bin"))))))
4198 #t)))))
4199 (native-inputs
4200 `(("perl-env-path" ,perl-env-path)
4201 ("perl-test-files" ,perl-test-files)
4202 ("perl-test-most" ,perl-test-most)
4203 ("perl-test-output" ,perl-test-output)))
4204 (inputs
4205 `(("perl-array-utils" ,perl-array-utils)
4206 ("bioperl" ,bioperl-minimal)
4207 ("perl-exception-class" ,perl-exception-class)
4208 ("perl-file-find-rule" ,perl-file-find-rule)
4209 ("perl-file-grep" ,perl-file-grep)
4210 ("perl-file-slurper" ,perl-file-slurper)
4211 ("perl-file-which" ,perl-file-which)
4212 ("perl-graph" ,perl-graph)
4213 ("perl-graph-readwrite" ,perl-graph-readwrite)
4214 ("perl-log-log4perl" ,perl-log-log4perl)
4215 ("perl-moose" ,perl-moose)
4216 ("perl-perlio-utf8_strict" ,perl-perlio-utf8_strict)
4217 ("perl-text-csv" ,perl-text-csv)
4218 ("bedtools" ,bedtools)
4219 ("cd-hit" ,cd-hit)
4220 ("blast+" ,blast+)
4221 ("mcl" ,mcl)
4222 ("parallel" ,parallel)
4223 ("prank" ,prank)
4224 ("mafft" ,mafft)
4225 ("fasttree" ,fasttree)
4226 ("grep" ,grep)
4227 ("sed" ,sed)
4228 ("gawk" ,gawk)
2d7c4ae3 4229 ("r-minimal" ,r-minimal)
ceb62d54
BW		 4230 ("r-ggplot2" ,r-ggplot2)
4231 ("coreutils" ,coreutils)))
4232 (home-page "http://sanger-pathogens.github.io/Roary")
4233 (synopsis "High speed stand-alone pan genome pipeline")
4234 (description
4235 "Roary is a high speed stand alone pan genome pipeline, which takes
4236annotated assemblies in GFF3 format (produced by the Prokka program) and
4237calculates the pan genome. Using a standard desktop PC, it can analyse
4238datasets with thousands of samples, without compromising the quality of the
4239results. 128 samples can be analysed in under 1 hour using 1 GB of RAM and a
4240single processor. Roary is not intended for metagenomics or for comparing
4241extremely diverse sets of genomes.")
4242 (license license:gpl3)))
4243
608dd932
BW		 4244(define-public raxml
4245 (package
4246 (name "raxml")
e9e0fab0 4247 (version "8.2.10")
608dd932
BW		 4248 (source
4249 (origin
4250 (method url-fetch)
4251 (uri
4252 (string-append
4253 "https://github.com/stamatak/standard-RAxML/archive/v"
4254 version ".tar.gz"))
4255 (file-name (string-append name "-" version ".tar.gz"))
4256 (sha256
4257 (base32
e9e0fab0 4258 "13s7aspfdcfr6asynwdg1x6vznys6pzap5f8wsffbnnwpkkg9ya8"))))
608dd932
BW		 4259 (build-system gnu-build-system)
4260 (arguments
4261 `(#:tests? #f ; There are no tests.
4262 ;; Use 'standard' Makefile rather than SSE or AVX ones.
4263 #:make-flags (list "-f" "Makefile.HYBRID.gcc")
4264 #:phases
4265 (modify-phases %standard-phases
4266 (delete 'configure)
4267 (replace 'install
4268 (lambda* (#:key outputs #:allow-other-keys)
4269 (let* ((out (assoc-ref outputs "out"))
4270 (bin (string-append out "/bin"))
4271 (executable "raxmlHPC-HYBRID"))
4272 (install-file executable bin)
4273 (symlink (string-append bin "/" executable) "raxml"))
4274 #t)))))
4275 (inputs
4276 `(("openmpi" ,openmpi)))
4277 (home-page "http://sco.h-its.org/exelixis/web/software/raxml/index.html")
4278 (synopsis "Randomized Axelerated Maximum Likelihood phylogenetic trees")
4279 (description
4280 "RAxML is a tool for phylogenetic analysis and post-analysis of large
4281phylogenies.")
2805f9b2
EF		 4282 ;; The source includes x86 specific code
4283 (supported-systems '("x86_64-linux" "i686-linux"))
608dd932 4284 (license license:gpl2+)))
af860475 4285
66e3eff1
RW		 4286(define-public rsem
4287 (package
4288 (name "rsem")
4289 (version "1.2.20")
4290 (source
4291 (origin
4292 (method url-fetch)
4293 (uri
4294 (string-append "http://deweylab.biostat.wisc.edu/rsem/src/rsem-"
4295 version ".tar.gz"))
4296 (sha256
4297 (base32 "0nzdc0j0hjllhsd5f2xli95dafm3nawskigs140xzvjk67xh0r9q"))
fc1adab1 4298 (patches (search-patches "rsem-makefile.patch"))
66e3eff1
RW		 4299 (modules '((guix build utils)))
4300 (snippet
4301 '(begin
4302 ;; remove bundled copy of boost
4303 (delete-file-recursively "boost")
4304 #t))))
4305 (build-system gnu-build-system)
4306 (arguments
4307 `(#:tests? #f ;no "check" target
4308 #:phases
4309 (modify-phases %standard-phases
4310 ;; No "configure" script.
4311 ;; Do not build bundled samtools library.
4312 (replace 'configure
4313 (lambda _
4314 (substitute* "Makefile"
4315 (("^all : sam/libbam.a") "all : "))
4316 #t))
4317 (replace 'install
4318 (lambda* (#:key outputs #:allow-other-keys)
4319 (let* ((out (string-append (assoc-ref outputs "out")))
4320 (bin (string-append out "/bin/"))
4321 (perl (string-append out "/lib/perl5/site_perl")))
4322 (mkdir-p bin)
4323 (mkdir-p perl)
4324 (for-each (lambda (file)
f3860753 4325 (install-file file bin))
66e3eff1 4326 (find-files "." "rsem-.*"))
f3860753 4327 (install-file "rsem_perl_utils.pm" perl))
66e3eff1
RW		 4328 #t))
4329 (add-after
4330 'install 'wrap-program
4331 (lambda* (#:key outputs #:allow-other-keys)
4332 (let ((out (assoc-ref outputs "out")))
4333 (for-each (lambda (prog)
4334 (wrap-program (string-append out "/bin/" prog)
4335 `("PERL5LIB" ":" prefix
4336 (,(string-append out "/lib/perl5/site_perl")))))
4337 '("rsem-plot-transcript-wiggles"
4338 "rsem-calculate-expression"
4339 "rsem-generate-ngvector"
4340 "rsem-run-ebseq"
4341 "rsem-prepare-reference")))
4342 #t)))))
4343 (inputs
4344 `(("boost" ,boost)
4345 ("ncurses" ,ncurses)
2d7c4ae3 4346 ("r-minimal" ,r-minimal)
66e3eff1
RW		 4347 ("perl" ,perl)
4348 ("samtools" ,samtools-0.1)
4349 ("zlib" ,zlib)))
4350 (home-page "http://deweylab.biostat.wisc.edu/rsem/")
4351 (synopsis "Estimate gene expression levels from RNA-Seq data")
4352 (description
4353 "RSEM is a software package for estimating gene and isoform expression
4354levels from RNA-Seq data. The RSEM package provides a user-friendly
4355interface, supports threads for parallel computation of the EM algorithm,
4356single-end and paired-end read data, quality scores, variable-length reads and
4357RSPD estimation. In addition, it provides posterior mean and 95% credibility
4358interval estimates for expression levels. For visualization, it can generate
4359BAM and Wiggle files in both transcript-coordinate and genomic-coordinate.")
4360 (license license:gpl3+)))
4361
8622a072
RW		 4362(define-public rseqc
4363 (package
4364 (name "rseqc")
4365 (version "2.6.1")
4366 (source
4367 (origin
4368 (method url-fetch)
4369 (uri
4370 (string-append "mirror://sourceforge/rseqc/"
de67e922 4371 "RSeQC-" version ".tar.gz"))
8622a072 4372 (sha256
8214b7fb 4373 (base32 "15ly0254yi032qzkdplg00q144qfdsd986gh62829rl5bkxhj330"))
8622a072
RW		 4374 (modules '((guix build utils)))
4375 (snippet
4376 '(begin
4377 ;; remove bundled copy of pysam
4378 (delete-file-recursively "lib/pysam")
4379 (substitute* "setup.py"
4380 ;; remove dependency on outdated "distribute" module
4381 (("^from distribute_setup import use_setuptools") "")
4382 (("^use_setuptools\\(\\)") "")
4383 ;; do not use bundled copy of pysam
4384 (("^have_pysam = False") "have_pysam = True"))))))
4385 (build-system python-build-system)
4386 (arguments `(#:python ,python-2))
4387 (inputs
4388 `(("python-cython" ,python2-cython)
4389 ("python-pysam" ,python2-pysam)
4390 ("python-numpy" ,python2-numpy)
8622a072
RW		 4391 ("zlib" ,zlib)))
4392 (native-inputs
4393 `(("python-nose" ,python2-nose)))
4394 (home-page "http://rseqc.sourceforge.net/")
4395 (synopsis "RNA-seq quality control package")
4396 (description
4397 "RSeQC provides a number of modules that can comprehensively evaluate
4398high throughput sequence data, especially RNA-seq data. Some basic modules
4399inspect sequence quality, nucleotide composition bias, PCR bias and GC bias,
4400while RNA-seq specific modules evaluate sequencing saturation, mapped reads
4401distribution, coverage uniformity, strand specificity, etc.")
4402 (license license:gpl3+)))
4403
ec946638
RW		 4404(define-public seek
4405 ;; There are no release tarballs. According to the installation
4406 ;; instructions at http://seek.princeton.edu/installation.jsp, the latest
4407 ;; stable release is identified by this changeset ID.
4408 (let ((changeset "2329130")
4409 (revision "1"))
4410 (package
4411 (name "seek")
4412 (version (string-append "0-" revision "." changeset))
4413 (source (origin
4414 (method hg-fetch)
4415 (uri (hg-reference
4416 (url "https://bitbucket.org/libsleipnir/sleipnir")
4417 (changeset changeset)))
4418 (sha256
4419 (base32
4420 "0qrvilwh18dpbhkf92qvxbmay0j75ra3jg2wrhz67gf538zzphsx"))))
4421 (build-system gnu-build-system)
4422 (arguments
4423 `(#:modules ((srfi srfi-1)
4424 (guix build gnu-build-system)
4425 (guix build utils))
4426 #:phases
4427 (let ((dirs '("SeekMiner"
4428 "SeekEvaluator"
4429 "SeekPrep"
4430 "Distancer"
4431 "Data2DB"
4432 "PCL2Bin")))
4433 (modify-phases %standard-phases
4561498e 4434 (add-before 'configure 'bootstrap
ec946638
RW		 4435 (lambda _
4436 (zero? (system* "bash" "gen_auto"))))
4437 (add-after 'build 'build-additional-tools
4438 (lambda* (#:key make-flags #:allow-other-keys)
4439 (every (lambda (dir)
4440 (with-directory-excursion (string-append "tools/" dir)
4441 (zero? (apply system* "make" make-flags))))
4442 dirs)))
4443 (add-after 'install 'install-additional-tools
4444 (lambda* (#:key make-flags #:allow-other-keys)
4445 (fold (lambda (dir result)
4446 (with-directory-excursion (string-append "tools/" dir)
4447 (and result
4448 (zero? (apply system*
4449 `("make" ,@make-flags "install"))))))
4450 #t dirs)))))))
4451 (inputs
4452 `(("gsl" ,gsl)
4453 ("boost" ,boost)
4454 ("libsvm" ,libsvm)
4455 ("readline" ,readline)
4456 ("gengetopt" ,gengetopt)
4457 ("log4cpp" ,log4cpp)))
4458 (native-inputs
4459 `(("autoconf" ,autoconf)
4460 ("automake" ,automake)
4461 ("perl" ,perl)))
4462 (home-page "http://seek.princeton.edu")
4463 (synopsis "Gene co-expression search engine")
4464 (description
4465 "SEEK is a computational gene co-expression search engine. SEEK provides
4466biologists with a way to navigate the massive human expression compendium that
4467now contains thousands of expression datasets. SEEK returns a robust ranking
4468of co-expressed genes in the biological area of interest defined by the user's
4469query genes. It also prioritizes thousands of expression datasets according
4470to the user's query of interest.")
4471 (license license:cc-by3.0))))
4472
4e10a221
RW		 4473(define-public samtools
4474 (package
4475 (name "samtools")
63fa0ffc 4476 (version "1.5")
4e10a221
RW		 4477 (source
4478 (origin
4479 (method url-fetch)
4480 (uri
de67e922 4481 (string-append "mirror://sourceforge/samtools/samtools/"
4e10a221
RW		 4482 version "/samtools-" version ".tar.bz2"))
4483 (sha256
4484 (base32
63fa0ffc 4485 "1xidmv0jmfy7l0kb32hdnlshcxgzi1hmygvig0cqrq1fhckdlhl5"))))
4e10a221
RW		 4486 (build-system gnu-build-system)
4487 (arguments
c4473411 4488 `(#:modules ((ice-9 ftw)
5bdda30b
RW		 4489 (ice-9 regex)
4490 (guix build gnu-build-system)
4491 (guix build utils))
c4473411 4492 #:make-flags (list (string-append "prefix=" (assoc-ref %outputs "out")))
4ab16440 4493 #:configure-flags (list "--with-ncurses" "--with-htslib=system")
4e10a221 4494 #:phases
c842d425
BW		 4495 (modify-phases %standard-phases
4496 (add-after 'unpack 'patch-tests
4497 (lambda _
4498 (substitute* "test/test.pl"
4499 ;; The test script calls out to /bin/bash
4500 (("/bin/bash") (which "bash")))
b2d748ed 4501 #t))
c842d425
BW		 4502 (add-after 'install 'install-library
4503 (lambda* (#:key outputs #:allow-other-keys)
4504 (let ((lib (string-append (assoc-ref outputs "out") "/lib")))
4505 (install-file "libbam.a" lib)
4506 #t)))
4507 (add-after 'install 'install-headers
4508 (lambda* (#:key outputs #:allow-other-keys)
4509 (let ((include (string-append (assoc-ref outputs "out")
4510 "/include/samtools/")))
4511 (for-each (lambda (file)
4512 (install-file file include))
4513 (scandir "." (lambda (name) (string-match "\\.h$" name))))
4514 #t))))))
4e10a221 4515 (native-inputs `(("pkg-config" ,pkg-config)))
4ab16440
BW		 4516 (inputs
4517 `(("htslib" ,htslib)
4518 ("ncurses" ,ncurses)
4519 ("perl" ,perl)
4520 ("python" ,python)
4521 ("zlib" ,zlib)))
4e10a221
RW		 4522 (home-page "http://samtools.sourceforge.net")
4523 (synopsis "Utilities to efficiently manipulate nucleotide sequence alignments")
4524 (description
4525 "Samtools implements various utilities for post-processing nucleotide
4526sequence alignments in the SAM, BAM, and CRAM formats, including indexing,
4527variant calling (in conjunction with bcftools), and a simple alignment
4528viewer.")
4529 (license license:expat)))
d3517eda 4530
0b84a0aa
RW		 4531(define-public samtools-0.1
4532 ;; This is the most recent version of the 0.1 line of samtools. The input
4533 ;; and output formats differ greatly from that used and produced by samtools
4534 ;; 1.x and is still used in many bioinformatics pipelines.
4535 (package (inherit samtools)
4536 (version "0.1.19")
4537 (source
4538 (origin
4539 (method url-fetch)
4540 (uri
de67e922 4541 (string-append "mirror://sourceforge/samtools/samtools/"
0b84a0aa
RW		 4542 version "/samtools-" version ".tar.bz2"))
4543 (sha256
4544 (base32 "1m33xsfwz0s8qi45lylagfllqg7fphf4dr0780rsvw75av9wk06h"))))
4545 (arguments
2309ed68
RW		 4546 `(#:tests? #f ;no "check" target
4547 ,@(substitute-keyword-arguments (package-arguments samtools)
4548 ((#:make-flags flags)
4549 `(cons "LIBCURSES=-lncurses" ,flags))
4550 ((#:phases phases)
4551 `(modify-phases ,phases
4552 (replace 'install
4553 (lambda* (#:key outputs #:allow-other-keys)
4554 (let ((bin (string-append
4555 (assoc-ref outputs "out") "/bin")))
4556 (mkdir-p bin)
f3860753 4557 (install-file "samtools" bin)
b2d748ed 4558 #t)))
2309ed68
RW		 4559 (delete 'patch-tests)
4560 (delete 'configure))))))))
0b84a0aa 4561
fe4c37c2 4562(define-public mosaik
698bd297 4563 (let ((commit "5c25216d3522d6a33e53875cd76a6d65001e4e67"))
fe4c37c2
RW		 4564 (package
4565 (name "mosaik")
4566 (version "2.2.30")
4567 (source (origin
4568 ;; There are no release tarballs nor tags.
4569 (method git-fetch)
4570 (uri (git-reference
4571 (url "https://github.com/wanpinglee/MOSAIK.git")
4572 (commit commit)))
4573 (file-name (string-append name "-" version))
4574 (sha256
4575 (base32
4576 "17gj3s07cm77r41z92awh0bim7w7q7fbn0sf5nkqmcm1vw052qgw"))))
4577 (build-system gnu-build-system)
4578 (arguments
4579 `(#:tests? #f ; no tests
4580 #:make-flags (list "CC=gcc")
4581 #:phases
4582 (modify-phases %standard-phases
4583 (replace 'configure
4584 (lambda _ (chdir "src") #t))
4585 (replace 'install
4586 (lambda* (#:key outputs #:allow-other-keys)
4587 (let ((bin (string-append (assoc-ref outputs "out")
4588 "/bin")))
4589 (mkdir-p bin)
4590 (copy-recursively "../bin" bin)
4591 #t))))))
4592 (inputs
4593 `(("perl" ,perl)
4594 ("zlib" ,zlib)))
029d9f77 4595 (supported-systems '("x86_64-linux"))
0c6c9c00 4596 (home-page "https://github.com/wanpinglee/MOSAIK")
fe4c37c2
RW		 4597 (synopsis "Map nucleotide sequence reads to reference genomes")
4598 (description
4599 "MOSAIK is a program for mapping second and third-generation sequencing
4600reads to a reference genome. MOSAIK can align reads generated by all the
4601major sequencing technologies, including Illumina, Applied Biosystems SOLiD,
4602Roche 454, Ion Torrent and Pacific BioSciences SMRT.")
4603 ;; MOSAIK is released under the GPLv2+ with the exception of third-party
4604 ;; code released into the public domain:
4605 ;; 1. fastlz by Ariya Hidayat - http://www.fastlz.org/
4606 ;; 2. MD5 implementation - RSA Data Security, RFC 1321
4607 (license (list license:gpl2+ license:public-domain)))))
4608
282c5087
RW		 4609(define-public ngs-sdk
4610 (package
4611 (name "ngs-sdk")
6c4ccf32 4612 (version "1.3.0")
282c5087
RW		 4613 (source
4614 (origin
4615 (method url-fetch)
4616 (uri
4617 (string-append "https://github.com/ncbi/ngs/archive/"
4618 version ".tar.gz"))
4619 (file-name (string-append name "-" version ".tar.gz"))
4620 (sha256
4621 (base32
6c4ccf32 4622 "1wiyf4c6nm2j87pv015cbi0qny5byf3pbvcw3likifz5dl56ag40"))))
282c5087
RW		 4623 (build-system gnu-build-system)
4624 (arguments
4625 `(#:parallel-build? #f ; not supported
4626 #:tests? #f ; no "check" target
4627 #:phases
dc1d3cde
KK		 4628 (modify-phases %standard-phases
4629 (replace 'configure
4630 (lambda* (#:key outputs #:allow-other-keys)
4631 (let ((out (assoc-ref outputs "out")))
4632 ;; Allow 'konfigure.perl' to find 'package.prl'.
4633 (setenv "PERL5LIB"
4634 (string-append ".:" (getenv "PERL5LIB")))
4635
4636 ;; The 'configure' script doesn't recognize things like
4637 ;; '--enable-fast-install'.
4638 (zero? (system* "./configure"
4639 (string-append "--build-prefix=" (getcwd) "/build")
4640 (string-append "--prefix=" out))))))
4641 (add-after 'unpack 'enter-dir
4642 (lambda _ (chdir "ngs-sdk") #t)))))
282c5087 4643 (native-inputs `(("perl" ,perl)))
a0dadf0c
AE		 4644 ;; According to the test
4645 ;; unless ($MARCH =~ /x86_64/i || $MARCH =~ /i?86/i)
4646 ;; in ngs-sdk/setup/konfigure.perl
ab29be81 4647 (supported-systems '("i686-linux" "x86_64-linux"))
282c5087
RW		 4648 (home-page "https://github.com/ncbi/ngs")
4649 (synopsis "API for accessing Next Generation Sequencing data")
4650 (description
4651 "NGS is a domain-specific API for accessing reads, alignments and pileups
4652produced from Next Generation Sequencing. The API itself is independent from
4653any particular back-end implementation, and supports use of multiple back-ends
4654simultaneously.")
4655 (license license:public-domain)))
4656
1ad15c16 4657(define-public java-ngs
2651a5e6 4658 (package (inherit ngs-sdk)
1ad15c16 4659 (name "java-ngs")
2651a5e6
RW		 4660 (arguments
4661 `(,@(substitute-keyword-arguments
4662 `(#:modules ((guix build gnu-build-system)
4663 (guix build utils)
4664 (srfi srfi-1)
4665 (srfi srfi-26))
4666 ,@(package-arguments ngs-sdk))
4667 ((#:phases phases)
614a8977
RW		 4668 `(modify-phases ,phases
4669 (replace 'enter-dir (lambda _ (chdir "ngs-java") #t)))))))
2651a5e6 4670 (inputs
d2540f80 4671 `(("jdk" ,icedtea "jdk")
2651a5e6
RW		 4672 ("ngs-sdk" ,ngs-sdk)))
4673 (synopsis "Java bindings for NGS SDK")))
4674
75dd2424
RW		 4675(define-public ncbi-vdb
4676 (package
4677 (name "ncbi-vdb")
5021f547 4678 (version "2.8.2")
75dd2424
RW		 4679 (source
4680 (origin
4681 (method url-fetch)
4682 (uri
4683 (string-append "https://github.com/ncbi/ncbi-vdb/archive/"
4684 version ".tar.gz"))
4685 (file-name (string-append name "-" version ".tar.gz"))
4686 (sha256
4687 (base32
5021f547 4688 "1acn4bv81mfl137qnbn9995mjjhwd36pm0b7qli1iw5skrxa9j8m"))))
75dd2424
RW		 4689 (build-system gnu-build-system)
4690 (arguments
4691 `(#:parallel-build? #f ; not supported
4692 #:tests? #f ; no "check" target
4693 #:phases
70111198 4694 (modify-phases %standard-phases
0691851a
BW		 4695 (add-before 'configure 'set-perl-search-path
4696 (lambda _
4697 ;; Work around "dotless @INC" build failure.
4698 (setenv "PERL5LIB"
4699 (string-append (getcwd) "/setup:"
4700 (getenv "PERL5LIB")))
4701 #t))
70111198
RW		 4702 (replace 'configure
4703 (lambda* (#:key inputs outputs #:allow-other-keys)
4704 (let ((out (assoc-ref outputs "out")))
4705 ;; Override include path for libmagic
4706 (substitute* "setup/package.prl"
4707 (("name => 'magic', Include => '/usr/include'")
4708 (string-append "name=> 'magic', Include => '"
4709 (assoc-ref inputs "libmagic")
4710 "/include" "'")))
75dd2424 4711
70111198
RW		 4712 ;; Install kdf5 library (needed by sra-tools)
4713 (substitute* "build/Makefile.install"
4714 (("LIBRARIES_TO_INSTALL =")
4715 "LIBRARIES_TO_INSTALL = kdf5.$(VERSION_LIBX) kdf5.$(VERSION_SHLX)"))
75dd2424 4716
70111198
RW		 4717 (substitute* "build/Makefile.env"
4718 (("CFLAGS =" prefix)
4719 (string-append prefix "-msse2 ")))
675d7ae2 4720
558e2307
RW		 4721 ;; Override search path for ngs-java
4722 (substitute* "setup/package.prl"
4723 (("/usr/local/ngs/ngs-java")
4724 (assoc-ref inputs "java-ngs")))
4725
70111198
RW		 4726 ;; The 'configure' script doesn't recognize things like
4727 ;; '--enable-fast-install'.
4728 (zero? (system*
4729 "./configure"
4730 (string-append "--build-prefix=" (getcwd) "/build")
4731 (string-append "--prefix=" (assoc-ref outputs "out"))
4732 (string-append "--debug")
4733 (string-append "--with-xml2-prefix="
4734 (assoc-ref inputs "libxml2"))
4735 (string-append "--with-ngs-sdk-prefix="
4736 (assoc-ref inputs "ngs-sdk"))
70111198
RW		 4737 (string-append "--with-hdf5-prefix="
4738 (assoc-ref inputs "hdf5")))))))
4739 (add-after 'install 'install-interfaces
4740 (lambda* (#:key outputs #:allow-other-keys)
4741 ;; Install interface libraries. On i686 the interface libraries
4742 ;; are installed to "linux/gcc/i386", so we need to use the Linux
4743 ;; architecture name ("i386") instead of the target system prefix
4744 ;; ("i686").
4745 (mkdir (string-append (assoc-ref outputs "out") "/ilib"))
4746 (copy-recursively (string-append "build/ncbi-vdb/linux/gcc/"
4747 ,(system->linux-architecture
4748 (or (%current-target-system)
4749 (%current-system)))
4750 "/rel/ilib")
4751 (string-append (assoc-ref outputs "out")
4752 "/ilib"))
4753 ;; Install interface headers
4754 (copy-recursively "interfaces"
4755 (string-append (assoc-ref outputs "out")
4756 "/include"))
d5e17162
RW		 4757 #t))
4758 ;; These files are needed by sra-tools.
4759 (add-after 'install 'install-configuration-files
4760 (lambda* (#:key outputs #:allow-other-keys)
4761 (let ((target (string-append (assoc-ref outputs "out") "/kfg")))
4762 (mkdir target)
4763 (install-file "libs/kfg/default.kfg" target)
4764 (install-file "libs/kfg/certs.kfg" target))
70111198 4765 #t)))))
75dd2424
RW		 4766 (inputs
4767 `(("libxml2" ,libxml2)
4768 ("ngs-sdk" ,ngs-sdk)
1ad15c16 4769 ("java-ngs" ,java-ngs)
75dd2424
RW		 4770 ("libmagic" ,file)
4771 ("hdf5" ,hdf5)))
4772 (native-inputs `(("perl" ,perl)))
675d7ae2
RW		 4773 ;; NCBI-VDB requires SSE capability.
4774 (supported-systems '("i686-linux" "x86_64-linux"))
75dd2424
RW		 4775 (home-page "https://github.com/ncbi/ncbi-vdb")
4776 (synopsis "Database engine for genetic information")
4777 (description
4778 "The NCBI-VDB library implements a highly compressed columnar data
4779warehousing engine that is most often used to store genetic information.
4780Databases are stored in a portable image within the file system, and can be
4781accessed/downloaded on demand across HTTP.")
4782 (license license:public-domain)))
4783
cc6ed477
RW		 4784(define-public plink
4785 (package
4786 (name "plink")
4787 (version "1.07")
4788 (source
4789 (origin
4790 (method url-fetch)
4791 (uri (string-append
4792 "http://pngu.mgh.harvard.edu/~purcell/plink/dist/plink-"
4793 version "-src.zip"))
4794 (sha256
4795 (base32 "0as8gxm4pjyc8dxmm1sl873rrd7wn5qs0l29nqfnl31x8i467xaa"))
0dbb7ac2
EF		 4796 (patches (search-patches "plink-1.07-unclobber-i.patch"
4797 "plink-endian-detection.patch"))))
cc6ed477
RW		 4798 (build-system gnu-build-system)
4799 (arguments
4800 '(#:tests? #f ;no "check" target
4801 #:make-flags (list (string-append "LIB_LAPACK="
4802 (assoc-ref %build-inputs "lapack")
4803 "/lib/liblapack.so")
4804 "WITH_LAPACK=1"
4805 "FORCE_DYNAMIC=1"
4806 ;; disable phoning home
4807 "WITH_WEBCHECK=")
4808 #:phases
4809 (modify-phases %standard-phases
4810 ;; no "configure" script
4811 (delete 'configure)
4812 (replace 'install
4813 (lambda* (#:key outputs #:allow-other-keys)
4814 (let ((bin (string-append (assoc-ref outputs "out")
4815 "/bin/")))
96c46210 4816 (install-file "plink" bin)
cc6ed477
RW		 4817 #t))))))
4818 (inputs
4819 `(("zlib" ,zlib)
4820 ("lapack" ,lapack)))
4821 (native-inputs
4822 `(("unzip" ,unzip)))
4823 (home-page "http://pngu.mgh.harvard.edu/~purcell/plink/")
4824 (synopsis "Whole genome association analysis toolset")
4825 (description
4826 "PLINK is a whole genome association analysis toolset, designed to
4827perform a range of basic, large-scale analyses in a computationally efficient
4828manner. The focus of PLINK is purely on analysis of genotype/phenotype data,
4829so there is no support for steps prior to this (e.g. study design and
4830planning, generating genotype or CNV calls from raw data). Through
4831integration with gPLINK and Haploview, there is some support for the
4832subsequent visualization, annotation and storage of results.")
4833 ;; Code is released under GPLv2, except for fisher.h, which is under
4834 ;; LGPLv2.1+
4835 (license (list license:gpl2 license:lgpl2.1+))))
4836
dad66da4
RW		 4837(define-public plink-ng
4838 (package (inherit plink)
4839 (name "plink-ng")
4840 (version "1.90b4")
4841 (source
4842 (origin
4843 (method url-fetch)
4844 (uri (string-append "https://github.com/chrchang/plink-ng/archive/v"
4845 version ".tar.gz"))
4846 (file-name (string-append name "-" version ".tar.gz"))
4847 (sha256
4848 (base32 "09ixrds009aczjswxr2alcb774mksq5g0v78dgjjn1h4dky0kf9a"))))
4849 (build-system gnu-build-system)
4850 (arguments
4851 '(#:tests? #f ;no "check" target
4852 #:make-flags (list "BLASFLAGS=-llapack -lopenblas"
4853 "CFLAGS=-Wall -O2 -DDYNAMIC_ZLIB=1"
4854 "ZLIB=-lz"
4855 "-f" "Makefile.std")
4856 #:phases
4857 (modify-phases %standard-phases
4858 (add-after 'unpack 'chdir
4859 (lambda _ (chdir "1.9") #t))
4860 (delete 'configure) ; no "configure" script
4861 (replace 'install
4862 (lambda* (#:key outputs #:allow-other-keys)
4863 (let ((bin (string-append (assoc-ref outputs "out")
4864 "/bin/")))
4865 (install-file "plink" bin)
4866 #t))))))
4867 (inputs
4868 `(("zlib" ,zlib)
4869 ("lapack" ,lapack)
4870 ("openblas" ,openblas)))
4871 (home-page "https://www.cog-genomics.org/plink/")
4872 (license license:gpl3+)))
4873
c6a24d6e
RW		 4874(define-public smithlab-cpp
4875 (let ((revision "1")
698bd297 4876 (commit "728a097bec88c6f4b8528b685932049e660eff2e"))
c6a24d6e
RW		 4877 (package
4878 (name "smithlab-cpp")
698bd297 4879 (version (string-append "0." revision "." (string-take commit 7)))
c6a24d6e
RW		 4880 (source (origin
4881 (method git-fetch)
4882 (uri (git-reference
4883 (url "https://github.com/smithlabcode/smithlab_cpp.git")
4884 (commit commit)))
4885 (file-name (string-append name "-" version "-checkout"))
4886 (sha256
4887 (base32
4888 "0d476lmj312xk77kr9fzrv7z1bv96yfyx0w7y62ycmnfbx32ll74"))))
4889 (build-system gnu-build-system)
4890 (arguments
4891 `(#:modules ((guix build gnu-build-system)
4892 (guix build utils)
4893 (srfi srfi-26))
4894 #:tests? #f ;no "check" target
4895 #:phases
4896 (modify-phases %standard-phases
4897 (add-after 'unpack 'use-samtools-headers
4898 (lambda _
4899 (substitute* '("SAM.cpp"
4900 "SAM.hpp")
4901 (("sam.h") "samtools/sam.h"))
4902 #t))
4903 (replace 'install
4904 (lambda* (#:key outputs #:allow-other-keys)
4905 (let* ((out (assoc-ref outputs "out"))
4906 (lib (string-append out "/lib"))
4907 (include (string-append out "/include/smithlab-cpp")))
4908 (mkdir-p lib)
4909 (mkdir-p include)
4910 (for-each (cut install-file <> lib)
4911 (find-files "." "\\.o$"))
4912 (for-each (cut install-file <> include)
4913 (find-files "." "\\.hpp$")))
4914 #t))
4915 (delete 'configure))))
4916 (inputs
4917 `(("samtools" ,samtools-0.1)
4918 ("zlib" ,zlib)))
4919 (home-page "https://github.com/smithlabcode/smithlab_cpp")
4920 (synopsis "C++ helper library for functions used in Smith lab projects")
4921 (description
4922 "Smithlab CPP is a C++ library that includes functions used in many of
4923the Smith lab bioinformatics projects, such as a wrapper around Samtools data
4924structures, classes for genomic regions, mapped sequencing reads, etc.")
4925 (license license:gpl3+))))
4926
56e373ef
RW		 4927(define-public preseq
4928 (package
4929 (name "preseq")
b49c5a58 4930 (version "2.0")
56e373ef
RW		 4931 (source (origin
4932 (method url-fetch)
b49c5a58
RW		 4933 (uri (string-append "https://github.com/smithlabcode/"
4934 "preseq/archive/v" version ".tar.gz"))
4935 (file-name (string-append name "-" version ".tar.gz"))
56e373ef 4936 (sha256
b49c5a58 4937 (base32 "08r684l50pnxjpvmhzjgqq56yv9rfw90k8vx0nsrnrzk8mf9hsdq"))
56e373ef
RW		 4938 (modules '((guix build utils)))
4939 (snippet
4940 ;; Remove bundled samtools.
b49c5a58 4941 '(delete-file-recursively "samtools"))))
56e373ef
RW		 4942 (build-system gnu-build-system)
4943 (arguments
4944 `(#:tests? #f ;no "check" target
4945 #:phases
4946 (modify-phases %standard-phases
56e373ef 4947 (delete 'configure))
b49c5a58
RW		 4948 #:make-flags
4949 (list (string-append "PREFIX="
4950 (assoc-ref %outputs "out"))
4951 (string-append "LIBBAM="
4952 (assoc-ref %build-inputs "samtools")
4953 "/lib/libbam.a")
4954 (string-append "SMITHLAB_CPP="
4955 (assoc-ref %build-inputs "smithlab-cpp")
4956 "/lib")
4957 "PROGS=preseq"
4958 "INCLUDEDIRS=$(SMITHLAB_CPP)/../include/smithlab-cpp $(SAMTOOLS_DIR)")))
56e373ef
RW		 4959 (inputs
4960 `(("gsl" ,gsl)
4961 ("samtools" ,samtools-0.1)
b49c5a58 4962 ("smithlab-cpp" ,smithlab-cpp)
56e373ef
RW		 4963 ("zlib" ,zlib)))
4964 (home-page "http://smithlabresearch.org/software/preseq/")
4965 (synopsis "Program for analyzing library complexity")
4966 (description
4967 "The preseq package is aimed at predicting and estimating the complexity
4968of a genomic sequencing library, equivalent to predicting and estimating the
4969number of redundant reads from a given sequencing depth and how many will be
4970expected from additional sequencing using an initial sequencing experiment.
4971The estimates can then be used to examine the utility of further sequencing,
4972optimize the sequencing depth, or to screen multiple libraries to avoid low
4973complexity samples.")
4974 (license license:gpl3+)))
4975
9ded1457
BW		 4976(define-public python-screed
4977 (package
4978 (name "python-screed")
4979 (version "0.9")
4980 (source
4981 (origin
4982 (method url-fetch)
4983 (uri (pypi-uri "screed" version))
4984 (sha256
4985 (base32
4986 "18czszp9fkx3j6jr7y5kp6dfialscgddk05mw1zkhh2zhn0jd8i0"))))
4987 (build-system python-build-system)
4988 (arguments
4989 `(#:phases
4990 (modify-phases %standard-phases
4991 (replace 'check
4992 (lambda _
4993 (setenv "PYTHONPATH"
4994 (string-append (getenv "PYTHONPATH") ":."))
4995 (zero? (system* "nosetests" "--attr" "!known_failing")))))))
4996 (native-inputs
4997 `(("python-nose" ,python-nose)))
4998 (inputs
4999 `(("python-bz2file" ,python-bz2file)))
7bf837fd 5000 (home-page "https://github.com/dib-lab/screed/")
9ded1457
BW		 5001 (synopsis "Short read sequence database utilities")
5002 (description "Screed parses FASTA and FASTQ files and generates databases.
5003Values such as sequence name, sequence description, sequence quality and the
5004sequence itself can be retrieved from these databases.")
5005 (license license:bsd-3)))
5006
5007(define-public python2-screed
5c31f4aa 5008 (package-with-python2 python-screed))
9ded1457 5009
51c64999
RW		 5010(define-public sra-tools
5011 (package
5012 (name "sra-tools")
79849358 5013 (version "2.8.2-1")
51c64999
RW		 5014 (source
5015 (origin
5016 (method url-fetch)
5017 (uri
5018 (string-append "https://github.com/ncbi/sra-tools/archive/"
5019 version ".tar.gz"))
5020 (file-name (string-append name "-" version ".tar.gz"))
5021 (sha256
5022 (base32
79849358 5023 "1camsijmvv2s45mb4iyf44ghl4gkd4rl0viphpcgl3ccchy32a0g"))))
51c64999
RW		 5024 (build-system gnu-build-system)
5025 (arguments
5026 `(#:parallel-build? #f ; not supported
5027 #:tests? #f ; no "check" target
2320e76b 5028 #:make-flags
79849358
RW		 5029 (list (string-append "DEFAULT_CRT="
5030 (assoc-ref %build-inputs "ncbi-vdb")
5031 "/kfg/certs.kfg")
5032 (string-append "DEFAULT_KFG="
5033 (assoc-ref %build-inputs "ncbi-vdb")
5034 "/kfg/default.kfg")
5035 (string-append "VDB_LIBDIR="
2320e76b
RW		 5036 (assoc-ref %build-inputs "ncbi-vdb")
5037 ,(if (string-prefix? "x86_64"
5038 (or (%current-target-system)
5039 (%current-system)))
5040 "/lib64"
5041 "/lib32")))
51c64999 5042 #:phases
beebe431 5043 (modify-phases %standard-phases
0691851a
BW		 5044 (add-before 'configure 'set-perl-search-path
5045 (lambda _
5046 ;; Work around "dotless @INC" build failure.
5047 (setenv "PERL5LIB"
5048 (string-append (getcwd) "/setup:"
5049 (getenv "PERL5LIB")))
5050 #t))
beebe431
RW		 5051 (replace 'configure
5052 (lambda* (#:key inputs outputs #:allow-other-keys)
5053 ;; The build system expects a directory containing the sources and
5054 ;; raw build output of ncbi-vdb, including files that are not
5055 ;; installed. Since we are building against an installed version of
5056 ;; ncbi-vdb, the following modifications are needed.
5057 (substitute* "setup/konfigure.perl"
5058 ;; Make the configure script look for the "ilib" directory of
5059 ;; "ncbi-vdb" without first checking for the existence of a
5060 ;; matching library in its "lib" directory.
5061 (("^ my \\$f = File::Spec->catdir\\(\\$libdir, \\$lib\\);")
5062 "my $f = File::Spec->catdir($ilibdir, $ilib);")
5063 ;; Look for interface libraries in ncbi-vdb's "ilib" directory.
5064 (("my \\$ilibdir = File::Spec->catdir\\(\\$builddir, 'ilib'\\);")
5065 "my $ilibdir = File::Spec->catdir($dir, 'ilib');"))
51c64999 5066
beebe431
RW		 5067 ;; Dynamic linking
5068 (substitute* "tools/copycat/Makefile"
5069 (("smagic-static") "lmagic"))
2320e76b 5070
beebe431
RW		 5071 ;; The 'configure' script doesn't recognize things like
5072 ;; '--enable-fast-install'.
5073 (zero? (system*
5074 "./configure"
5075 (string-append "--build-prefix=" (getcwd) "/build")
5076 (string-append "--prefix=" (assoc-ref outputs "out"))
5077 (string-append "--debug")
5078 (string-append "--with-fuse-prefix="
5079 (assoc-ref inputs "fuse"))
5080 (string-append "--with-magic-prefix="
5081 (assoc-ref inputs "libmagic"))
5082 ;; TODO: building with libxml2 fails with linker errors
5083 ;; (string-append "--with-xml2-prefix="
5084 ;; (assoc-ref inputs "libxml2"))
5085 (string-append "--with-ncbi-vdb-sources="
5086 (assoc-ref inputs "ncbi-vdb"))
5087 (string-append "--with-ncbi-vdb-build="
5088 (assoc-ref inputs "ncbi-vdb"))
5089 (string-append "--with-ngs-sdk-prefix="
5090 (assoc-ref inputs "ngs-sdk"))
5091 (string-append "--with-hdf5-prefix="
1108a920
RW		 5092 (assoc-ref inputs "hdf5"))))))
5093 ;; This version of sra-tools fails to build with glibc because of a
5094 ;; naming conflict. glibc-2.25/include/bits/mathcalls.h already
5095 ;; contains a definition of "canonicalize", so we rename it.
5096 ;;
5097 ;; See upstream bug report:
5098 ;; https://github.com/ncbi/sra-tools/issues/67
5099 (add-after 'unpack 'patch-away-glibc-conflict
5100 (lambda _
5101 (substitute* "tools/bam-loader/bam.c"
5102 (("canonicalize\\(" line)
5103 (string-append "sra_tools_" line)))
5104 #t)))))
51c64999
RW		 5105 (native-inputs `(("perl" ,perl)))
5106 (inputs
5107 `(("ngs-sdk" ,ngs-sdk)
5108 ("ncbi-vdb" ,ncbi-vdb)
5109 ("libmagic" ,file)
5110 ("fuse" ,fuse)
5111 ("hdf5" ,hdf5)
5112 ("zlib" ,zlib)))
5113 (home-page "http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software")
5114 (synopsis "Tools and libraries for reading and writing sequencing data")
5115 (description
5116 "The SRA Toolkit from NCBI is a collection of tools and libraries for
5117reading of sequencing files from the Sequence Read Archive (SRA) database and
5118writing files into the .sra format.")
5119 (license license:public-domain)))
5120
d3517eda
RW		 5121(define-public seqan
5122 (package
5123 (name "seqan")
5124 (version "1.4.2")
5125 (source (origin
5126 (method url-fetch)
5127 (uri (string-append "http://packages.seqan.de/seqan-library/"
5128 "seqan-library-" version ".tar.bz2"))
5129 (sha256
5130 (base32
5131 "05s3wrrwn50f81aklfm65i4a749zag1vr8z03k21xm0pdxy47yvp"))))
5132 ;; The documentation is 7.8MB and the includes are 3.6MB heavy, so it
5133 ;; makes sense to split the outputs.
5134 (outputs '("out" "doc"))
5135 (build-system trivial-build-system)
5136 (arguments
5137 `(#:modules ((guix build utils))
5138 #:builder
5139 (begin
5140 (use-modules (guix build utils))
5141 (let ((tar (assoc-ref %build-inputs "tar"))
5142 (bzip (assoc-ref %build-inputs "bzip2"))
5143 (out (assoc-ref %outputs "out"))
5144 (doc (assoc-ref %outputs "doc")))
5145 (setenv "PATH" (string-append tar "/bin:" bzip "/bin"))
5146 (system* "tar" "xvf" (assoc-ref %build-inputs "source"))
5147 (chdir (string-append "seqan-library-" ,version))
5148 (copy-recursively "include" (string-append out "/include"))
5149 (copy-recursively "share" (string-append doc "/share"))))))
5150 (native-inputs
5151 `(("source" ,source)
5152 ("tar" ,tar)
5153 ("bzip2" ,bzip2)))
5154 (home-page "http://www.seqan.de")
5155 (synopsis "Library for nucleotide sequence analysis")
5156 (description
5157 "SeqAn is a C++ library of efficient algorithms and data structures for
5158the analysis of sequences with the focus on biological data. It contains
5159algorithms and data structures for string representation and their
5160manipulation, online and indexed string search, efficient I/O of
5161bioinformatics file formats, sequence alignment, and more.")
5162 (license license:bsd-3)))
ce7155d5 5163
d708b7a9
BW		 5164(define-public seqmagick
5165 (package
5166 (name "seqmagick")
5167 (version "0.6.1")
5168 (source
5169 (origin
5170 (method url-fetch)
5171 (uri (string-append
5172 "https://pypi.python.org/packages/source/s/seqmagick/seqmagick-"
5173 version ".tar.gz"))
5174 (sha256
5175 (base32
5176 "0cgn477n74gsl4qdaakrrhi953kcsd4q3ivk2lr18x74s3g4ma1d"))))
5177 (build-system python-build-system)
5178 (arguments
5179 ;; python2 only, see https://github.com/fhcrc/seqmagick/issues/56
5180 `(#:python ,python-2
5181 #:phases
5182 (modify-phases %standard-phases
5183 ;; Current test in setup.py does not work as of 0.6.1,
5184 ;; so use nose to run tests instead for now. See
5185 ;; https://github.com/fhcrc/seqmagick/issues/55
5186 (replace 'check (lambda _ (zero? (system* "nosetests")))))))
5187 (inputs
7bba95b7
BW		 5188 ;; biopython-1.66 is required due to
5189 ;; https://github.com/fhcrc/seqmagick/issues/59
5190 ;; When that issue is resolved the 'python2-biopython-1.66' package
5191 ;; should be removed.
5192 `(("python-biopython" ,python2-biopython-1.66)))
d708b7a9 5193 (native-inputs
f3b98f4f 5194 `(("python-nose" ,python2-nose)))
7bf837fd 5195 (home-page "https://github.com/fhcrc/seqmagick")
d708b7a9
BW		 5196 (synopsis "Tools for converting and modifying sequence files")
5197 (description
5198 "Bioinformaticians often have to convert sequence files between formats
5199and do little manipulations on them, and it's not worth writing scripts for
5200that. Seqmagick is a utility to expose the file format conversion in
5201BioPython in a convenient way. Instead of having a big mess of scripts, there
5202is one that takes arguments.")
5203 (license license:gpl3)))
5204
66daf78c
BW		 5205(define-public seqtk
5206 (package
5207 (name "seqtk")
5208 (version "1.2")
5209 (source (origin
5210 (method url-fetch)
5211 (uri (string-append
5212 "https://github.com/lh3/seqtk/archive/v"
5213 version ".tar.gz"))
5214 (file-name (string-append name "-" version ".tar.gz"))
5215 (sha256
5216 (base32
5217 "0ywdyzpmfiz2wp6ampbzqg4y8bj450nfgqarpamg045b8mk32lxx"))
5218 (modules '((guix build utils)))
5219 (snippet
5220 '(begin
5221 ;; Remove extraneous header files, as is done in the seqtk
5222 ;; master branch.
5223 (for-each (lambda (file) (delete-file file))
5224 (list "ksort.h" "kstring.h" "kvec.h"))
5225 #t))))
5226 (build-system gnu-build-system)
5227 (arguments
5228 `(#:phases
5229 (modify-phases %standard-phases
5230 (delete 'configure)
5231 (replace 'check
5232 ;; There are no tests, so we just run a sanity check.
5233 (lambda _ (zero? (system* "./seqtk" "seq"))))
5234 (replace 'install
5235 (lambda* (#:key outputs #:allow-other-keys)
5236 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5237 (install-file "seqtk" bin)))))))
5238 (inputs
5239 `(("zlib" ,zlib)))
5240 (home-page "https://github.com/lh3/seqtk")
5241 (synopsis "Toolkit for processing biological sequences in FASTA/Q format")
5242 (description
5243 "Seqtk is a fast and lightweight tool for processing sequences in the
5244FASTA or FASTQ format. It parses both FASTA and FASTQ files which can be
5245optionally compressed by gzip.")
5246 (license license:expat)))
5247
5f7e17be
BW		 5248(define-public snap-aligner
5249 (package
5250 (name "snap-aligner")
5251 (version "1.0beta.18")
5252 (source (origin
5253 (method url-fetch)
5254 (uri (string-append
5255 "https://github.com/amplab/snap/archive/v"
5256 version ".tar.gz"))
5257 (file-name (string-append name "-" version ".tar.gz"))
5258 (sha256
5259 (base32
5260 "1vnsjwv007k1fl1q7d681kbwn6bc66cgw6h16hym6gvyy71qv2ly"))))
5261 (build-system gnu-build-system)
5262 (arguments
5263 '(#:phases
5264 (modify-phases %standard-phases
5265 (delete 'configure)
5266 (replace 'check (lambda _ (zero? (system* "./unit_tests"))))
5267 (replace 'install
5268 (lambda* (#:key outputs #:allow-other-keys)
5269 (let* ((out (assoc-ref outputs "out"))
5270 (bin (string-append out "/bin")))
5f7e17be
BW		 5271 (install-file "snap-aligner" bin)
5272 (install-file "SNAPCommand" bin)
5273 #t))))))
5274 (native-inputs
5275 `(("zlib" ,zlib)))
5276 (home-page "http://snap.cs.berkeley.edu/")
5277 (synopsis "Short read DNA sequence aligner")
5278 (description
5279 "SNAP is a fast and accurate aligner for short DNA reads. It is
5280optimized for modern read lengths of 100 bases or higher, and takes advantage
5281of these reads to align data quickly through a hash-based indexing scheme.")
3e6fdd5f
EF		 5282 ;; 32-bit systems are not supported by the unpatched code.
5283 ;; Following the bug reports https://github.com/amplab/snap/issues/68 and
5284 ;; https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=812378 we see that
5285 ;; systems without a lot of memory cannot make good use of this program.
5286 (supported-systems '("x86_64-linux"))
5f7e17be
BW		 5287 (license license:asl2.0)))
5288
bcadaf00
BW		 5289(define-public sortmerna
5290 (package
5291 (name "sortmerna")
849485f5 5292 (version "2.1b")
bcadaf00
BW		 5293 (source
5294 (origin
5295 (method url-fetch)
5296 (uri (string-append
5297 "https://github.com/biocore/sortmerna/archive/"
5298 version ".tar.gz"))
5299 (file-name (string-append name "-" version ".tar.gz"))
5300 (sha256
5301 (base32
849485f5 5302 "1ghaghvd82af9j5adavxh77g7hm247d1r69m3fbi6f1jdivj5ldk"))))
bcadaf00
BW		 5303 (build-system gnu-build-system)
5304 (outputs '("out" ;for binaries
5305 "db")) ;for sequence databases
5306 (arguments
5307 `(#:phases
5308 (modify-phases %standard-phases
5309 (replace 'install
5310 (lambda* (#:key outputs #:allow-other-keys)
5311 (let* ((out (assoc-ref outputs "out"))
5312 (bin (string-append out "/bin"))
5313 (db (assoc-ref outputs "db"))
5314 (share
5315 (string-append db "/share/sortmerna/rRNA_databases")))
5316 (install-file "sortmerna" bin)
5317 (install-file "indexdb_rna" bin)
5318 (for-each (lambda (file)
5319 (install-file file share))
5320 (find-files "rRNA_databases" ".*fasta"))
5321 #t))))))
849485f5
BW		 5322 (inputs
5323 `(("zlib" ,zlib)))
bcadaf00
BW		 5324 (home-page "http://bioinfo.lifl.fr/RNA/sortmerna")
5325 (synopsis "Biological sequence analysis tool for NGS reads")
5326 (description
5327 "SortMeRNA is a biological sequence analysis tool for filtering, mapping
5328and operational taxonomic unit (OTU) picking of next generation
5329sequencing (NGS) reads. The core algorithm is based on approximate seeds and
5330allows for fast and sensitive analyses of nucleotide sequences. The main
5331application of SortMeRNA is filtering rRNA from metatranscriptomic data.")
2324541b
EF		 5332 ;; The source includes x86 specific code
5333 (supported-systems '("x86_64-linux" "i686-linux"))
bcadaf00
BW		 5334 (license license:lgpl3)))
5335
ce7155d5
RW		 5336(define-public star
5337 (package
5338 (name "star")
939b6b11 5339 (version "2.5.3a")
ce7155d5
RW		 5340 (source (origin
5341 (method url-fetch)
3062d750
RW		 5342 (uri (string-append "https://github.com/alexdobin/STAR/archive/"
5343 version ".tar.gz"))
5344 (file-name (string-append name "-" version ".tar.gz"))
ce7155d5
RW		 5345 (sha256
5346 (base32
939b6b11 5347 "013wirlz8lllgjyagl48l75n1isxyabqb3sj7qlsl0x1rmvqw99a"))
ce7155d5
RW		 5348 (modules '((guix build utils)))
5349 (snippet
3062d750
RW		 5350 '(begin
5351 (substitute* "source/Makefile"
5352 (("/bin/rm") "rm"))
5353 ;; Remove pre-built binaries and bundled htslib sources.
5354 (delete-file-recursively "bin/MacOSX_x86_64")
5355 (delete-file-recursively "bin/Linux_x86_64")
c9e9154e 5356 (delete-file-recursively "bin/Linux_x86_64_static")
3062d750
RW		 5357 (delete-file-recursively "source/htslib")
5358 #t))))
ce7155d5
RW		 5359 (build-system gnu-build-system)
5360 (arguments
5361 '(#:tests? #f ;no check target
5362 #:make-flags '("STAR")
5363 #:phases
c0266e8d
RW		 5364 (modify-phases %standard-phases
5365 (add-after 'unpack 'enter-source-dir
5366 (lambda _ (chdir "source") #t))
3062d750
RW		 5367 (add-after 'enter-source-dir 'do-not-use-bundled-htslib
5368 (lambda _
5369 (substitute* "Makefile"
5370 (("(Depend.list: \\$\\(SOURCES\\) parametersDefault\\.xxd) htslib"
5371 _ prefix) prefix))
5372 (substitute* '("BAMfunctions.cpp"
5373 "signalFromBAM.h"
5374 "bam_cat.h"
5375 "bam_cat.c"
5376 "STAR.cpp"
5377 "bamRemoveDuplicates.cpp")
5378 (("#include \"htslib/([^\"]+\\.h)\"" _ header)
5379 (string-append "#include <" header ">")))
5380 (substitute* "IncludeDefine.h"
5381 (("\"htslib/(htslib/[^\"]+.h)\"" _ header)
5382 (string-append "<" header ">")))
5383 #t))
c0266e8d
RW		 5384 (replace 'install
5385 (lambda* (#:key outputs #:allow-other-keys)
5386 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5387 (install-file "STAR" bin))
5388 #t))
5389 (delete 'configure))))
ce7155d5 5390 (native-inputs
9fc513ad 5391 `(("xxd" ,xxd)))
ce7155d5 5392 (inputs
3062d750
RW		 5393 `(("htslib" ,htslib)
5394 ("zlib" ,zlib)))
ce7155d5
RW		 5395 (home-page "https://github.com/alexdobin/STAR")
5396 (synopsis "Universal RNA-seq aligner")
5397 (description
5398 "The Spliced Transcripts Alignment to a Reference (STAR) software is
5399based on a previously undescribed RNA-seq alignment algorithm that uses
5400sequential maximum mappable seed search in uncompressed suffix arrays followed
5401by seed clustering and stitching procedure. In addition to unbiased de novo
5402detection of canonical junctions, STAR can discover non-canonical splices and
5403chimeric (fusion) transcripts, and is also capable of mapping full-length RNA
5404sequences.")
c11f79a4
BW		 5405 ;; Only 64-bit systems are supported according to the README.
5406 (supported-systems '("x86_64-linux" "mips64el-linux"))
ce7155d5
RW		 5407 ;; STAR is licensed under GPLv3 or later; htslib is MIT-licensed.
5408 (license license:gpl3+)))
de07c0db 5409
dbf4ed7c
RW		 5410(define-public subread
5411 (package
5412 (name "subread")
c180533b 5413 (version "1.5.1")
dbf4ed7c
RW		 5414 (source (origin
5415 (method url-fetch)
de67e922
LF		 5416 (uri (string-append "mirror://sourceforge/subread/subread-"
5417 version "/subread-" version "-source.tar.gz"))
dbf4ed7c
RW		 5418 (sha256
5419 (base32
c180533b 5420 "0gn5zhbvllks0mmdg3qlmsbg91p2mpdc2wixwfqpi85yzfrh8hcy"))))
dbf4ed7c
RW		 5421 (build-system gnu-build-system)
5422 (arguments
5423 `(#:tests? #f ;no "check" target
104c1986
RW		 5424 ;; The CC and CCFLAGS variables are set to contain a lot of x86_64
5425 ;; optimizations by default, so we override these flags such that x86_64
5426 ;; flags are only added when the build target is an x86_64 system.
5427 #:make-flags
5428 (list (let ((system ,(or (%current-target-system)
5429 (%current-system)))
5430 (flags '("-ggdb" "-fomit-frame-pointer"
5431 "-ffast-math" "-funroll-loops"
5432 "-fmessage-length=0"
5433 "-O9" "-Wall" "-DMAKE_FOR_EXON"
5434 "-DMAKE_STANDALONE"
5435 "-DSUBREAD_VERSION=\\\"${SUBREAD_VERSION}\\\""))
5436 (flags64 '("-mmmx" "-msse" "-msse2" "-msse3")))
5437 (if (string-prefix? "x86_64" system)
5438 (string-append "CCFLAGS=" (string-join (append flags flags64)))
5439 (string-append "CCFLAGS=" (string-join flags))))
5440 "-f" "Makefile.Linux"
5441 "CC=gcc ${CCFLAGS}")
dbf4ed7c 5442 #:phases
dc1d3cde
KK		 5443 (modify-phases %standard-phases
5444 (add-after 'unpack 'enter-dir
5445 (lambda _ (chdir "src") #t))
5446 (replace 'install
5447 (lambda* (#:key outputs #:allow-other-keys)
5448 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5449 (mkdir-p bin)
5450 (copy-recursively "../bin" bin))))
dbf4ed7c 5451 ;; no "configure" script
dc1d3cde 5452 (delete 'configure))))
dbf4ed7c
RW		 5453 (inputs `(("zlib" ,zlib)))
5454 (home-page "http://bioinf.wehi.edu.au/subread-package/")
5455 (synopsis "Tool kit for processing next-gen sequencing data")
5456 (description
5457 "The subread package contains the following tools: subread aligner, a
5458general-purpose read aligner; subjunc aligner: detecting exon-exon junctions
5459and mapping RNA-seq reads; featureCounts: counting mapped reads for genomic
5460features; exactSNP: a SNP caller that discovers SNPs by testing signals
5461against local background noises.")
5462 (license license:gpl3+)))
5463
d15d981e
RW		 5464(define-public stringtie
5465 (package
5466 (name "stringtie")
5467 (version "1.2.1")
5468 (source (origin
5469 (method url-fetch)
5470 (uri (string-append "http://ccb.jhu.edu/software/stringtie/dl/"
5471 "stringtie-" version ".tar.gz"))
5472 (sha256
5473 (base32
5474 "1cqllsc1maq4kh92isi8yadgzbmnf042hlnalpk3y59aph1z3bfz"))
5475 (modules '((guix build utils)))
5476 (snippet
5477 '(begin
5478 (delete-file-recursively "samtools-0.1.18")
5479 #t))))
5480 (build-system gnu-build-system)
5481 (arguments
5482 `(#:tests? #f ;no test suite
5483 #:phases
5484 (modify-phases %standard-phases
5485 ;; no configure script
5486 (delete 'configure)
5487 (add-before 'build 'use-system-samtools
5488 (lambda _
5489 (substitute* "Makefile"
5490 (("stringtie: \\$\\{BAM\\}/libbam\\.a")
5491 "stringtie: "))
5492 (substitute* '("gclib/GBam.h"
5493 "gclib/GBam.cpp")
5494 (("#include \"(bam|sam|kstring).h\"" _ header)
5495 (string-append "#include <samtools/" header ".h>")))
5496 #t))
0d2c0562
RW		 5497 (add-after 'unpack 'remove-duplicate-typedef
5498 (lambda _
5499 ;; This typedef conflicts with the typedef in
5500 ;; glibc-2.25/include/bits/types.h
5501 (substitute* "gclib/GThreads.h"
5502 (("typedef long long __intmax_t;") ""))
5503 #t))
d15d981e
RW		 5504 (replace 'install
5505 (lambda* (#:key outputs #:allow-other-keys)
5506 (let ((bin (string-append (assoc-ref outputs "out") "/bin/")))
5507 (install-file "stringtie" bin)
5508 #t))))))
5509 (inputs
5510 `(("samtools" ,samtools-0.1)
5511 ("zlib" ,zlib)))
5512 (home-page "http://ccb.jhu.edu/software/stringtie/")
5513 (synopsis "Transcript assembly and quantification for RNA-Seq data")
5514 (description
5515 "StringTie is a fast and efficient assembler of RNA-Seq sequence
5516alignments into potential transcripts. It uses a novel network flow algorithm
5517as well as an optional de novo assembly step to assemble and quantitate
5518full-length transcripts representing multiple splice variants for each gene
5519locus. Its input can include not only the alignments of raw reads used by
5520other transcript assemblers, but also alignments of longer sequences that have
5521been assembled from those reads. To identify differentially expressed genes
5522between experiments, StringTie's output can be processed either by the
5523Cuffdiff or Ballgown programs.")
5524 (license license:artistic2.0)))
5525
ad0ae297
BW		 5526(define-public taxtastic
5527 (package
5528 (name "taxtastic")
3cbfc149 5529 (version "0.6.4")
ad0ae297
BW		 5530 (source (origin
5531 (method url-fetch)
3cbfc149 5532 (uri (pypi-uri "taxtastic" version))
ad0ae297
BW		 5533 (sha256
5534 (base32
3cbfc149 5535 "0s79z8kfl853x7l4h8ms05k31q87aw62nrchlk20w9n227j35929"))))
ad0ae297
BW		 5536 (build-system python-build-system)
5537 (arguments
5538 `(#:python ,python-2
5539 #:phases
5540 (modify-phases %standard-phases
5541 (replace 'check
5542 (lambda _
5543 (zero? (system* "python" "-m" "unittest" "discover" "-v")))))))
5544 (propagated-inputs
5545 `(("python-sqlalchemy" ,python2-sqlalchemy)
5546 ("python-decorator" ,python2-decorator)
5547 ("python-biopython" ,python2-biopython)
5548 ("python-pandas" ,python2-pandas)))
5549 (home-page "https://github.com/fhcrc/taxtastic")
5550 (synopsis "Tools for taxonomic naming and annotation")
5551 (description
5552 "Taxtastic is software written in python used to build and maintain
5553reference packages i.e. collections of reference trees, reference alignments,
5554profiles, and associated taxonomic information.")
5555 (license license:gpl3+)))
5556
de07c0db
RW		 5557(define-public vcftools
5558 (package
5559 (name "vcftools")
f4322542 5560 (version "0.1.15")
de07c0db
RW		 5561 (source (origin
5562 (method url-fetch)
5563 (uri (string-append
9b36e256
RJ		 5564 "https://github.com/vcftools/vcftools/releases/download/v"
5565 version "/vcftools-" version ".tar.gz"))
de07c0db
RW		 5566 (sha256
5567 (base32
f4322542 5568 "1qw30c45wihgy632rbz4rh3njnwj4msj46l1rsgdhyg6bgypmr1i"))))
de07c0db
RW		 5569 (build-system gnu-build-system)
5570 (arguments
5571 `(#:tests? #f ; no "check" target
5572 #:make-flags (list
7c3958e1 5573 "CFLAGS=-O2" ; override "-m64" flag
de07c0db
RW		 5574 (string-append "PREFIX=" (assoc-ref %outputs "out"))
5575 (string-append "MANDIR=" (assoc-ref %outputs "out")
9b36e256
RJ		 5576 "/share/man/man1"))))
5577 (native-inputs
5578 `(("pkg-config" ,pkg-config)))
de07c0db
RW		 5579 (inputs
5580 `(("perl" ,perl)
5581 ("zlib" ,zlib)))
9b36e256 5582 (home-page "https://vcftools.github.io/")
de07c0db
RW		 5583 (synopsis "Tools for working with VCF files")
5584 (description
5585 "VCFtools is a program package designed for working with VCF files, such
5586as those generated by the 1000 Genomes Project. The aim of VCFtools is to
5587provide easily accessible methods for working with complex genetic variation
5588data in the form of VCF files.")
5589 ;; The license is declared as LGPLv3 in the README and
9b36e256 5590 ;; at https://vcftools.github.io/license.html
de07c0db 5591 (license license:lgpl3)))
9c38b540 5592
35aa90a1
RW		 5593(define-public infernal
5594 (package
5595 (name "infernal")
5596 (version "1.1.2")
5597 (source (origin
5598 (method url-fetch)
5599 (uri (string-append "http://eddylab.org/software/infernal/"
5600 "infernal-" version ".tar.gz"))
5601 (sha256
5602 (base32
5603 "0sr2hiz3qxfwqpz3whxr6n82p3x27336v3f34iqznp10hks2935c"))))
5604 (build-system gnu-build-system)
5605 (native-inputs
5606 `(("perl" ,perl))) ; for tests
5607 (home-page "http://eddylab.org/infernal/")
5608 (synopsis "Inference of RNA alignments")
5609 (description "Infernal (\"INFERence of RNA ALignment\") is a tool for
5610searching DNA sequence databases for RNA structure and sequence similarities.
5611It is an implementation of a special case of profile stochastic context-free
5612grammars called @dfn{covariance models} (CMs). A CM is like a sequence
5613profile, but it scores a combination of sequence consensus and RNA secondary
5614structure consensus, so in many cases, it is more capable of identifying RNA
5615homologs that conserve their secondary structure more than their primary
5616sequence.")
48409ef2
EF		 5617 ;; Infernal 1.1.2 requires VMX or SSE capability for parallel instructions.
5618 (supported-systems '("i686-linux" "x86_64-linux"))
35aa90a1
RW		 5619 (license license:bsd-3)))
5620
b91cfa22
RW		 5621(define-public r-centipede
5622 (package
5623 (name "r-centipede")
5624 (version "1.2")
5625 (source (origin
5626 (method url-fetch)
5627 (uri (string-append "http://download.r-forge.r-project.org/"
5628 "src/contrib/CENTIPEDE_" version ".tar.gz"))
5629 (sha256
5630 (base32
5631 "1hsx6qgwr0i67fhy9257zj7s0ppncph2hjgbia5nn6nfmj0ax6l9"))))
5632 (build-system r-build-system)
5633 (home-page "http://centipede.uchicago.edu/")
5634 (synopsis "Predict transcription factor binding sites")
5635 (description
5636 "CENTIPEDE applies a hierarchical Bayesian mixture model to infer regions
5637of the genome that are bound by particular transcription factors. It starts
5638by identifying a set of candidate binding sites, and then aims to classify the
5639sites according to whether each site is bound or not bound by a transcription
5640factor. CENTIPEDE is an unsupervised learning algorithm that discriminates
5641between two different types of motif instances using as much relevant
5642information as possible.")
5643 (license (list license:gpl2+ license:gpl3+))))
5644
7b3df1e5
BW		 5645(define-public r-vegan
5646 (package
5647 (name "r-vegan")
db2e4386 5648 (version "2.4-4")
7b3df1e5
BW		 5649 (source
5650 (origin
5651 (method url-fetch)
5652 (uri (cran-uri "vegan" version))
5653 (sha256
5654 (base32
db2e4386 5655 "1n57dzv2aid6iqd9fkqik401sidqanhzsawyak94qbiyh6dbd1x9"))))
7b3df1e5 5656 (build-system r-build-system)
7b3df1e5 5657 (native-inputs
db2e4386 5658 `(("gfortran" ,gfortran)))
7b3df1e5
BW		 5659 (propagated-inputs
5660 `(("r-cluster" ,r-cluster)
5661 ("r-lattice" ,r-lattice)
aeb64f3c 5662 ("r-mass" ,r-mass)
7b3df1e5
BW		 5663 ("r-mgcv" ,r-mgcv)
5664 ("r-permute" ,r-permute)))
5665 (home-page "https://cran.r-project.org/web/packages/vegan")
5666 (synopsis "Functions for community ecology")
5667 (description
5668 "The vegan package provides tools for descriptive community ecology. It
5669has most basic functions of diversity analysis, community ordination and
5670dissimilarity analysis. Most of its multivariate tools can be used for other
5671data types as well.")
5672 (license license:gpl2+)))
5673
8c6de588
RW		 5674(define-public r-annotate
5675 (package
5676 (name "r-annotate")
dbab99b6 5677 (version "1.54.0")
8c6de588
RW		 5678 (source
5679 (origin
5680 (method url-fetch)
5681 (uri (bioconductor-uri "annotate" version))
5682 (sha256
5683 (base32
dbab99b6 5684 "03hmbvp3i6lvd307fqdg7akxi2qp322rlky3bzw0zccgm0i0221g"))))
8c6de588
RW		 5685 (build-system r-build-system)
5686 (propagated-inputs
5687 `(("r-annotationdbi" ,r-annotationdbi)
5688 ("r-biobase" ,r-biobase)
5689 ("r-biocgenerics" ,r-biocgenerics)
5690 ("r-dbi" ,r-dbi)
d0f0579e 5691 ("r-rcurl" ,r-rcurl)
8c6de588
RW		 5692 ("r-xml" ,r-xml)
5693 ("r-xtable" ,r-xtable)))
5694 (home-page
5713bbf1 5695 "https://bioconductor.org/packages/annotate")
8c6de588 5696 (synopsis "Annotation for microarrays")
d1e4ad1b 5697 (description "This package provides R environments for the annotation of
8c6de588
RW		 5698microarrays.")
5699 (license license:artistic2.0)))
5700
07a664cd
RW		 5701(define-public r-geneplotter
5702 (package
5703 (name "r-geneplotter")
97a50e4f 5704 (version "1.54.0")
07a664cd
RW		 5705 (source
5706 (origin
5707 (method url-fetch)
5708 (uri (bioconductor-uri "geneplotter" version))
5709 (sha256
5710 (base32
97a50e4f 5711 "0a0ajns21db5rrjl16bq6wawggsnxr00fg184pc38nmfghv4z4b6"))))
07a664cd
RW		 5712 (build-system r-build-system)
5713 (propagated-inputs
5714 `(("r-annotate" ,r-annotate)
5715 ("r-annotationdbi" ,r-annotationdbi)
5716 ("r-biobase" ,r-biobase)
5717 ("r-biocgenerics" ,r-biocgenerics)
5718 ("r-lattice" ,r-lattice)
5719 ("r-rcolorbrewer" ,r-rcolorbrewer)))
5713bbf1 5720 (home-page "https://bioconductor.org/packages/geneplotter")
07a664cd
RW		 5721 (synopsis "Graphics functions for genomic data")
5722 (description
5723 "This package provides functions for plotting genomic data.")
5724 (license license:artistic2.0)))
5725
2301fd3e
RW		 5726(define-public r-genefilter
5727 (package
5728 (name "r-genefilter")
b5bc3a39 5729 (version "1.58.0")
2301fd3e
RW		 5730 (source
5731 (origin
5732 (method url-fetch)
5733 (uri (bioconductor-uri "genefilter" version))
5734 (sha256
5735 (base32
b5bc3a39 5736 "0sf2hdi9nv6r83vn1y65m4jiba8pffddpj46d6yjn5rlsixplmqg"))))
2301fd3e
RW		 5737 (build-system r-build-system)
5738 (native-inputs
5739 `(("gfortran" ,gfortran)))
5740 (propagated-inputs
5741 `(("r-annotate" ,r-annotate)
5742 ("r-annotationdbi" ,r-annotationdbi)
5743 ("r-biobase" ,r-biobase)
aeb64f3c
RW		 5744 ("r-s4vectors" ,r-s4vectors)
5745 ("r-survival" ,r-survival)))
5713bbf1 5746 (home-page "https://bioconductor.org/packages/genefilter")
2301fd3e
RW		 5747 (synopsis "Filter genes from high-throughput experiments")
5748 (description
5749 "This package provides basic functions for filtering genes from
5750high-throughput sequencing experiments.")
5751 (license license:artistic2.0)))
5752
ad34f0ac
RW		 5753(define-public r-deseq2
5754 (package
5755 (name "r-deseq2")
f1915a5d 5756 (version "1.16.1")
ad34f0ac
RW		 5757 (source
5758 (origin
5759 (method url-fetch)
5760 (uri (bioconductor-uri "DESeq2" version))
5761 (sha256
5762 (base32
f1915a5d 5763 "01pvyljxkwazxl510v7h0971nx65iqd2bdkbdhw3xzind0n9pdvq"))))
ad34f0ac
RW		 5764 (properties `((upstream-name . "DESeq2")))
5765 (build-system r-build-system)
ad34f0ac
RW		 5766 (propagated-inputs
5767 `(("r-biobase" ,r-biobase)
5768 ("r-biocgenerics" ,r-biocgenerics)
5769 ("r-biocparallel" ,r-biocparallel)
5770 ("r-genefilter" ,r-genefilter)
5771 ("r-geneplotter" ,r-geneplotter)
5772 ("r-genomicranges" ,r-genomicranges)
5773 ("r-ggplot2" ,r-ggplot2)
5774 ("r-hmisc" ,r-hmisc)
5775 ("r-iranges" ,r-iranges)
5776 ("r-locfit" ,r-locfit)
5777 ("r-rcpp" ,r-rcpp)
5778 ("r-rcpparmadillo" ,r-rcpparmadillo)
5779 ("r-s4vectors" ,r-s4vectors)
5780 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5781 (home-page "https://bioconductor.org/packages/DESeq2")
ad34f0ac
RW		 5782 (synopsis "Differential gene expression analysis")
5783 (description
5784 "This package provides functions to estimate variance-mean dependence in
5785count data from high-throughput nucleotide sequencing assays and test for
5786differential expression based on a model using the negative binomial
5787distribution.")
5788 (license license:lgpl3+)))
5789
86763fdd
RW		 5790(define-public r-dexseq
5791 (package
5792 (name "r-dexseq")
5793 (version "1.22.0")
5794 (source
5795 (origin
5796 (method url-fetch)
5797 (uri (bioconductor-uri "DEXSeq" version))
5798 (sha256
5799 (base32
5800 "085aqk1wlzzqcqcqhvz74y099kr2ln5dwdxd3rl6zan806mgwahg"))))
5801 (properties `((upstream-name . "DEXSeq")))
5802 (build-system r-build-system)
5803 (propagated-inputs
5804 `(("r-annotationdbi" ,r-annotationdbi)
5805 ("r-biobase" ,r-biobase)
5806 ("r-biocgenerics" ,r-biocgenerics)
5807 ("r-biocparallel" ,r-biocparallel)
5808 ("r-biomart" ,r-biomart)
5809 ("r-deseq2" ,r-deseq2)
5810 ("r-genefilter" ,r-genefilter)
5811 ("r-geneplotter" ,r-geneplotter)
5812 ("r-genomicranges" ,r-genomicranges)
5813 ("r-hwriter" ,r-hwriter)
5814 ("r-iranges" ,r-iranges)
5815 ("r-rcolorbrewer" ,r-rcolorbrewer)
5816 ("r-rsamtools" ,r-rsamtools)
5817 ("r-s4vectors" ,r-s4vectors)
5818 ("r-statmod" ,r-statmod)
5819 ("r-stringr" ,r-stringr)
5820 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 5821 (home-page "https://bioconductor.org/packages/DEXSeq")
86763fdd
RW		 5822 (synopsis "Inference of differential exon usage in RNA-Seq")
5823 (description
5824 "This package is focused on finding differential exon usage using RNA-seq
5825exon counts between samples with different experimental designs. It provides
5826functions that allows the user to make the necessary statistical tests based
5827on a model that uses the negative binomial distribution to estimate the
5828variance between biological replicates and generalized linear models for
5829testing. The package also provides functions for the visualization and
5830exploration of the results.")
5831 (license license:gpl3+)))
5832
e8163773
RW		 5833(define-public r-annotationforge
5834 (package
5835 (name "r-annotationforge")
fd215c29 5836 (version "1.18.1")
e8163773
RW		 5837 (source
5838 (origin
5839 (method url-fetch)
5840 (uri (bioconductor-uri "AnnotationForge" version))
5841 (sha256
5842 (base32
fd215c29 5843 "1366qvykd9cpcvwgc5g9mm9adw9rxw6p4814dd6l5fyb0pwpmysx"))))
e8163773
RW		 5844 (properties
5845 `((upstream-name . "AnnotationForge")))
5846 (build-system r-build-system)
5847 (propagated-inputs
5848 `(("r-annotationdbi" ,r-annotationdbi)
5849 ("r-biobase" ,r-biobase)
5850 ("r-biocgenerics" ,r-biocgenerics)
5851 ("r-dbi" ,r-dbi)
55cd914c 5852 ("r-rcurl" ,r-rcurl)
e8163773
RW		 5853 ("r-rsqlite" ,r-rsqlite)
5854 ("r-s4vectors" ,r-s4vectors)
5855 ("r-xml" ,r-xml)))
5713bbf1 5856 (home-page "https://bioconductor.org/packages/AnnotationForge")
e8163773
RW		 5857 (synopsis "Code for building annotation database packages")
5858 (description
5859 "This package provides code for generating Annotation packages and their
5860databases. Packages produced are intended to be used with AnnotationDbi.")
5861 (license license:artistic2.0)))
5862
cd9e7dc7
RW		 5863(define-public r-rbgl
5864 (package
5865 (name "r-rbgl")
8ab6f241 5866 (version "1.52.0")
cd9e7dc7
RW		 5867 (source
5868 (origin
5869 (method url-fetch)
5870 (uri (bioconductor-uri "RBGL" version))
5871 (sha256
5872 (base32
8ab6f241 5873 "11db6kvz453ypj9ds3xpjqzwrrjck84ijn4wlhkfyz2dzdgd5ryv"))))
cd9e7dc7
RW		 5874 (properties `((upstream-name . "RBGL")))
5875 (build-system r-build-system)
5876 (propagated-inputs `(("r-graph" ,r-graph)))
5713bbf1 5877 (home-page "https://www.bioconductor.org/packages/RBGL")
cd9e7dc7
RW		 5878 (synopsis "Interface to the Boost graph library")
5879 (description
5880 "This package provides a fairly extensive and comprehensive interface to
5881the graph algorithms contained in the Boost library.")
5882 (license license:artistic2.0)))
5883
ad740ff8
RW		 5884(define-public r-gseabase
5885 (package
5886 (name "r-gseabase")
c584a07f 5887 (version "1.38.0")
ad740ff8
RW		 5888 (source
5889 (origin
5890 (method url-fetch)
5891 (uri (bioconductor-uri "GSEABase" version))
5892 (sha256
5893 (base32
c584a07f 5894 "1c6i6g4fj3b8wjyxyygr7i3v8sxrq1ffb2bbicya5ah2gdaclfad"))))
ad740ff8
RW		 5895 (properties `((upstream-name . "GSEABase")))
5896 (build-system r-build-system)
5897 (propagated-inputs
5898 `(("r-annotate" ,r-annotate)
5899 ("r-annotationdbi" ,r-annotationdbi)
5900 ("r-biobase" ,r-biobase)
5901 ("r-biocgenerics" ,r-biocgenerics)
5902 ("r-graph" ,r-graph)
5903 ("r-xml" ,r-xml)))
5713bbf1 5904 (home-page "https://bioconductor.org/packages/GSEABase")
ad740ff8
RW		 5905 (synopsis "Gene set enrichment data structures and methods")
5906 (description
5907 "This package provides classes and methods to support @dfn{Gene Set
5908Enrichment Analysis} (GSEA).")
5909 (license license:artistic2.0)))
5910
1a1931f7
RW		 5911(define-public r-category
5912 (package
5913 (name "r-category")
bd4a0bab 5914 (version "2.42.1")
1a1931f7
RW		 5915 (source
5916 (origin
5917 (method url-fetch)
5918 (uri (bioconductor-uri "Category" version))
5919 (sha256
5920 (base32
bd4a0bab 5921 "1w186nhc85bglcgmbcrsdbb8l6rph21pl5kdwjqwkp0jnr9z0ifn"))))
1a1931f7
RW		 5922 (properties `((upstream-name . "Category")))
5923 (build-system r-build-system)
5924 (propagated-inputs
5925 `(("r-annotate" ,r-annotate)
5926 ("r-annotationdbi" ,r-annotationdbi)
5927 ("r-biobase" ,r-biobase)
5928 ("r-biocgenerics" ,r-biocgenerics)
5929 ("r-genefilter" ,r-genefilter)
5930 ("r-graph" ,r-graph)
5931 ("r-gseabase" ,r-gseabase)
5932 ("r-matrix" ,r-matrix)
5933 ("r-rbgl" ,r-rbgl)
5934 ("r-rsqlite" ,r-rsqlite)))
5713bbf1 5935 (home-page "https://bioconductor.org/packages/Category")
1a1931f7
RW		 5936 (synopsis "Category analysis")
5937 (description
5938 "This package provides a collection of tools for performing category
5939analysis.")
5940 (license license:artistic2.0)))
5941
89f40c5e
RW		 5942(define-public r-gostats
5943 (package
5944 (name "r-gostats")
95117734 5945 (version "2.42.0")
89f40c5e
RW		 5946 (source
5947 (origin
5948 (method url-fetch)
5949 (uri (bioconductor-uri "GOstats" version))
5950 (sha256
5951 (base32
95117734 5952 "0qvqjgfnd9ap4rikvyxa9p4dhcnccvkw8phzv88vghh6pq463d62"))))
89f40c5e
RW		 5953 (properties `((upstream-name . "GOstats")))
5954 (build-system r-build-system)
5955 (propagated-inputs
5956 `(("r-annotate" ,r-annotate)
5957 ("r-annotationdbi" ,r-annotationdbi)
5958 ("r-annotationforge" ,r-annotationforge)
5959 ("r-biobase" ,r-biobase)
5960 ("r-category" ,r-category)
5961 ("r-go-db" ,r-go-db)
5962 ("r-graph" ,r-graph)
5963 ("r-rbgl" ,r-rbgl)))
5713bbf1 5964 (home-page "https://bioconductor.org/packages/GOstats")
89f40c5e
RW		 5965 (synopsis "Tools for manipulating GO and microarrays")
5966 (description
5967 "This package provides a set of tools for interacting with GO and
5968microarray data. A variety of basic manipulation tools for graphs, hypothesis
5969testing and other simple calculations.")
5970 (license license:artistic2.0)))
5971
cb99d457
RW		 5972(define-public r-shortread
5973 (package
5974 (name "r-shortread")
34c276f7 5975 (version "1.34.0")
cb99d457
RW		 5976 (source
5977 (origin
5978 (method url-fetch)
5979 (uri (bioconductor-uri "ShortRead" version))
5980 (sha256
5981 (base32
34c276f7 5982 "0ayk3d5625ymb5g2gycq6banzqmyd642xrwjzhdshz2dwid7kly8"))))
cb99d457
RW		 5983 (properties `((upstream-name . "ShortRead")))
5984 (build-system r-build-system)
5985 (inputs
5986 `(("zlib" ,zlib)))
5987 (propagated-inputs
5988 `(("r-biobase" ,r-biobase)
5989 ("r-biocgenerics" ,r-biocgenerics)
5990 ("r-biocparallel" ,r-biocparallel)
5991 ("r-biostrings" ,r-biostrings)
5992 ("r-genomeinfodb" ,r-genomeinfodb)
5993 ("r-genomicalignments" ,r-genomicalignments)
5994 ("r-genomicranges" ,r-genomicranges)
5995 ("r-hwriter" ,r-hwriter)
5996 ("r-iranges" ,r-iranges)
5997 ("r-lattice" ,r-lattice)
5998 ("r-latticeextra" ,r-latticeextra)
5999 ("r-rsamtools" ,r-rsamtools)
6000 ("r-s4vectors" ,r-s4vectors)
6001 ("r-xvector" ,r-xvector)
6002 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 6003 (home-page "https://bioconductor.org/packages/ShortRead")
cb99d457
RW		 6004 (synopsis "FASTQ input and manipulation tools")
6005 (description
6006 "This package implements sampling, iteration, and input of FASTQ files.
6007It includes functions for filtering and trimming reads, and for generating a
6008quality assessment report. Data are represented as
6009@code{DNAStringSet}-derived objects, and easily manipulated for a diversity of
6010purposes. The package also contains legacy support for early single-end,
6011ungapped alignment formats.")
6012 (license license:artistic2.0)))
6013
7f903d73
RW		 6014(define-public r-systempiper
6015 (package
6016 (name "r-systempiper")
026dc7e4 6017 (version "1.10.0")
7f903d73
RW		 6018 (source
6019 (origin
6020 (method url-fetch)
6021 (uri (bioconductor-uri "systemPipeR" version))
6022 (sha256
6023 (base32
026dc7e4 6024 "0c3m5rq63ypv15yca97yag5d4vgd7xj9by2a4sd8z0pcmpajz0hw"))))
7f903d73
RW		 6025 (properties `((upstream-name . "systemPipeR")))
6026 (build-system r-build-system)
6027 (propagated-inputs
6028 `(("r-annotate" ,r-annotate)
6029 ("r-batchjobs" ,r-batchjobs)
6030 ("r-biocgenerics" ,r-biocgenerics)
6031 ("r-biostrings" ,r-biostrings)
6032 ("r-deseq2" ,r-deseq2)
6033 ("r-edger" ,r-edger)
6034 ("r-genomicfeatures" ,r-genomicfeatures)
6035 ("r-genomicranges" ,r-genomicranges)
6036 ("r-ggplot2" ,r-ggplot2)
6037 ("r-go-db" ,r-go-db)
6038 ("r-gostats" ,r-gostats)
6039 ("r-limma" ,r-limma)
6040 ("r-pheatmap" ,r-pheatmap)
6041 ("r-rjson" ,r-rjson)
6042 ("r-rsamtools" ,r-rsamtools)
6043 ("r-shortread" ,r-shortread)
6044 ("r-summarizedexperiment" ,r-summarizedexperiment)
6045 ("r-variantannotation" ,r-variantannotation)))
6046 (home-page "https://github.com/tgirke/systemPipeR")
6047 (synopsis "Next generation sequencing workflow and reporting environment")
6048 (description
6049 "This R package provides tools for building and running automated
6050end-to-end analysis workflows for a wide range of @dfn{next generation
6051sequence} (NGS) applications such as RNA-Seq, ChIP-Seq, VAR-Seq and Ribo-Seq.
6052Important features include a uniform workflow interface across different NGS
6053applications, automated report generation, and support for running both R and
6054command-line software, such as NGS aligners or peak/variant callers, on local
6055computers or compute clusters. Efficient handling of complex sample sets and
6056experimental designs is facilitated by a consistently implemented sample
6057annotation infrastructure.")
6058 (license license:artistic2.0)))
6059
684f29bd
RW		 6060(define-public r-grohmm
6061 (package
6062 (name "r-grohmm")
795a6045 6063 (version "1.10.0")
684f29bd
RW		 6064 (source
6065 (origin
6066 (method url-fetch)
6067 (uri (bioconductor-uri "groHMM" version))
6068 (sha256
6069 (base32
795a6045 6070 "16k1kp4sbhh0vp7dzywafq52csq42ksqfrqfy4bdv1qbd7536dpd"))))
684f29bd
RW		 6071 (properties `((upstream-name . "groHMM")))
6072 (build-system r-build-system)
6073 (propagated-inputs
6074 `(("r-genomeinfodb" ,r-genomeinfodb)
6075 ("r-genomicalignments" ,r-genomicalignments)
6076 ("r-genomicranges" ,r-genomicranges)
6077 ("r-iranges" ,r-iranges)
aeb64f3c 6078 ("r-mass" ,r-mass)
684f29bd
RW		 6079 ("r-rtracklayer" ,r-rtracklayer)
6080 ("r-s4vectors" ,r-s4vectors)))
6081 (home-page "https://github.com/Kraus-Lab/groHMM")
6082 (synopsis "GRO-seq analysis pipeline")
6083 (description
6084 "This package provides a pipeline for the analysis of GRO-seq data.")
6085 (license license:gpl3+)))
6086
f3cfe451
RW		 6087(define-public r-txdb-hsapiens-ucsc-hg19-knowngene
6088 (package
6089 (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
6090 (version "3.2.2")
6091 (source (origin
6092 (method url-fetch)
6093 ;; We cannot use bioconductor-uri here because this tarball is
6094 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 6095 (uri (string-append "https://bioconductor.org/packages/"
f3cfe451
RW		 6096 "release/data/annotation/src/contrib"
6097 "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
6098 version ".tar.gz"))
6099 (sha256
6100 (base32
6101 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
6102 (properties
6103 `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
6104 (build-system r-build-system)
6105 ;; As this package provides little more than a very large data file it
6106 ;; doesn't make sense to build substitutes.
6107 (arguments `(#:substitutable? #f))
6108 (propagated-inputs
6109 `(("r-genomicfeatures" ,r-genomicfeatures)))
6110 (home-page
5713bbf1 6111 "https://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
f3cfe451
RW		 6112 (synopsis "Annotation package for human genome in TxDb format")
6113 (description
6114 "This package provides an annotation database of Homo sapiens genome
6115data. It is derived from the UCSC hg19 genome and based on the \"knownGene\"
6116track. The database is exposed as a @code{TxDb} object.")
6117 (license license:artistic2.0)))
6118
325c039c
RJ		 6119(define-public r-sparql
6120 (package
6121 (name "r-sparql")
6122 (version "1.16")
6123 (source (origin
6124 (method url-fetch)
6125 (uri (cran-uri "SPARQL" version))
6126 (sha256
6127 (base32
6128 "0gak1q06yyhdmcxb2n3v0h9gr1vqd0viqji52wpw211qp6r6dcrc"))))
6129 (properties `((upstream-name . "SPARQL")))
6130 (build-system r-build-system)
6131 (propagated-inputs
6132 `(("r-rcurl" ,r-rcurl)
6133 ("r-xml" ,r-xml)))
6134 (home-page "http://cran.r-project.org/web/packages/SPARQL")
6135 (synopsis "SPARQL client for R")
6136 (description "This package provides an interface to use SPARQL to pose
6137SELECT or UPDATE queries to an end-point.")
6138 ;; The only license indication is found in the DESCRIPTION file,
6139 ;; which states GPL-3. So we cannot assume GPLv3+.
6140 (license license:gpl3)))
6141
a2950fa4
BW		 6142(define-public vsearch
6143 (package
6144 (name "vsearch")
03d6e9dd 6145 (version "2.5.2")
a2950fa4
BW		 6146 (source
6147 (origin
6148 (method url-fetch)
6149 (uri (string-append
6150 "https://github.com/torognes/vsearch/archive/v"
6151 version ".tar.gz"))
6152 (file-name (string-append name "-" version ".tar.gz"))
6153 (sha256
6154 (base32
03d6e9dd 6155 "1z10psrwhflchwzw1fvzwnsn59xglf94qrn7msj93fpnjgafvmks"))
206af46f 6156 (patches (search-patches "vsearch-unbundle-cityhash.patch"))
a2950fa4
BW		 6157 (snippet
6158 '(begin
206af46f
BW		 6159 ;; Remove bundled cityhash sources. The vsearch source is adjusted
6160 ;; for this in the patch.
cf6edaba
BW		 6161 (delete-file "src/city.h")
6162 (delete-file "src/citycrc.h")
6163 (delete-file "src/city.cc")
a2950fa4
BW		 6164 #t))))
6165 (build-system gnu-build-system)
6166 (arguments
6167 `(#:phases
6168 (modify-phases %standard-phases
d10092b8
KK		 6169 (add-after 'unpack 'autogen
6170 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
a2950fa4
BW		 6171 (inputs
6172 `(("zlib" ,zlib)
6173 ("bzip2" ,bzip2)
6174 ("cityhash" ,cityhash)))
6175 (native-inputs
6176 `(("autoconf" ,autoconf)
6177 ("automake" ,automake)))
6178 (synopsis "Sequence search tools for metagenomics")
6179 (description
6180 "VSEARCH supports DNA sequence searching, clustering, chimera detection,
6181dereplication, pairwise alignment, shuffling, subsampling, sorting and
6182masking. The tool takes advantage of parallelism in the form of SIMD
6183vectorization as well as multiple threads to perform accurate alignments at
6184high speed. VSEARCH uses an optimal global aligner (full dynamic programming
6185Needleman-Wunsch).")
6186 (home-page "https://github.com/torognes/vsearch")
6f04e515
BW		 6187 ;; vsearch uses non-portable SSE intrinsics so building fails on other
6188 ;; platforms.
6189 (supported-systems '("x86_64-linux"))
a2950fa4
BW		 6190 ;; Dual licensed; also includes public domain source.
6191 (license (list license:gpl3 license:bsd-2))))
6192
07837874
RW		 6193(define-public pardre
6194 (package
6195 (name "pardre")
7922ab8f
BW		 6196 ;; The source of 1.1.5 changed in place, so we append "-1" to the version.
6197 (version "1.1.5-1")
07837874
RW		 6198 (source
6199 (origin
6200 (method url-fetch)
6201 (uri (string-append "mirror://sourceforge/pardre/ParDRe-rel"
7922ab8f 6202 "1.1.5" ".tar.gz"))
07837874
RW		 6203 (sha256
6204 (base32
7922ab8f 6205 "17j73nc0viq4f6qj50nrndsrif5d6b71q8fl87m54psiv0ilns2b"))))
07837874
RW		 6206 (build-system gnu-build-system)
6207 (arguments
6208 `(#:tests? #f ; no tests included
6209 #:phases
6210 (modify-phases %standard-phases
6211 (delete 'configure)
6212 (replace 'install
6213 (lambda* (#:key outputs #:allow-other-keys)
6214 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
07837874
RW		 6215 (install-file "ParDRe" bin)
6216 #t))))))
6217 (inputs
6218 `(("openmpi" ,openmpi)
6219 ("zlib" ,zlib)))
6220 (synopsis "Parallel tool to remove duplicate DNA reads")
6221 (description
6222 "ParDRe is a parallel tool to remove duplicate genetic sequence reads.
6223Duplicate reads can be seen as identical or nearly identical sequences with
6224some mismatches. This tool lets users avoid the analysis of unnecessary
6225reads, reducing the time of subsequent procedures with the
6226dataset (e.g. assemblies, mappings, etc.). The tool is implemented with MPI
6227in order to exploit the parallel capabilities of multicore clusters. It is
6228faster than multithreaded counterparts (end of 2015) for the same number of
6229cores and, thanks to the message-passing technology, it can be executed on
6230clusters.")
6231 (home-page "https://sourceforge.net/projects/pardre/")
6232 (license license:gpl3+)))
6233
e4a44a6a
BW		 6234(define-public ruby-bio-kseq
6235 (package
6236 (name "ruby-bio-kseq")
6237 (version "0.0.2")
6238 (source
6239 (origin
6240 (method url-fetch)
6241 (uri (rubygems-uri "bio-kseq" version))
6242 (sha256
6243 (base32
6244 "1xyaha46khb5jc6wzkbf7040jagac49jbimn0vcrzid0j8jdikrz"))))
6245 (build-system ruby-build-system)
6246 (arguments
6247 `(#:test-target "spec"))
6248 (native-inputs
6249 `(("bundler" ,bundler)
6250 ("ruby-rspec" ,ruby-rspec)
6251 ("ruby-rake-compiler" ,ruby-rake-compiler)))
6252 (inputs
6253 `(("zlib" ,zlib)))
6254 (synopsis "Ruby bindings for the kseq.h FASTA/Q parser")
6255 (description
6256 "@code{Bio::Kseq} provides ruby bindings to the @code{kseq.h} FASTA and
6257FASTQ parsing code. It provides a fast iterator over sequences and their
6258quality scores.")
6259 (home-page "https://github.com/gusevfe/bio-kseq")
6260 (license license:expat)))
6261
9c38b540
PP		 6262(define-public bio-locus
6263 (package
6264 (name "bio-locus")
6265 (version "0.0.7")
6266 (source
6267 (origin
6268 (method url-fetch)
6269 (uri (rubygems-uri "bio-locus" version))
6270 (sha256
6271 (base32
6272 "02vmrxyimkj9sahsp4zhfhnmbvz6dbbqz1y01vglf8cbwvkajfl0"))))
6273 (build-system ruby-build-system)
6274 (native-inputs
6275 `(("ruby-rspec" ,ruby-rspec)))
6276 (synopsis "Tool for fast querying of genome locations")
6277 (description
6278 "Bio-locus is a tabix-like tool for fast querying of genome
6279locations. Many file formats in bioinformatics contain records that
6280start with a chromosome name and a position for a SNP, or a start-end
6281position for indels. Bio-locus allows users to store this chr+pos or
6282chr+pos+alt information in a database.")
6283 (home-page "https://github.com/pjotrp/bio-locus")
6284 (license license:expat)))
edb15985 6285
b2bddb07
PP		 6286(define-public bio-blastxmlparser
6287 (package
6288 (name "bio-blastxmlparser")
6289 (version "2.0.4")
6290 (source (origin
6291 (method url-fetch)
6292 (uri (rubygems-uri "bio-blastxmlparser" version))
6293 (sha256
6294 (base32
6295 "1wf4qygcmdjgcqm6flmvsagfr1gs9lf63mj32qv3z1f481zc5692"))))
6296 (build-system ruby-build-system)
6297 (propagated-inputs
6298 `(("ruby-bio-logger" ,ruby-bio-logger)
6299 ("ruby-nokogiri" ,ruby-nokogiri)))
6300 (inputs
6301 `(("ruby-rspec" ,ruby-rspec)))
6302 (synopsis "Fast big data BLAST XML parser and library")
6303 (description
6304 "Very fast parallel big-data BLAST XML file parser which can be used as
6305command line utility. Use blastxmlparser to: Parse BLAST XML; filter output;
6306generate FASTA, JSON, YAML, RDF, JSON-LD, HTML, CSV, tabular output etc.")
7bf837fd 6307 (home-page "https://github.com/pjotrp/blastxmlparser")
b2bddb07
PP		 6308 (license license:expat)))
6309
edb15985
PP		 6310(define-public bioruby
6311 (package
6312 (name "bioruby")
dbf9d371 6313 (version "1.5.1")
edb15985
PP		 6314 (source
6315 (origin
6316 (method url-fetch)
6317 (uri (rubygems-uri "bio" version))
6318 (sha256
6319 (base32
dbf9d371 6320 "0hdl0789c9n4mprnx5pgd46bfwl8d000rqpamj5h6kkjgspijv49"))))
edb15985
PP		 6321 (build-system ruby-build-system)
6322 (propagated-inputs
6323 `(("ruby-libxml" ,ruby-libxml)))
6324 (native-inputs
6325 `(("which" ,which))) ; required for test phase
6326 (arguments
6327 `(#:phases
6328 (modify-phases %standard-phases
6329 (add-before 'build 'patch-test-command
6330 (lambda _
6331 (substitute* '("test/functional/bio/test_command.rb")
6332 (("/bin/sh") (which "sh")))
6333 (substitute* '("test/functional/bio/test_command.rb")
6334 (("/bin/ls") (which "ls")))
6335 (substitute* '("test/functional/bio/test_command.rb")
6336 (("which") (which "which")))
6337 (substitute* '("test/functional/bio/test_command.rb",
6338 "test/data/command/echoarg2.sh")
6339 (("/bin/echo") (which "echo")))
6340 #t)))))
6341 (synopsis "Ruby library, shell and utilities for bioinformatics")
6342 (description "BioRuby comes with a comprehensive set of Ruby development
6343tools and libraries for bioinformatics and molecular biology. BioRuby has
6344components for sequence analysis, pathway analysis, protein modelling and
6345phylogenetic analysis; it supports many widely used data formats and provides
6346easy access to databases, external programs and public web services, including
6347BLAST, KEGG, GenBank, MEDLINE and GO.")
6348 (home-page "http://bioruby.org/")
6349 ;; Code is released under Ruby license, except for setup
6350 ;; (LGPLv2.1+) and scripts in samples (which have GPL2 and GPL2+)
6351 (license (list license:ruby license:lgpl2.1+ license:gpl2+ ))))
a5002ae7 6352
9fba89e8
RW		 6353(define-public r-acsnminer
6354 (package
6355 (name "r-acsnminer")
0b54b4c9 6356 (version "0.16.8.25")
9fba89e8
RW		 6357 (source (origin
6358 (method url-fetch)
6359 (uri (cran-uri "ACSNMineR" version))
6360 (sha256
6361 (base32
0b54b4c9 6362 "0gh604s8qall6zfjlwcg2ilxjvz08dplf9k5g47idhv43scm748l"))))
9fba89e8
RW		 6363 (properties `((upstream-name . "ACSNMineR")))
6364 (build-system r-build-system)
6365 (propagated-inputs
6366 `(("r-ggplot2" ,r-ggplot2)
6367 ("r-gridextra" ,r-gridextra)))
6368 (home-page "http://cran.r-project.org/web/packages/ACSNMineR")
6369 (synopsis "Gene enrichment analysis")
6370 (description
6371 "This package provides tools to compute and represent gene set enrichment
6372or depletion from your data based on pre-saved maps from the @dfn{Atlas of
6373Cancer Signalling Networks} (ACSN) or user imported maps. The gene set
6374enrichment can be run with hypergeometric test or Fisher exact test, and can
6375use multiple corrections. Visualization of data can be done either by
6376barplots or heatmaps.")
6377 (license license:gpl2+)))
6378
d29b25c4
RW		 6379(define-public r-biocgenerics
6380 (package
6381 (name "r-biocgenerics")
28a5b3f2 6382 (version "0.22.0")
d29b25c4
RW		 6383 (source (origin
6384 (method url-fetch)
6385 (uri (bioconductor-uri "BiocGenerics" version))
6386 (sha256
6387 (base32
28a5b3f2 6388 "0qbmz2qxwwi30xpxpvp2h1h7l494rbbz5d9pls5cfhqdv3wnpzsv"))))
d29b25c4 6389 (properties
1d216b6e 6390 `((upstream-name . "BiocGenerics")))
d29b25c4 6391 (build-system r-build-system)
5713bbf1 6392 (home-page "https://bioconductor.org/packages/BiocGenerics")
d29b25c4
RW		 6393 (synopsis "S4 generic functions for Bioconductor")
6394 (description
6395 "This package provides S4 generic functions needed by many Bioconductor
6396packages.")
6397 (license license:artistic2.0)))
6398
eb24341f
RJ		 6399(define-public r-biocinstaller
6400 (package
6401 (name "r-biocinstaller")
3bef1528 6402 (version "1.26.0")
eb24341f
RJ		 6403 (source (origin
6404 (method url-fetch)
6405 (uri (bioconductor-uri "BiocInstaller" version))
6406 (sha256
6407 (base32
3bef1528 6408 "0njw2q3lq1rrjx8qzw5d2130l72bmd3g2z8qlxqmkdcbmmgliyj2"))))
eb24341f
RJ		 6409 (properties
6410 `((upstream-name . "BiocInstaller")))
6411 (build-system r-build-system)
5713bbf1 6412 (home-page "https://bioconductor.org/packages/BiocInstaller")
eb24341f
RJ		 6413 (synopsis "Install Bioconductor packages")
6414 (description "This package is used to install and update R packages from
6415Bioconductor, CRAN, and Github.")
6416 (license license:artistic2.0)))
6417
207ce8fb
RJ		 6418(define-public r-biocviews
6419 (package
6420 (name "r-biocviews")
68eac349 6421 (version "1.44.0")
207ce8fb
RJ		 6422 (source (origin
6423 (method url-fetch)
6424 (uri (bioconductor-uri "biocViews" version))
6425 (sha256
6426 (base32
68eac349 6427 "17hi8w0w63f5yc43kid5pbld3ca78sj6n8x9dmkbl8h48818xbga"))))
207ce8fb
RJ		 6428 (properties
6429 `((upstream-name . "biocViews")))
6430 (build-system r-build-system)
6431 (propagated-inputs
6432 `(("r-biobase" ,r-biobase)
6433 ("r-graph" ,r-graph)
6434 ("r-rbgl" ,r-rbgl)
6435 ("r-rcurl" ,r-rcurl)
6436 ("r-xml" ,r-xml)
6437 ("r-knitr" ,r-knitr)
6438 ("r-runit" ,r-runit)))
5713bbf1 6439 (home-page "https://bioconductor.org/packages/biocViews")
207ce8fb
RJ		 6440 (synopsis "Bioconductor package categorization helper")
6441 (description "The purpose of biocViews is to create HTML pages that
6442categorize packages in a Bioconductor package repository according to keywords,
6443also known as views, in a controlled vocabulary.")
6444 (license license:artistic2.0)))
6445
2abfc5b8
RJ		 6446(define-public r-bookdown
6447 (package
6448 (name "r-bookdown")
6bd27981 6449 (version "0.5")
2abfc5b8
RJ		 6450 (source (origin
6451 (method url-fetch)
6452 (uri (cran-uri "bookdown" version))
6453 (sha256
6454 (base32
6bd27981 6455 "0zm63kr4f4kja4qpwkzl119zzyciqj7ihajfqgfjpgb4dzaiycxp"))))
2abfc5b8
RJ		 6456 (build-system r-build-system)
6457 (propagated-inputs
6458 `(("r-htmltools" ,r-htmltools)
6459 ("r-knitr" ,r-knitr)
6460 ("r-rmarkdown" ,r-rmarkdown)
6461 ("r-yaml" ,r-yaml)))
6462 (home-page "https://github.com/rstudio/bookdown")
6463 (synopsis "Authoring books and technical documents with R markdown")
6464 (description "This package provides output formats and utilities for
6465authoring books and technical documents with R Markdown.")
6466 (license license:gpl3)))
6467
99df12cd
RJ		 6468(define-public r-biocstyle
6469 (package
6470 (name "r-biocstyle")
0ff47834 6471 (version "2.4.1")
99df12cd
RJ		 6472 (source (origin
6473 (method url-fetch)
6474 (uri (bioconductor-uri "BiocStyle" version))
6475 (sha256
6476 (base32
0ff47834 6477 "0bmgmsfll923v573g0kyzlmjd7gly5jwgd8vkrcwvbam1gz75f2c"))))
99df12cd
RJ		 6478 (properties
6479 `((upstream-name . "BiocStyle")))
6480 (build-system r-build-system)
3bef24c9
RJ		 6481 (propagated-inputs
6482 `(("r-bookdown" ,r-bookdown)
6483 ("r-knitr" ,r-knitr)
6484 ("r-rmarkdown" ,r-rmarkdown)
6485 ("r-yaml" ,r-yaml)))
5713bbf1 6486 (home-page "https://bioconductor.org/packages/BiocStyle")
99df12cd
RJ		 6487 (synopsis "Bioconductor formatting styles")
6488 (description "This package provides standard formatting styles for
6489Bioconductor PDF and HTML documents. Package vignettes illustrate use and
6490functionality.")
6491 (license license:artistic2.0)))
6492
4644644a
RJ		 6493(define-public r-bioccheck
6494 (package
6495 (name "r-bioccheck")
2e11beec 6496 (version "1.12.0")
4644644a
RJ		 6497 (source (origin
6498 (method url-fetch)
6499 (uri (bioconductor-uri "BiocCheck" version))
6500 (sha256
6501 (base32
2e11beec 6502 "01zkw5hggzvn0wj4msac71k1mknq4h2inn1c2hwqgw4cy1675wl0"))))
4644644a
RJ		 6503 (properties
6504 `((upstream-name . "BiocCheck")))
6505 (build-system r-build-system)
6506 (arguments
6507 '(#:phases
6508 (modify-phases %standard-phases
6509 ;; This package can be used by calling BiocCheck(<package>) from
6510 ;; within R, or by running R CMD BiocCheck <package>. This phase
6511 ;; makes sure the latter works. For this to work, the BiocCheck
6512 ;; script must be somewhere on the PATH (not the R bin directory).
6513 (add-after 'install 'install-bioccheck-subcommand
6514 (lambda* (#:key outputs #:allow-other-keys)
6515 (let* ((out (assoc-ref outputs "out"))
6516 (dest-dir (string-append out "/bin"))
6517 (script-dir
6518 (string-append out "/site-library/BiocCheck/script/")))
6519 (mkdir-p dest-dir)
6520 (symlink (string-append script-dir "/checkBadDeps.R")
6521 (string-append dest-dir "/checkBadDeps.R"))
6522 (symlink (string-append script-dir "/BiocCheck")
6523 (string-append dest-dir "/BiocCheck")))
6524 #t)))))
6525 (native-inputs
6526 `(("which" ,which)))
6527 (propagated-inputs
aeb64f3c
RW		 6528 `(("r-codetools" ,r-codetools)
6529 ("r-graph" ,r-graph)
4644644a
RJ		 6530 ("r-httr" ,r-httr)
6531 ("r-optparse" ,r-optparse)
4644644a
RJ		 6532 ("r-biocinstaller" ,r-biocinstaller)
6533 ("r-biocviews" ,r-biocviews)))
5713bbf1 6534 (home-page "https://bioconductor.org/packages/BiocCheck")
4644644a
RJ		 6535 (synopsis "Executes Bioconductor-specific package checks")
6536 (description "This package contains tools to perform additional quality
6537checks on R packages that are to be submitted to the Bioconductor repository.")
6538 (license license:artistic2.0)))
6539
2acaaee5
RJ		 6540(define-public r-getopt
6541 (package
6542 (name "r-getopt")
6543 (version "1.20.0")
6544 (source
6545 (origin
6546 (method url-fetch)
6547 (uri (cran-uri "getopt" version))
6548 (sha256
6549 (base32
6550 "00f57vgnzmg7cz80rjmjz1556xqcmx8nhrlbbhaq4w7gl2ibl87r"))))
6551 (build-system r-build-system)
6552 (home-page "https://github.com/trevorld/getopt")
6553 (synopsis "Command-line option processor for R")
6554 (description
6555 "This package is designed to be used with Rscript to write shebang
6556scripts that accept short and long options. Many users will prefer to
6557use the packages @code{optparse} or @code{argparse} which add extra
6558features like automatically generated help options and usage texts,
6559support for default values, positional argument support, etc.")
6560 (license license:gpl2+)))
6561
c79ad57a
RJ		 6562(define-public r-optparse
6563 (package
6564 (name "r-optparse")
7150f1c3 6565 (version "1.4.4")
c79ad57a
RJ		 6566 (source
6567 (origin
6568 (method url-fetch)
6569 (uri (cran-uri "optparse" version))
6570 (sha256
6571 (base32
7150f1c3 6572 "1ff4wmsszrb3spwfp7ynfs8w11qpy1sdzfxm1wk8dqqvdwris7qb"))))
c79ad57a
RJ		 6573 (build-system r-build-system)
6574 (propagated-inputs
6575 `(("r-getopt" ,r-getopt)))
6576 (home-page
6577 "https://github.com/trevorld/optparse")
6578 (synopsis "Command line option parser")
6579 (description
6580 "This package provides a command line parser inspired by Python's
6581@code{optparse} library to be used with Rscript to write shebang scripts
6582that accept short and long options.")
6583 (license license:gpl2+)))
6584
247d498a
RJ		 6585(define-public r-dnacopy
6586 (package
6587 (name "r-dnacopy")
9c4348d8 6588 (version "1.50.1")
247d498a
RJ		 6589 (source (origin
6590 (method url-fetch)
6591 (uri (bioconductor-uri "DNAcopy" version))
6592 (sha256
6593 (base32
9c4348d8 6594 "0f0x83db7rm5xf9fg5pjhvs4i165qqaf01lbwb8kj13fsqpwx15p"))))
247d498a
RJ		 6595 (properties
6596 `((upstream-name . "DNAcopy")))
6597 (build-system r-build-system)
6598 (inputs
6599 `(("gfortran" ,gfortran)))
5697fdc3 6600 (home-page "https://bioconductor.org/packages/DNAcopy")
247d498a
RJ		 6601 (synopsis "Implementation of a circular binary segmentation algorithm")
6602 (description "This package implements the circular binary segmentation (CBS)
6603algorithm to segment DNA copy number data and identify genomic regions with
6604abnormal copy number.")
6605 (license license:gpl2+)))
6606
7485129e
RW		 6607(define-public r-s4vectors
6608 (package
6609 (name "r-s4vectors")
3a4b0b12 6610 (version "0.14.3")
7485129e
RW		 6611 (source (origin
6612 (method url-fetch)
6613 (uri (bioconductor-uri "S4Vectors" version))
6614 (sha256
6615 (base32
3a4b0b12 6616 "1r7s4pfw026qazzic090mhk8d9m39j2nwl87dyqcpdylyq7gq5qs"))))
7485129e 6617 (properties
1d216b6e 6618 `((upstream-name . "S4Vectors")))
7485129e
RW		 6619 (build-system r-build-system)
6620 (propagated-inputs
6621 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6622 (home-page "https://bioconductor.org/packages/S4Vectors")
7485129e
RW		 6623 (synopsis "S4 implementation of vectors and lists")
6624 (description
6625 "The S4Vectors package defines the @code{Vector} and @code{List} virtual
6626classes and a set of generic functions that extend the semantic of ordinary
6627vectors and lists in R. Package developers can easily implement vector-like
6628or list-like objects as concrete subclasses of @code{Vector} or @code{List}.
6629In addition, a few low-level concrete subclasses of general interest (e.g.
6630@code{DataFrame}, @code{Rle}, and @code{Hits}) are implemented in the
6631S4Vectors package itself.")
6632 (license license:artistic2.0)))
6633
274da826
RW		 6634(define-public r-seqinr
6635 (package
6636 (name "r-seqinr")
023aa8ff 6637 (version "3.4-5")
274da826
RW		 6638 (source
6639 (origin
6640 (method url-fetch)
6641 (uri (cran-uri "seqinr" version))
6642 (sha256
6643 (base32
023aa8ff 6644 "17zv0n5cji17izwmwg0jcbxbjl3w5rls91w15svcnlpxjms38ahn"))))
274da826
RW		 6645 (build-system r-build-system)
6646 (propagated-inputs
3b851cd4
RW		 6647 `(("r-ade4" ,r-ade4)
6648 ("r-segmented" ,r-segmented)))
274da826
RW		 6649 (inputs
6650 `(("zlib" ,zlib)))
6651 (home-page "http://seqinr.r-forge.r-project.org/")
6652 (synopsis "Biological sequences retrieval and analysis")
6653 (description
6654 "This package provides tools for exploratory data analysis and data
6655visualization of biological sequence (DNA and protein) data. It also includes
6656utilities for sequence data management under the ACNUC system.")
6657 (license license:gpl2+)))
6658
78addcb0
RW		 6659(define-public r-iranges
6660 (package
6661 (name "r-iranges")
3ec25eb2 6662 (version "2.10.2")
78addcb0
RW		 6663 (source (origin
6664 (method url-fetch)
6665 (uri (bioconductor-uri "IRanges" version))
6666 (sha256
6667 (base32
3ec25eb2 6668 "1brmzs3rsf97gymridrh9c9r3vws8b3rpghaanxnniw36lmcajfy"))))
78addcb0 6669 (properties
1d216b6e 6670 `((upstream-name . "IRanges")))
78addcb0
RW		 6671 (build-system r-build-system)
6672 (propagated-inputs
6673 `(("r-biocgenerics" ,r-biocgenerics)
6674 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6675 (home-page "https://bioconductor.org/packages/IRanges")
78addcb0
RW		 6676 (synopsis "Infrastructure for manipulating intervals on sequences")
6677 (description
6678 "This package provides efficient low-level and highly reusable S4 classes
6679for storing ranges of integers, RLE vectors (Run-Length Encoding), and, more
6680generally, data that can be organized sequentially (formally defined as
6681@code{Vector} objects), as well as views on these @code{Vector} objects.
6682Efficient list-like classes are also provided for storing big collections of
6683instances of the basic classes. All classes in the package use consistent
6684naming and share the same rich and consistent \"Vector API\" as much as
6685possible.")
6686 (license license:artistic2.0)))
6687
ffef27f3
RJ		 6688(define-public r-genomeinfodbdata
6689 (package
6690 (name "r-genomeinfodbdata")
6691 (version "0.99.0")
6692 (source (origin
6693 (method url-fetch)
90f83099
EF		 6694 ;; We cannot use bioconductor-uri here because this tarball is
6695 ;; located under "data/annotation/" instead of "bioc/".
6696 (uri (string-append "https://bioconductor.org/packages/release/"
6697 "data/annotation/src/contrib/GenomeInfoDbData_"
6698 version ".tar.gz"))
ffef27f3
RJ		 6699 (sha256
6700 (base32
6701 "120qvhb0pvkzd65lsgja62vyrgc37si6fh68q4cg4w5x9f04jw25"))))
6702 (properties
6703 `((upstream-name . "GenomeInfoDbData")))
6704 (build-system r-build-system)
5713bbf1 6705 (home-page "https://bioconductor.org/packages/GenomeInfoDbData")
ffef27f3
RJ		 6706 (synopsis "Species and taxonomy ID look up tables for GenomeInfoDb")
6707 (description "This package contains data for mapping between NCBI taxonomy
6708ID and species. It is used by functions in the GenomeInfoDb package.")
6709 (license license:artistic2.0)))
6710
bf7764b7
RW		 6711(define-public r-genomeinfodb
6712 (package
6713 (name "r-genomeinfodb")
6252dd0f 6714 (version "1.12.2")
bf7764b7
RW		 6715 (source (origin
6716 (method url-fetch)
6717 (uri (bioconductor-uri "GenomeInfoDb" version))
6718 (sha256
6719 (base32
6252dd0f 6720 "1hjxgmcnrngp1307ipqaq9hgxz4j0ldn7d46knhzs30k2r4qnrfp"))))
bf7764b7 6721 (properties
1d216b6e 6722 `((upstream-name . "GenomeInfoDb")))
bf7764b7
RW		 6723 (build-system r-build-system)
6724 (propagated-inputs
6725 `(("r-biocgenerics" ,r-biocgenerics)
38b99ccc 6726 ("r-genomeinfodbdata" ,r-genomeinfodbdata)
bf7764b7 6727 ("r-iranges" ,r-iranges)
4cd07e48 6728 ("r-rcurl" ,r-rcurl)
bf7764b7 6729 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6730 (home-page "https://bioconductor.org/packages/GenomeInfoDb")
bf7764b7
RW		 6731 (synopsis "Utilities for manipulating chromosome identifiers")
6732 (description
6733 "This package contains data and functions that define and allow
6734translation between different chromosome sequence naming conventions (e.g.,
6735\"chr1\" versus \"1\"), including a function that attempts to place sequence
6736names in their natural, rather than lexicographic, order.")
6737 (license license:artistic2.0)))
6738
744004a3
RJ		 6739(define-public r-edger
6740 (package
6741 (name "r-edger")
00ffd55f 6742 (version "3.18.0")
744004a3
RJ		 6743 (source (origin
6744 (method url-fetch)
6745 (uri (bioconductor-uri "edgeR" version))
6746 (sha256
6747 (base32
00ffd55f 6748 "02l17i6xh33dv10swwvyfxrmv5kp23iv278iwvkfq2mnipasfnb9"))))
744004a3
RJ		 6749 (properties `((upstream-name . "edgeR")))
6750 (build-system r-build-system)
6751 (propagated-inputs
5e48005f 6752 `(("r-limma" ,r-limma)
47055b27
RW		 6753 ("r-locfit" ,r-locfit)
6754 ("r-statmod" ,r-statmod))) ;for estimateDisp
744004a3
RJ		 6755 (home-page "http://bioinf.wehi.edu.au/edgeR")
6756 (synopsis "EdgeR does empirical analysis of digital gene expression data")
6757 (description "This package can do differential expression analysis of
6758RNA-seq expression profiles with biological replication. It implements a range
6759of statistical methodology based on the negative binomial distributions,
6760including empirical Bayes estimation, exact tests, generalized linear models
6761and quasi-likelihood tests. It be applied to differential signal analysis of
6762other types of genomic data that produce counts, including ChIP-seq, SAGE and
6763CAGE.")
6764 (license license:gpl2+)))
6765
b669d9c4
RJ		 6766(define-public r-variantannotation
6767 (package
6768 (name "r-variantannotation")
48f5d283 6769 (version "1.22.3")
b669d9c4
RJ		 6770 (source (origin
6771 (method url-fetch)
6772 (uri (bioconductor-uri "VariantAnnotation" version))
6773 (sha256
6774 (base32
48f5d283 6775 "0sr3vdn85x5zdxh80cfwlpfdpi2hmjy3fwi00ac3jya4v145vawr"))))
b669d9c4
RJ		 6776 (properties
6777 `((upstream-name . "VariantAnnotation")))
6778 (inputs
6779 `(("zlib" ,zlib)))
6780 (propagated-inputs
6781 `(("r-annotationdbi" ,r-annotationdbi)
37d96f1d 6782 ("r-biobase" ,r-biobase)
b669d9c4 6783 ("r-biocgenerics" ,r-biocgenerics)
37d96f1d 6784 ("r-biostrings" ,r-biostrings)
b669d9c4
RJ		 6785 ("r-bsgenome" ,r-bsgenome)
6786 ("r-dbi" ,r-dbi)
6787 ("r-genomeinfodb" ,r-genomeinfodb)
6788 ("r-genomicfeatures" ,r-genomicfeatures)
6789 ("r-genomicranges" ,r-genomicranges)
37d96f1d 6790 ("r-iranges" ,r-iranges)
b669d9c4
RJ		 6791 ("r-summarizedexperiment" ,r-summarizedexperiment)
6792 ("r-rsamtools" ,r-rsamtools)
37d96f1d
RW		 6793 ("r-rtracklayer" ,r-rtracklayer)
6794 ("r-s4vectors" ,r-s4vectors)
6795 ("r-xvector" ,r-xvector)
b669d9c4
RJ		 6796 ("r-zlibbioc" ,r-zlibbioc)))
6797 (build-system r-build-system)
6798 (home-page "https://bioconductor.org/packages/VariantAnnotation")
6799 (synopsis "Package for annotation of genetic variants")
6800 (description "This R package can annotate variants, compute amino acid
6801coding changes and predict coding outcomes.")
6802 (license license:artistic2.0)))
6803
7d4224d7
RJ		 6804(define-public r-limma
6805 (package
6806 (name "r-limma")
1b4738e2 6807 (version "3.32.5")
7d4224d7
RJ		 6808 (source (origin
6809 (method url-fetch)
6810 (uri (bioconductor-uri "limma" version))
6811 (sha256
6812 (base32
1b4738e2 6813 "0p2ayha9g9w5r8s7pgdf16mkmdbqwh6f35jh07g3b8gyra48gwiw"))))
7d4224d7
RJ		 6814 (build-system r-build-system)
6815 (home-page "http://bioinf.wehi.edu.au/limma")
6816 (synopsis "Package for linear models for microarray and RNA-seq data")
6817 (description "This package can be used for the analysis of gene expression
6818studies, especially the use of linear models for analysing designed experiments
6819and the assessment of differential expression. The analysis methods apply to
6820different technologies, including microarrays, RNA-seq, and quantitative PCR.")
6821 (license license:gpl2+)))
6822
0e7d5560
RW		 6823(define-public r-xvector
6824 (package
6825 (name "r-xvector")
17a02f51 6826 (version "0.16.0")
0e7d5560
RW		 6827 (source (origin
6828 (method url-fetch)
6829 (uri (bioconductor-uri "XVector" version))
6830 (sha256
6831 (base32
17a02f51 6832 "01n09f4jdm60684lzikp02zf9gjan8bdrjx740vggr21q9fa69wn"))))
0e7d5560 6833 (properties
1d216b6e 6834 `((upstream-name . "XVector")))
0e7d5560
RW		 6835 (build-system r-build-system)
6836 (arguments
6837 `(#:phases
6838 (modify-phases %standard-phases
6839 (add-after 'unpack 'use-system-zlib
6840 (lambda _
6841 (substitute* "DESCRIPTION"
6842 (("zlibbioc, ") ""))
6843 (substitute* "NAMESPACE"
6844 (("import\\(zlibbioc\\)") ""))
6845 #t)))))
6846 (inputs
6847 `(("zlib" ,zlib)))
6848 (propagated-inputs
6849 `(("r-biocgenerics" ,r-biocgenerics)
6850 ("r-iranges" ,r-iranges)
6851 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6852 (home-page "https://bioconductor.org/packages/XVector")
0e7d5560
RW		 6853 (synopsis "Representation and manpulation of external sequences")
6854 (description
6855 "This package provides memory efficient S4 classes for storing sequences
6856\"externally\" (behind an R external pointer, or on disk).")
6857 (license license:artistic2.0)))
6858
e2cd1d0f
RW		 6859(define-public r-genomicranges
6860 (package
6861 (name "r-genomicranges")
a35a1e9c 6862 (version "1.28.4")
e2cd1d0f
RW		 6863 (source (origin
6864 (method url-fetch)
6865 (uri (bioconductor-uri "GenomicRanges" version))
6866 (sha256
6867 (base32
a35a1e9c 6868 "1y15kg1q81h8rmga83ljiwr8whkajcargfjiljr212d6if17ys1z"))))
e2cd1d0f 6869 (properties
1d216b6e 6870 `((upstream-name . "GenomicRanges")))
e2cd1d0f
RW		 6871 (build-system r-build-system)
6872 (propagated-inputs
6873 `(("r-biocgenerics" ,r-biocgenerics)
6874 ("r-genomeinfodb" ,r-genomeinfodb)
92a740af
RW		 6875 ("r-iranges" ,r-iranges)
6876 ("r-s4vectors" ,r-s4vectors)
e2cd1d0f 6877 ("r-xvector" ,r-xvector)))
5713bbf1 6878 (home-page "https://bioconductor.org/packages/GenomicRanges")
e2cd1d0f
RW		 6879 (synopsis "Representation and manipulation of genomic intervals")
6880 (description
6881 "This package provides tools to efficiently represent and manipulate
6882genomic annotations and alignments is playing a central role when it comes to
6883analyzing high-throughput sequencing data (a.k.a. NGS data). The
6884GenomicRanges package defines general purpose containers for storing and
6885manipulating genomic intervals and variables defined along a genome.")
6886 (license license:artistic2.0)))
6887
555e3399
RW		 6888(define-public r-biobase
6889 (package
6890 (name "r-biobase")
1898a783 6891 (version "2.36.2")
555e3399
RW		 6892 (source (origin
6893 (method url-fetch)
6894 (uri (bioconductor-uri "Biobase" version))
6895 (sha256
6896 (base32
1898a783 6897 "0sr48nqx5bqid4g6lr9zr9286xh842w717yvmssvddb5xxynib6f"))))
555e3399
RW		 6898 (properties
6899 `((upstream-name . "Biobase")))
6900 (build-system r-build-system)
6901 (propagated-inputs
6902 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 6903 (home-page "https://bioconductor.org/packages/Biobase")
555e3399
RW		 6904 (synopsis "Base functions for Bioconductor")
6905 (description
6906 "This package provides functions that are needed by many other packages
6907on Bioconductor or which replace R functions.")
6908 (license license:artistic2.0)))
6909
8b7bce74
RW		 6910(define-public r-annotationdbi
6911 (package
6912 (name "r-annotationdbi")
62eb93a7 6913 (version "1.38.2")
8b7bce74
RW		 6914 (source (origin
6915 (method url-fetch)
6916 (uri (bioconductor-uri "AnnotationDbi" version))
6917 (sha256
6918 (base32
62eb93a7 6919 "1lsamnbf07zzsy5asy5hn97n2a4layv58w2bzd90ikcdx0gmzarj"))))
8b7bce74
RW		 6920 (properties
6921 `((upstream-name . "AnnotationDbi")))
6922 (build-system r-build-system)
6923 (propagated-inputs
6924 `(("r-biobase" ,r-biobase)
6925 ("r-biocgenerics" ,r-biocgenerics)
6926 ("r-dbi" ,r-dbi)
6927 ("r-iranges" ,r-iranges)
6928 ("r-rsqlite" ,r-rsqlite)
6929 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 6930 (home-page "https://bioconductor.org/packages/AnnotationDbi")
8b7bce74
RW		 6931 (synopsis "Annotation database interface")
6932 (description
6933 "This package provides user interface and database connection code for
6934annotation data packages using SQLite data storage.")
6935 (license license:artistic2.0)))
6936
c465fa72
RW		 6937(define-public r-biomart
6938 (package
6939 (name "r-biomart")
0c6494a1 6940 (version "2.32.1")
c465fa72
RW		 6941 (source (origin
6942 (method url-fetch)
6943 (uri (bioconductor-uri "biomaRt" version))
6944 (sha256
6945 (base32
0c6494a1 6946 "0fhpbjlsgbqxrpj6nzhhk9q3ph81n5x4p7mmd097xjjn6b05w1d8"))))
c465fa72
RW		 6947 (properties
6948 `((upstream-name . "biomaRt")))
6949 (build-system r-build-system)
6950 (propagated-inputs
6951 `(("r-annotationdbi" ,r-annotationdbi)
6952 ("r-rcurl" ,r-rcurl)
6953 ("r-xml" ,r-xml)))
5713bbf1 6954 (home-page "https://bioconductor.org/packages/biomaRt")
c465fa72
RW		 6955 (synopsis "Interface to BioMart databases")
6956 (description
6957 "biomaRt provides an interface to a growing collection of databases
6958implementing the @url{BioMart software suite, http://www.biomart.org}. The
6959package enables retrieval of large amounts of data in a uniform way without
6960the need to know the underlying database schemas or write complex SQL queries.
6961Examples of BioMart databases are Ensembl, COSMIC, Uniprot, HGNC, Gramene,
6962Wormbase and dbSNP mapped to Ensembl. These major databases give biomaRt
6963users direct access to a diverse set of data and enable a wide range of
6964powerful online queries from gene annotation to database mining.")
6965 (license license:artistic2.0)))
6966
e91d362e
RW		 6967(define-public r-biocparallel
6968 (package
6969 (name "r-biocparallel")
1ff74d07 6970 (version "1.10.1")
e91d362e
RW		 6971 (source (origin
6972 (method url-fetch)
6973 (uri (bioconductor-uri "BiocParallel" version))
6974 (sha256
6975 (base32
1ff74d07 6976 "08mdfxyk9nwz77v0xhlvs19p2wj0phgm5c5b25vm0xh3749njsp0"))))
e91d362e
RW		 6977 (properties
6978 `((upstream-name . "BiocParallel")))
6979 (build-system r-build-system)
6980 (propagated-inputs
6981 `(("r-futile-logger" ,r-futile-logger)
6982 ("r-snow" ,r-snow)))
5713bbf1 6983 (home-page "https://bioconductor.org/packages/BiocParallel")
e91d362e
RW		 6984 (synopsis "Bioconductor facilities for parallel evaluation")
6985 (description
6986 "This package provides modified versions and novel implementation of
6987functions for parallel evaluation, tailored to use with Bioconductor
6988objects.")
6989 (license (list license:gpl2+ license:gpl3+))))
6990
bf159353
RW		 6991(define-public r-biostrings
6992 (package
6993 (name "r-biostrings")
5077bc3e 6994 (version "2.44.2")
bf159353
RW		 6995 (source (origin
6996 (method url-fetch)
6997 (uri (bioconductor-uri "Biostrings" version))
6998 (sha256
6999 (base32
5077bc3e 7000 "12c5abgshwq86357jr0r9039y6vl4d6ngysy89rsnr23ldnsirjp"))))
bf159353
RW		 7001 (properties
7002 `((upstream-name . "Biostrings")))
7003 (build-system r-build-system)
7004 (propagated-inputs
7005 `(("r-biocgenerics" ,r-biocgenerics)
7006 ("r-iranges" ,r-iranges)
7007 ("r-s4vectors" ,r-s4vectors)
7008 ("r-xvector" ,r-xvector)))
5713bbf1 7009 (home-page "https://bioconductor.org/packages/Biostrings")
bf159353
RW		 7010 (synopsis "String objects and algorithms for biological sequences")
7011 (description
7012 "This package provides memory efficient string containers, string
7013matching algorithms, and other utilities, for fast manipulation of large
7014biological sequences or sets of sequences.")
7015 (license license:artistic2.0)))
7016
f8d74f70
RW		 7017(define-public r-rsamtools
7018 (package
7019 (name "r-rsamtools")
6bf5543b 7020 (version "1.28.0")
f8d74f70
RW		 7021 (source (origin
7022 (method url-fetch)
7023 (uri (bioconductor-uri "Rsamtools" version))
7024 (sha256
7025 (base32
6bf5543b 7026 "1zx6vhdz3lksbfy98jj3qzl7cxgspigm2pbsd5835b12r6mc5w6d"))))
f8d74f70
RW		 7027 (properties
7028 `((upstream-name . "Rsamtools")))
7029 (build-system r-build-system)
7030 (arguments
7031 `(#:phases
7032 (modify-phases %standard-phases
7033 (add-after 'unpack 'use-system-zlib
7034 (lambda _
7035 (substitute* "DESCRIPTION"
7036 (("zlibbioc, ") ""))
7037 (substitute* "NAMESPACE"
7038 (("import\\(zlibbioc\\)") ""))
7039 #t)))))
7040 (inputs
7041 `(("zlib" ,zlib)))
7042 (propagated-inputs
7043 `(("r-biocgenerics" ,r-biocgenerics)
7044 ("r-biocparallel" ,r-biocparallel)
7045 ("r-biostrings" ,r-biostrings)
7046 ("r-bitops" ,r-bitops)
7047 ("r-genomeinfodb" ,r-genomeinfodb)
7048 ("r-genomicranges" ,r-genomicranges)
7049 ("r-iranges" ,r-iranges)
7050 ("r-s4vectors" ,r-s4vectors)
7051 ("r-xvector" ,r-xvector)))
5713bbf1 7052 (home-page "https://bioconductor.org/packages/release/bioc/html/Rsamtools.html")
f8d74f70
RW		 7053 (synopsis "Interface to samtools, bcftools, and tabix")
7054 (description
7055 "This package provides an interface to the 'samtools', 'bcftools', and
7056'tabix' utilities for manipulating SAM (Sequence Alignment / Map), FASTA,
7057binary variant call (BCF) and compressed indexed tab-delimited (tabix)
7058files.")
7059 (license license:expat)))
7060
71e34e6b
RJ		 7061(define-public r-delayedarray
7062 (package
7063 (name "r-delayedarray")
adf5d0a0 7064 (version "0.2.7")
71e34e6b
RJ		 7065 (source (origin
7066 (method url-fetch)
7067 (uri (bioconductor-uri "DelayedArray" version))
7068 (sha256
7069 (base32
adf5d0a0 7070 "02dfqp4md9xaqjj712ijc3jswghmipr5hwkd5hr0x1xi6l2fb69g"))))
71e34e6b
RJ		 7071 (properties
7072 `((upstream-name . "DelayedArray")))
7073 (build-system r-build-system)
7074 (propagated-inputs
7075 `(("r-biocgenerics" ,r-biocgenerics)
7076 ("r-s4vectors" ,r-s4vectors)
7077 ("r-iranges" ,r-iranges)
7078 ("r-matrixstats" ,r-matrixstats)))
5713bbf1 7079 (home-page "https://bioconductor.org/packages/DelayedArray")
71e34e6b
RJ		 7080 (synopsis "Delayed operations on array-like objects")
7081 (description
7082 "Wrapping an array-like object (typically an on-disk object) in a
7083@code{DelayedArray} object allows one to perform common array operations on it
7084without loading the object in memory. In order to reduce memory usage and
7085optimize performance, operations on the object are either delayed or executed
7086using a block processing mechanism. Note that this also works on in-memory
7087array-like objects like @code{DataFrame} objects (typically with Rle columns),
7088@code{Matrix} objects, and ordinary arrays and data frames.")
7089 (license license:artistic2.0)))
7090
6e76dda2
RW		 7091(define-public r-summarizedexperiment
7092 (package
7093 (name "r-summarizedexperiment")
f41150f1 7094 (version "1.6.3")
6e76dda2
RW		 7095 (source (origin
7096 (method url-fetch)
7097 (uri (bioconductor-uri "SummarizedExperiment" version))
7098 (sha256
7099 (base32
f41150f1 7100 "0j7xn7pk52d383fb1wplcggacl2586c4zi0alkgfc3wz7qq9w13s"))))
6e76dda2
RW		 7101 (properties
7102 `((upstream-name . "SummarizedExperiment")))
7103 (build-system r-build-system)
7104 (propagated-inputs
7105 `(("r-biobase" ,r-biobase)
7106 ("r-biocgenerics" ,r-biocgenerics)
d006ee31 7107 ("r-delayedarray" ,r-delayedarray)
6e76dda2
RW		 7108 ("r-genomeinfodb" ,r-genomeinfodb)
7109 ("r-genomicranges" ,r-genomicranges)
7110 ("r-iranges" ,r-iranges)
aeb64f3c 7111 ("r-matrix" ,r-matrix)
6e76dda2 7112 ("r-s4vectors" ,r-s4vectors)))
5713bbf1 7113 (home-page "https://bioconductor.org/packages/SummarizedExperiment")
6e76dda2
RW		 7114 (synopsis "Container for representing genomic ranges by sample")
7115 (description
7116 "The SummarizedExperiment container contains one or more assays, each
7117represented by a matrix-like object of numeric or other mode. The rows
7118typically represent genomic ranges of interest and the columns represent
7119samples.")
7120 (license license:artistic2.0)))
7121
d8a828af
RW		 7122(define-public r-genomicalignments
7123 (package
7124 (name "r-genomicalignments")
81195d13 7125 (version "1.12.2")
d8a828af
RW		 7126 (source (origin
7127 (method url-fetch)
7128 (uri (bioconductor-uri "GenomicAlignments" version))
7129 (sha256
7130 (base32
81195d13 7131 "03ysxi9fdd3bcfj05iaysya9knn2aa2irwpypb5srg0xwv92bdb9"))))
d8a828af
RW		 7132 (properties
7133 `((upstream-name . "GenomicAlignments")))
7134 (build-system r-build-system)
7135 (propagated-inputs
7136 `(("r-biocgenerics" ,r-biocgenerics)
7137 ("r-biocparallel" ,r-biocparallel)
7138 ("r-biostrings" ,r-biostrings)
7139 ("r-genomeinfodb" ,r-genomeinfodb)
7140 ("r-genomicranges" ,r-genomicranges)
7141 ("r-iranges" ,r-iranges)
7142 ("r-rsamtools" ,r-rsamtools)
7143 ("r-s4vectors" ,r-s4vectors)
7144 ("r-summarizedexperiment" ,r-summarizedexperiment)))
5713bbf1 7145 (home-page "https://bioconductor.org/packages/GenomicAlignments")
d8a828af
RW		 7146 (synopsis "Representation and manipulation of short genomic alignments")
7147 (description
7148 "This package provides efficient containers for storing and manipulating
7149short genomic alignments (typically obtained by aligning short reads to a
7150reference genome). This includes read counting, computing the coverage,
7151junction detection, and working with the nucleotide content of the
7152alignments.")
7153 (license license:artistic2.0)))
7154
317755ff
RW		 7155(define-public r-rtracklayer
7156 (package
7157 (name "r-rtracklayer")
15afa9bf 7158 (version "1.36.4")
317755ff
RW		 7159 (source (origin
7160 (method url-fetch)
7161 (uri (bioconductor-uri "rtracklayer" version))
7162 (sha256
7163 (base32
15afa9bf 7164 "050q1rv04w31168ljr975vxva31n9lqdx84rnmsk6zcr6p640ffp"))))
317755ff
RW		 7165 (build-system r-build-system)
7166 (arguments
7167 `(#:phases
7168 (modify-phases %standard-phases
7169 (add-after 'unpack 'use-system-zlib
7170 (lambda _
7171 (substitute* "DESCRIPTION"
7172 (("zlibbioc, ") ""))
7173 (substitute* "NAMESPACE"
7174 (("import\\(zlibbioc\\)") ""))
7175 #t)))))
7176 (inputs
7177 `(("zlib" ,zlib)))
7178 (propagated-inputs
7179 `(("r-biocgenerics" ,r-biocgenerics)
7180 ("r-biostrings" ,r-biostrings)
7181 ("r-genomeinfodb" ,r-genomeinfodb)
7182 ("r-genomicalignments" ,r-genomicalignments)
7183 ("r-genomicranges" ,r-genomicranges)
7184 ("r-iranges" ,r-iranges)
7185 ("r-rcurl" ,r-rcurl)
7186 ("r-rsamtools" ,r-rsamtools)
7187 ("r-s4vectors" ,r-s4vectors)
7188 ("r-xml" ,r-xml)
7189 ("r-xvector" ,r-xvector)))
5713bbf1 7190 (home-page "https://bioconductor.org/packages/rtracklayer")
317755ff
RW		 7191 (synopsis "R interface to genome browsers and their annotation tracks")
7192 (description
7193 "rtracklayer is an extensible framework for interacting with multiple
7194genome browsers (currently UCSC built-in) and manipulating annotation tracks
7195in various formats (currently GFF, BED, bedGraph, BED15, WIG, BigWig and 2bit
7196built-in). The user may export/import tracks to/from the supported browsers,
7197as well as query and modify the browser state, such as the current viewport.")
7198 (license license:artistic2.0)))
7199
2fd7c049
RW		 7200(define-public r-genomicfeatures
7201 (package
7202 (name "r-genomicfeatures")
8cfaa9ab 7203 (version "1.28.4")
2fd7c049
RW		 7204 (source (origin
7205 (method url-fetch)
7206 (uri (bioconductor-uri "GenomicFeatures" version))
7207 (sha256
7208 (base32
8cfaa9ab 7209 "01ylvg275iy0cvsbxkbfxcf9pi9al597v5wnlqi2xdpmrcxyc3q0"))))
2fd7c049
RW		 7210 (properties
7211 `((upstream-name . "GenomicFeatures")))
7212 (build-system r-build-system)
7213 (propagated-inputs
7214 `(("r-annotationdbi" ,r-annotationdbi)
7215 ("r-biobase" ,r-biobase)
7216 ("r-biocgenerics" ,r-biocgenerics)
7217 ("r-biomart" ,r-biomart)
7218 ("r-biostrings" ,r-biostrings)
7219 ("r-dbi" ,r-dbi)
7220 ("r-genomeinfodb" ,r-genomeinfodb)
7221 ("r-genomicranges" ,r-genomicranges)
7222 ("r-iranges" ,r-iranges)
7223 ("r-rcurl" ,r-rcurl)
7224 ("r-rsqlite" ,r-rsqlite)
7225 ("r-rtracklayer" ,r-rtracklayer)
7226 ("r-s4vectors" ,r-s4vectors)
7227 ("r-xvector" ,r-xvector)))
5713bbf1 7228 (home-page "https://bioconductor.org/packages/GenomicFeatures")
2fd7c049
RW		 7229 (synopsis "Tools for working with transcript centric annotations")
7230 (description
7231 "This package provides a set of tools and methods for making and
7232manipulating transcript centric annotations. With these tools the user can
7233easily download the genomic locations of the transcripts, exons and cds of a
7234given organism, from either the UCSC Genome Browser or a BioMart
7235database (more sources will be supported in the future). This information is
7236then stored in a local database that keeps track of the relationship between
7237transcripts, exons, cds and genes. Flexible methods are provided for
7238extracting the desired features in a convenient format.")
7239 (license license:artistic2.0)))
7240
fb25d880
RW		 7241(define-public r-go-db
7242 (package
7243 (name "r-go-db")
d1b1587c 7244 (version "3.4.0")
fb25d880
RW		 7245 (source (origin
7246 (method url-fetch)
5713bbf1 7247 (uri (string-append "https://www.bioconductor.org/packages/"
f82c8c3c
PP		 7248 "release/data/annotation/src/contrib/GO.db_"
7249 version ".tar.gz"))
fb25d880
RW		 7250 (sha256
7251 (base32
d1b1587c 7252 "02cj8kqi5w39jwcs8gp1dgj08sah262ppxnkz4h3qd0w191y8yyl"))))
fb25d880
RW		 7253 (properties
7254 `((upstream-name . "GO.db")))
7255 (build-system r-build-system)
3141b83d
RW		 7256 (propagated-inputs
7257 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7258 (home-page "https://bioconductor.org/packages/GO.db")
fb25d880
RW		 7259 (synopsis "Annotation maps describing the entire Gene Ontology")
7260 (description
7261 "The purpose of this GO.db annotation package is to provide detailed
7262information about the latest version of the Gene Ontologies.")
7263 (license license:artistic2.0)))
7264
d1dbde6a
RW		 7265(define-public r-graph
7266 (package
7267 (name "r-graph")
56373771 7268 (version "1.54.0")
d1dbde6a
RW		 7269 (source (origin
7270 (method url-fetch)
7271 (uri (bioconductor-uri "graph" version))
7272 (sha256
7273 (base32
56373771 7274 "0hx9wslnrci7c5i1gd1zlpjmgrkdnx9gabfgmzzwfnykk1mdvvna"))))
d1dbde6a
RW		 7275 (build-system r-build-system)
7276 (propagated-inputs
7277 `(("r-biocgenerics" ,r-biocgenerics)))
5713bbf1 7278 (home-page "https://bioconductor.org/packages/graph")
d1dbde6a
RW		 7279 (synopsis "Handle graph data structures in R")
7280 (description
7281 "This package implements some simple graph handling capabilities for R.")
7282 (license license:artistic2.0)))
7283
d547ce5e
RW		 7284(define-public r-topgo
7285 (package
7286 (name "r-topgo")
6e70b6d6 7287 (version "2.28.0")
d547ce5e
RW		 7288 (source (origin
7289 (method url-fetch)
7290 (uri (bioconductor-uri "topGO" version))
7291 (sha256
7292 (base32
6e70b6d6 7293 "04kvxz9qsxfz0np7wlrzp4r95nykz94x7mqpyyk76f90g6m66vcc"))))
d547ce5e
RW		 7294 (properties
7295 `((upstream-name . "topGO")))
7296 (build-system r-build-system)
7297 (propagated-inputs
7298 `(("r-annotationdbi" ,r-annotationdbi)
30ec4de7 7299 ("r-dbi" ,r-dbi)
d547ce5e
RW		 7300 ("r-biobase" ,r-biobase)
7301 ("r-biocgenerics" ,r-biocgenerics)
7302 ("r-go-db" ,r-go-db)
6d415db2 7303 ("r-graph" ,r-graph)
aeb64f3c
RW		 7304 ("r-lattice" ,r-lattice)
7305 ("r-matrixstats" ,r-matrixstats)
d547ce5e 7306 ("r-sparsem" ,r-sparsem)))
5713bbf1 7307 (home-page "https://bioconductor.org/packages/topGO")
d547ce5e
RW		 7308 (synopsis "Enrichment analysis for gene ontology")
7309 (description
7310 "The topGO package provides tools for testing @dfn{gene ontology} (GO)
7311terms while accounting for the topology of the GO graph. Different test
7312statistics and different methods for eliminating local similarities and
7313dependencies between GO terms can be implemented and applied.")
7314 ;; Any version of the LGPL applies.
7315 (license license:lgpl2.1+)))
7316
c63cef66
RW		 7317(define-public r-bsgenome
7318 (package
7319 (name "r-bsgenome")
c54a702a 7320 (version "1.44.0")
c63cef66
RW		 7321 (source (origin
7322 (method url-fetch)
7323 (uri (bioconductor-uri "BSgenome" version))
7324 (sha256
7325 (base32
c54a702a 7326 "18dlknwk3xvi52hamxf9pl5bjc5806mpw98xwvx5xajn8mrxiy5h"))))
c63cef66
RW		 7327 (properties
7328 `((upstream-name . "BSgenome")))
7329 (build-system r-build-system)
7330 (propagated-inputs
7331 `(("r-biocgenerics" ,r-biocgenerics)
7332 ("r-biostrings" ,r-biostrings)
7333 ("r-genomeinfodb" ,r-genomeinfodb)
7334 ("r-genomicranges" ,r-genomicranges)
7335 ("r-iranges" ,r-iranges)
7336 ("r-rsamtools" ,r-rsamtools)
7337 ("r-rtracklayer" ,r-rtracklayer)
7338 ("r-s4vectors" ,r-s4vectors)
7339 ("r-xvector" ,r-xvector)))
5713bbf1 7340 (home-page "https://bioconductor.org/packages/BSgenome")
c63cef66
RW		 7341 (synopsis "Infrastructure for Biostrings-based genome data packages")
7342 (description
7343 "This package provides infrastructure shared by all Biostrings-based
7344genome data packages and support for efficient SNP representation.")
7345 (license license:artistic2.0)))
7346
aa3eeeb5
RJ		 7347(define-public r-bsgenome-hsapiens-1000genomes-hs37d5
7348 (package
7349 (name "r-bsgenome-hsapiens-1000genomes-hs37d5")
7350 (version "0.99.1")
7351 (source (origin
7352 (method url-fetch)
7353 ;; We cannot use bioconductor-uri here because this tarball is
7354 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7355 (uri (string-append "https://www.bioconductor.org/packages/"
aa3eeeb5
RJ		 7356 "release/data/annotation/src/contrib/"
7357 "BSgenome.Hsapiens.1000genomes.hs37d5_"
7358 version ".tar.gz"))
7359 (sha256
7360 (base32
7361 "1cg0g5fqmsvwyw2p9hp2yy4ilk21jkbbrnpgqvb5c36ihjwvc7sr"))))
7362 (properties
7363 `((upstream-name . "BSgenome.Hsapiens.1000genomes.hs37d5")))
7364 (build-system r-build-system)
7365 ;; As this package provides little more than a very large data file it
7366 ;; doesn't make sense to build substitutes.
7367 (arguments `(#:substitutable? #f))
7368 (propagated-inputs
7369 `(("r-bsgenome" ,r-bsgenome)))
7370 (home-page
5713bbf1 7371 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.1000genomes.hs37d5/")
aa3eeeb5
RJ		 7372 (synopsis "Full genome sequences for Homo sapiens")
7373 (description
7374 "This package provides full genome sequences for Homo sapiens from
73751000genomes phase2 reference genome sequence (hs37d5), based on NCBI GRCh37.")
7376 (license license:artistic2.0)))
7377
c43a011d
RW		 7378(define-public r-impute
7379 (package
7380 (name "r-impute")
5999359f 7381 (version "1.50.0")
c43a011d
RW		 7382 (source (origin
7383 (method url-fetch)
7384 (uri (bioconductor-uri "impute" version))
7385 (sha256
7386 (base32
5999359f 7387 "0va45vfixy3np549md87h3b3rbavm54gfnmnjhpr9hf02lll6zxs"))))
c43a011d
RW		 7388 (inputs
7389 `(("gfortran" ,gfortran)))
7390 (build-system r-build-system)
5713bbf1 7391 (home-page "https://bioconductor.org/packages/impute")
c43a011d
RW		 7392 (synopsis "Imputation for microarray data")
7393 (description
7394 "This package provides a function to impute missing gene expression
7395microarray data, using nearest neighbor averaging.")
7396 (license license:gpl2+)))
7397
03ea5a35
RW		 7398(define-public r-seqpattern
7399 (package
7400 (name "r-seqpattern")
08b9224d 7401 (version "1.8.0")
03ea5a35
RW		 7402 (source (origin
7403 (method url-fetch)
7404 (uri (bioconductor-uri "seqPattern" version))
7405 (sha256
7406 (base32
08b9224d 7407 "0yw5251sirv3283xgmfmfbf900xp8dwvqfk88mr945s3by5hx99v"))))
03ea5a35
RW		 7408 (properties
7409 `((upstream-name . "seqPattern")))
7410 (build-system r-build-system)
7411 (propagated-inputs
7412 `(("r-biostrings" ,r-biostrings)
7413 ("r-genomicranges" ,r-genomicranges)
7414 ("r-iranges" ,r-iranges)
e92dd6f5 7415 ("r-kernsmooth" ,r-kernsmooth)
03ea5a35 7416 ("r-plotrix" ,r-plotrix)))
5713bbf1 7417 (home-page "https://bioconductor.org/packages/seqPattern")
03ea5a35
RW		 7418 (synopsis "Visualising oligonucleotide patterns and motif occurrences")
7419 (description
7420 "This package provides tools to visualize oligonucleotide patterns and
7421sequence motif occurrences across a large set of sequences centred at a common
7422reference point and sorted by a user defined feature.")
7423 (license license:gpl3+)))
7424
cb933df6
RW		 7425(define-public r-genomation
7426 (package
7427 (name "r-genomation")
3ae82c8d 7428 (version "1.8.0")
cb933df6
RW		 7429 (source (origin
7430 (method url-fetch)
7431 (uri (bioconductor-uri "genomation" version))
7432 (sha256
7433 (base32
3ae82c8d 7434 "0r71nf8p3aq3yffqxg4yir1zmm7rar5a57nmy1kpqpn8kaf2svjl"))))
cb933df6
RW		 7435 (build-system r-build-system)
7436 (propagated-inputs
7437 `(("r-biostrings" ,r-biostrings)
7438 ("r-bsgenome" ,r-bsgenome)
7439 ("r-data-table" ,r-data-table)
7440 ("r-genomeinfodb" ,r-genomeinfodb)
7441 ("r-genomicalignments" ,r-genomicalignments)
7442 ("r-genomicranges" ,r-genomicranges)
7443 ("r-ggplot2" ,r-ggplot2)
7444 ("r-gridbase" ,r-gridbase)
7445 ("r-impute" ,r-impute)
7446 ("r-iranges" ,r-iranges)
7447 ("r-matrixstats" ,r-matrixstats)
7448 ("r-plotrix" ,r-plotrix)
7449 ("r-plyr" ,r-plyr)
51c3c490 7450 ("r-rcpp" ,r-rcpp)
cb933df6
RW		 7451 ("r-readr" ,r-readr)
7452 ("r-reshape2" ,r-reshape2)
51c3c490 7453 ("r-rhtslib" ,r-rhtslib)
cb933df6
RW		 7454 ("r-rsamtools" ,r-rsamtools)
7455 ("r-rtracklayer" ,r-rtracklayer)
51c3c490
RW		 7456 ("r-runit" ,r-runit)
7457 ("r-s4vectors" ,r-s4vectors)
cb933df6 7458 ("r-seqpattern" ,r-seqpattern)))
51c3c490
RW		 7459 (inputs
7460 `(("zlib" ,zlib)))
cb933df6
RW		 7461 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7462 (synopsis "Summary, annotation and visualization of genomic data")
7463 (description
7464 "This package provides a package for summary and annotation of genomic
7465intervals. Users can visualize and quantify genomic intervals over
7466pre-defined functional regions, such as promoters, exons, introns, etc. The
7467genomic intervals represent regions with a defined chromosome position, which
7468may be associated with a score, such as aligned reads from HT-seq experiments,
7469TF binding sites, methylation scores, etc. The package can use any tabular
7470genomic feature data as long as it has minimal information on the locations of
7471genomic intervals. In addition, it can use BAM or BigWig files as input.")
7472 (license license:artistic2.0)))
7473
64efa307
RW		 7474(define-public r-genomationdata
7475 (package
7476 (name "r-genomationdata")
59198f8f 7477 (version "1.6.0")
64efa307
RW		 7478 (source (origin
7479 (method url-fetch)
7480 ;; We cannot use bioconductor-uri here because this tarball is
7481 ;; located under "data/annotation/" instead of "bioc/".
7482 (uri (string-append "https://bioconductor.org/packages/"
7483 "release/data/experiment/src/contrib/"
7484 "genomationData_" version ".tar.gz"))
7485 (sha256
7486 (base32
59198f8f 7487 "16dqwb7wx1igx77zdbcskx5m1hs4g4gp2hl56zzm70hcagnlkz8y"))))
64efa307
RW		 7488 (build-system r-build-system)
7489 ;; As this package provides little more than large data files, it doesn't
7490 ;; make sense to build substitutes.
7491 (arguments `(#:substitutable? #f))
7492 (native-inputs
7493 `(("r-knitr" ,r-knitr)))
7494 (home-page "http://bioinformatics.mdc-berlin.de/genomation/")
7495 (synopsis "Experimental data for use with the genomation package")
7496 (description
7497 "This package contains experimental genetic data for use with the
7498genomation package. Included are Chip Seq, Methylation and Cage data,
7499downloaded from Encode.")
7500 (license license:gpl3+)))
7501
486da491
RW		 7502(define-public r-org-hs-eg-db
7503 (package
7504 (name "r-org-hs-eg-db")
83f9a6fb 7505 (version "3.4.0")
486da491
RW		 7506 (source (origin
7507 (method url-fetch)
7508 ;; We cannot use bioconductor-uri here because this tarball is
7509 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7510 (uri (string-append "https://www.bioconductor.org/packages/"
486da491
RW		 7511 "release/data/annotation/src/contrib/"
7512 "org.Hs.eg.db_" version ".tar.gz"))
7513 (sha256
7514 (base32
83f9a6fb 7515 "19mg64pw8zcvb9yxzzyf7caz1kvdrkfsj1hd84bzq7crrh8kc4y6"))))
486da491
RW		 7516 (properties
7517 `((upstream-name . "org.Hs.eg.db")))
7518 (build-system r-build-system)
7519 (propagated-inputs
7520 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7521 (home-page "https://www.bioconductor.org/packages/org.Hs.eg.db/")
486da491
RW		 7522 (synopsis "Genome wide annotation for Human")
7523 (description
7524 "This package provides mappings from Entrez gene identifiers to various
7525annotations for the human genome.")
7526 (license license:artistic2.0)))
7527
fefedf98
RW		 7528(define-public r-org-ce-eg-db
7529 (package
7530 (name "r-org-ce-eg-db")
f04a3eff 7531 (version "3.4.0")
fefedf98
RW		 7532 (source (origin
7533 (method url-fetch)
7534 ;; We cannot use bioconductor-uri here because this tarball is
7535 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7536 (uri (string-append "https://www.bioconductor.org/packages/"
fefedf98
RW		 7537 "release/data/annotation/src/contrib/"
7538 "org.Ce.eg.db_" version ".tar.gz"))
7539 (sha256
7540 (base32
f04a3eff 7541 "12llfzrrc09kj2wzbisdspv38qzkzgpsbn8kv7qkwg746k3pq436"))))
fefedf98
RW		 7542 (properties
7543 `((upstream-name . "org.Ce.eg.db")))
7544 (build-system r-build-system)
7545 (propagated-inputs
7546 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7547 (home-page "https://www.bioconductor.org/packages/org.Ce.eg.db/")
fefedf98
RW		 7548 (synopsis "Genome wide annotation for Worm")
7549 (description
7550 "This package provides mappings from Entrez gene identifiers to various
7551annotations for the genome of the model worm Caenorhabditis elegans.")
7552 (license license:artistic2.0)))
7553
16c53a1e
RW		 7554(define-public r-org-dm-eg-db
7555 (package
7556 (name "r-org-dm-eg-db")
0cfac6f2 7557 (version "3.4.0")
16c53a1e
RW		 7558 (source (origin
7559 (method url-fetch)
7560 ;; We cannot use bioconductor-uri here because this tarball is
7561 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7562 (uri (string-append "https://www.bioconductor.org/packages/"
16c53a1e
RW		 7563 "release/data/annotation/src/contrib/"
7564 "org.Dm.eg.db_" version ".tar.gz"))
7565 (sha256
7566 (base32
0cfac6f2 7567 "1vzbphbrh1cf7xi5cksia9xy9a9l42js2z2qsajvjxvddiphrb7j"))))
16c53a1e
RW		 7568 (properties
7569 `((upstream-name . "org.Dm.eg.db")))
7570 (build-system r-build-system)
7571 (propagated-inputs
7572 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7573 (home-page "https://www.bioconductor.org/packages/org.Dm.eg.db/")
16c53a1e
RW		 7574 (synopsis "Genome wide annotation for Fly")
7575 (description
7576 "This package provides mappings from Entrez gene identifiers to various
7577annotations for the genome of the model fruit fly Drosophila melanogaster.")
7578 (license license:artistic2.0)))
7579
e761beb9
RW		 7580(define-public r-org-mm-eg-db
7581 (package
7582 (name "r-org-mm-eg-db")
7c45670b 7583 (version "3.4.0")
e761beb9
RW		 7584 (source (origin
7585 (method url-fetch)
7586 ;; We cannot use bioconductor-uri here because this tarball is
7587 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7588 (uri (string-append "https://www.bioconductor.org/packages/"
e761beb9
RW		 7589 "release/data/annotation/src/contrib/"
7590 "org.Mm.eg.db_" version ".tar.gz"))
7591 (sha256
7592 (base32
7c45670b 7593 "1lykjqjaf01fmgg3cvfcvwd5xjq6zc5vbxnm5r4l32fzvl89q50c"))))
e761beb9
RW		 7594 (properties
7595 `((upstream-name . "org.Mm.eg.db")))
7596 (build-system r-build-system)
7597 (propagated-inputs
7598 `(("r-annotationdbi" ,r-annotationdbi)))
5713bbf1 7599 (home-page "https://www.bioconductor.org/packages/org.Mm.eg.db/")
e761beb9
RW		 7600 (synopsis "Genome wide annotation for Mouse")
7601 (description
7602 "This package provides mappings from Entrez gene identifiers to various
7603annotations for the genome of the model mouse Mus musculus.")
7604 (license license:artistic2.0)))
7605
936e7d67
RW		 7606(define-public r-seqlogo
7607 (package
7608 (name "r-seqlogo")
0fce7d9b 7609 (version "1.42.0")
936e7d67
RW		 7610 (source
7611 (origin
7612 (method url-fetch)
7613 (uri (bioconductor-uri "seqLogo" version))
7614 (sha256
7615 (base32
0fce7d9b 7616 "19d5zmy7m8svljwgbmrb4vxkq18slq0f3all6k2ayv42b8w44h6q"))))
936e7d67
RW		 7617 (properties `((upstream-name . "seqLogo")))
7618 (build-system r-build-system)
5713bbf1 7619 (home-page "https://bioconductor.org/packages/seqLogo")
936e7d67
RW		 7620 (synopsis "Sequence logos for DNA sequence alignments")
7621 (description
7622 "seqLogo takes the position weight matrix of a DNA sequence motif and
7623plots the corresponding sequence logo as introduced by Schneider and
7624Stephens (1990).")
7625 (license license:lgpl2.0+)))
7626
c90a4baf
RW		 7627(define-public r-bsgenome-hsapiens-ucsc-hg19
7628 (package
7629 (name "r-bsgenome-hsapiens-ucsc-hg19")
7630 (version "1.4.0")
7631 (source (origin
7632 (method url-fetch)
7633 ;; We cannot use bioconductor-uri here because this tarball is
7634 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7635 (uri (string-append "https://www.bioconductor.org/packages/"
c90a4baf
RW		 7636 "release/data/annotation/src/contrib/"
7637 "BSgenome.Hsapiens.UCSC.hg19_"
7638 version ".tar.gz"))
7639 (sha256
7640 (base32
9d217d27 7641 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
c90a4baf
RW		 7642 (properties
7643 `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
7644 (build-system r-build-system)
7645 ;; As this package provides little more than a very large data file it
7646 ;; doesn't make sense to build substitutes.
7647 (arguments `(#:substitutable? #f))
7648 (propagated-inputs
7649 `(("r-bsgenome" ,r-bsgenome)))
7650 (home-page
5713bbf1 7651 "https://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
c90a4baf
RW		 7652 (synopsis "Full genome sequences for Homo sapiens")
7653 (description
7654 "This package provides full genome sequences for Homo sapiens as provided
7655by UCSC (hg19, February 2009) and stored in Biostrings objects.")
7656 (license license:artistic2.0)))
7657
a3e90287
RW		 7658(define-public r-bsgenome-mmusculus-ucsc-mm9
7659 (package
7660 (name "r-bsgenome-mmusculus-ucsc-mm9")
7661 (version "1.4.0")
7662 (source (origin
7663 (method url-fetch)
7664 ;; We cannot use bioconductor-uri here because this tarball is
7665 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7666 (uri (string-append "https://www.bioconductor.org/packages/"
a3e90287
RW		 7667 "release/data/annotation/src/contrib/"
7668 "BSgenome.Mmusculus.UCSC.mm9_"
7669 version ".tar.gz"))
7670 (sha256
7671 (base32
7672 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
7673 (properties
7674 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
7675 (build-system r-build-system)
7676 ;; As this package provides little more than a very large data file it
7677 ;; doesn't make sense to build substitutes.
7678 (arguments `(#:substitutable? #f))
7679 (propagated-inputs
7680 `(("r-bsgenome" ,r-bsgenome)))
7681 (home-page
5713bbf1 7682 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
a3e90287
RW		 7683 (synopsis "Full genome sequences for Mouse")
7684 (description
7685 "This package provides full genome sequences for Mus musculus (Mouse) as
7686provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
7687 (license license:artistic2.0)))
7688
4714d521
RW		 7689(define-public r-bsgenome-mmusculus-ucsc-mm10
7690 (package
7691 (name "r-bsgenome-mmusculus-ucsc-mm10")
7692 (version "1.4.0")
7693 (source (origin
7694 (method url-fetch)
7695 ;; We cannot use bioconductor-uri here because this tarball is
7696 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7697 (uri (string-append "https://www.bioconductor.org/packages/"
4714d521
RW		 7698 "release/data/annotation/src/contrib/"
7699 "BSgenome.Mmusculus.UCSC.mm10_"
7700 version ".tar.gz"))
7701 (sha256
7702 (base32
7703 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
7704 (properties
7705 `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
7706 (build-system r-build-system)
7707 ;; As this package provides little more than a very large data file it
7708 ;; doesn't make sense to build substitutes.
7709 (arguments `(#:substitutable? #f))
7710 (propagated-inputs
7711 `(("r-bsgenome" ,r-bsgenome)))
7712 (home-page
5713bbf1 7713 "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
4714d521
RW		 7714 (synopsis "Full genome sequences for Mouse")
7715 (description
7716 "This package provides full genome sequences for Mus
7717musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
7718in Biostrings objects.")
7719 (license license:artistic2.0)))
7720
c5173d74
RJ		 7721(define-public r-txdb-mmusculus-ucsc-mm10-knowngene
7722 (package
7723 (name "r-txdb-mmusculus-ucsc-mm10-knowngene")
7724 (version "3.4.0")
7725 (source (origin
7726 (method url-fetch)
7727 ;; We cannot use bioconductor-uri here because this tarball is
7728 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7729 (uri (string-append "https://www.bioconductor.org/packages/"
c5173d74
RJ		 7730 "release/data/annotation/src/contrib/"
7731 "TxDb.Mmusculus.UCSC.mm10.knownGene_"
7732 version ".tar.gz"))
7733 (sha256
7734 (base32
7735 "08gava9wsvpcqz51k2sni3pj03n5155v32d9riqbf305nbirqbkb"))))
7736 (properties
7737 `((upstream-name . "TxDb.Mmusculus.UCSC.mm10.knownGene")))
7738 (build-system r-build-system)
7739 ;; As this package provides little more than a very large data file it
7740 ;; doesn't make sense to build substitutes.
7741 (arguments `(#:substitutable? #f))
7742 (propagated-inputs
7743 `(("r-bsgenome" ,r-bsgenome)
7744 ("r-genomicfeatures" ,r-genomicfeatures)
7745 ("r-annotationdbi" ,r-annotationdbi)))
7746 (home-page
5713bbf1 7747 "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm10.knownGene/")
c5173d74
RJ		 7748 (synopsis "Annotation package for TxDb knownGene object(s) for Mouse")
7749 (description
7750 "This package loads a TxDb object, which is an R interface to
7751prefabricated databases contained in this package. This package provides
7752the TxDb object of Mouse data as provided by UCSC (mm10, December 2011)
7753based on the knownGene track.")
7754 (license license:artistic2.0)))
7755
943bd627
RW		 7756(define-public r-bsgenome-celegans-ucsc-ce6
7757 (package
7758 (name "r-bsgenome-celegans-ucsc-ce6")
7759 (version "1.4.0")
7760 (source (origin
7761 (method url-fetch)
7762 ;; We cannot use bioconductor-uri here because this tarball is
7763 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7764 (uri (string-append "https://www.bioconductor.org/packages/"
943bd627
RW		 7765 "release/data/annotation/src/contrib/"
7766 "BSgenome.Celegans.UCSC.ce6_"
7767 version ".tar.gz"))
7768 (sha256
7769 (base32
7770 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
7771 (properties
7772 `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
7773 (build-system r-build-system)
7774 ;; As this package provides little more than a very large data file it
7775 ;; doesn't make sense to build substitutes.
7776 (arguments `(#:substitutable? #f))
7777 (propagated-inputs
7778 `(("r-bsgenome" ,r-bsgenome)))
7779 (home-page
5713bbf1 7780 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
943bd627
RW		 7781 (synopsis "Full genome sequences for Worm")
7782 (description
7783 "This package provides full genome sequences for Caenorhabditis
7784elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
7785objects.")
7786 (license license:artistic2.0)))
7787
fc47c7d6
RW		 7788(define-public r-bsgenome-celegans-ucsc-ce10
7789 (package
7790 (name "r-bsgenome-celegans-ucsc-ce10")
7791 (version "1.4.0")
7792 (source (origin
7793 (method url-fetch)
7794 ;; We cannot use bioconductor-uri here because this tarball is
7795 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7796 (uri (string-append "https://www.bioconductor.org/packages/"
fc47c7d6
RW		 7797 "release/data/annotation/src/contrib/"
7798 "BSgenome.Celegans.UCSC.ce10_"
7799 version ".tar.gz"))
7800 (sha256
7801 (base32
7802 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
7803 (properties
7804 `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
7805 (build-system r-build-system)
7806 ;; As this package provides little more than a very large data file it
7807 ;; doesn't make sense to build substitutes.
7808 (arguments `(#:substitutable? #f))
7809 (propagated-inputs
7810 `(("r-bsgenome" ,r-bsgenome)))
7811 (home-page
5713bbf1 7812 "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
fc47c7d6
RW		 7813 (synopsis "Full genome sequences for Worm")
7814 (description
7815 "This package provides full genome sequences for Caenorhabditis
7816elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
7817objects.")
7818 (license license:artistic2.0)))
7819
6dc60998
RW		 7820(define-public r-bsgenome-dmelanogaster-ucsc-dm3
7821 (package
7822 (name "r-bsgenome-dmelanogaster-ucsc-dm3")
7823 (version "1.4.0")
7824 (source (origin
7825 (method url-fetch)
7826 ;; We cannot use bioconductor-uri here because this tarball is
7827 ;; located under "data/annotation/" instead of "bioc/".
5713bbf1 7828 (uri (string-append "https://www.bioconductor.org/packages/"
6dc60998
RW		 7829 "release/data/annotation/src/contrib/"
7830 "BSgenome.Dmelanogaster.UCSC.dm3_"
7831 version ".tar.gz"))
7832 (sha256
7833 (base32
7834 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
7835 (properties
7836 `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
7837 (build-system r-build-system)
7838 ;; As this package provides little more than a very large data file it
7839 ;; doesn't make sense to build substitutes.
7840 (arguments `(#:substitutable? #f))
7841 (propagated-inputs
7842 `(("r-bsgenome" ,r-bsgenome)))
7843 (home-page
5713bbf1 7844 "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
6dc60998
RW		 7845 (synopsis "Full genome sequences for Fly")
7846 (description
7847 "This package provides full genome sequences for Drosophila
7848melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
7849Biostrings objects.")
7850 (license license:artistic2.0)))
7851
ae2462f7
RW		 7852(define-public r-motifrg
7853 (package
7854 (name "r-motifrg")
a7b90ea9 7855 (version "1.20.0")
ae2462f7
RW		 7856 (source
7857 (origin
7858 (method url-fetch)
7859 (uri (bioconductor-uri "motifRG" version))
7860 (sha256
7861 (base32
a7b90ea9 7862 "0lxxpqyljiyk73vyq6ss46q13g81pm24q87mkzdsqlr5zx427ch6"))))
ae2462f7
RW		 7863 (properties `((upstream-name . "motifRG")))
7864 (build-system r-build-system)
7865 (propagated-inputs
7866 `(("r-biostrings" ,r-biostrings)
7867 ("r-bsgenome" ,r-bsgenome)
7868 ("r-bsgenome.hsapiens.ucsc.hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7869 ("r-iranges" ,r-iranges)
7870 ("r-seqlogo" ,r-seqlogo)
7871 ("r-xvector" ,r-xvector)))
5713bbf1 7872 (home-page "https://bioconductor.org/packages/motifRG")
ae2462f7
RW		 7873 (synopsis "Discover motifs in high throughput sequencing data")
7874 (description
7875 "This package provides tools for discriminative motif discovery in high
7876throughput genetic sequencing data sets using regression methods.")
7877 (license license:artistic2.0)))
7878
a5002ae7
AE		 7879(define-public r-qtl
7880 (package
7881 (name "r-qtl")
c8a9b2bc 7882 (version "1.41-6")
a5002ae7
AE		 7883 (source
7884 (origin
7885 (method url-fetch)
7886 (uri (string-append "mirror://cran/src/contrib/qtl_"
7887 version ".tar.gz"))
7888 (sha256
7889 (base32
c8a9b2bc 7890 "067az4v432zxp6lxck8d7vlh9w4r13r0mvw5zsglyaqwsh3d9sad"))))
a5002ae7
AE		 7891 (build-system r-build-system)
7892 (home-page "http://rqtl.org/")
7893 (synopsis "R package for analyzing QTL experiments in genetics")
7894 (description "R/qtl is an extension library for the R statistics
7895system. It is used to analyze experimental crosses for identifying
7896genes contributing to variation in quantitative traits (so-called
7897quantitative trait loci, QTLs).
7898
7899Using a hidden Markov model, R/qtl allows to estimate genetic maps, to
7900identify genotyping errors, and to perform single-QTL and two-QTL,
7901two-dimensional genome scans.")
7902 (license license:gpl3)))
d1e32822 7903
9e3ba31c
RJ		 7904(define-public r-zlibbioc
7905 (package
7906 (name "r-zlibbioc")
8444f606 7907 (version "1.22.0")
9e3ba31c
RJ		 7908 (source (origin
7909 (method url-fetch)
7910 (uri (bioconductor-uri "zlibbioc" version))
7911 (sha256
7912 (base32
8444f606 7913 "1kdgwwlh39mgwzj3zq71za4iv40sq625ghnyrndv5wrivrcr2igv"))))
9e3ba31c
RJ		 7914 (properties
7915 `((upstream-name . "zlibbioc")))
7916 (build-system r-build-system)
7917 (home-page "https://bioconductor.org/packages/zlibbioc")
7918 (synopsis "Provider for zlib-1.2.5 to R packages")
7919 (description "This package uses the source code of zlib-1.2.5 to create
7920libraries for systems that do not have these available via other means.")
7921 (license license:artistic2.0)))
7922
e619a5c2
RW		 7923(define-public r-r4rna
7924 (package
7925 (name "r-r4rna")
7926 (version "0.1.4")
7927 (source
7928 (origin
7929 (method url-fetch)
7930 (uri (string-append "http://www.e-rna.org/r-chie/files/R4RNA_"
7931 version ".tar.gz"))
7932 (sha256
7933 (base32
7934 "1p0i78wh76jfgmn9jphbwwaz6yy6pipzfg08xs54cxavxg2j81p5"))))
7935 (build-system r-build-system)
7936 (propagated-inputs
7937 `(("r-optparse" ,r-optparse)
7938 ("r-rcolorbrewer" ,r-rcolorbrewer)))
7939 (home-page "http://www.e-rna.org/r-chie/index.cgi")
7940 (synopsis "Analysis framework for RNA secondary structure")
7941 (description
7942 "The R4RNA package aims to be a general framework for the analysis of RNA
7943secondary structure and comparative analysis in R.")
7944 (license license:gpl3+)))
7945
52765a63
RW		 7946(define-public r-rhtslib
7947 (package
7948 (name "r-rhtslib")
37631ddc 7949 (version "1.8.0")
52765a63
RW		 7950 (source
7951 (origin
7952 (method url-fetch)
7953 (uri (bioconductor-uri "Rhtslib" version))
7954 (sha256
7955 (base32
37631ddc 7956 "0jj5h81z5gyf4j3lp2l3zsm6pgbmalgrngr8qdpygc20phndpp0b"))))
52765a63
RW		 7957 (properties `((upstream-name . "Rhtslib")))
7958 (build-system r-build-system)
7959 (propagated-inputs
7960 `(("r-zlibbioc" ,r-zlibbioc)))
7961 (inputs
7962 `(("zlib" ,zlib)))
53ca52f0
RW		 7963 (native-inputs
7964 `(("autoconf" ,autoconf)))
52765a63
RW		 7965 (home-page "https://github.com/nhayden/Rhtslib")
7966 (synopsis "High-throughput sequencing library as an R package")
7967 (description
7968 "This package provides the HTSlib C library for high-throughput
7969nucleotide sequence analysis. The package is primarily useful to developers
7970of other R packages who wish to make use of HTSlib.")
7971 (license license:lgpl2.0+)))
7972
fe02c4c9
RW		 7973(define-public r-bamsignals
7974 (package
7975 (name "r-bamsignals")
ccd2aade 7976 (version "1.8.0")
fe02c4c9
RW		 7977 (source
7978 (origin
7979 (method url-fetch)
7980 (uri (bioconductor-uri "bamsignals" version))
7981 (sha256
7982 (base32
ccd2aade 7983 "0knx69zzdaak2sjim8k9mygmcjxpa705m8013ld5zwpgi8dag9mc"))))
fe02c4c9
RW		 7984 (build-system r-build-system)
7985 (propagated-inputs
7986 `(("r-biocgenerics" ,r-biocgenerics)
7987 ("r-genomicranges" ,r-genomicranges)
7988 ("r-iranges" ,r-iranges)
7989 ("r-rcpp" ,r-rcpp)
7990 ("r-rhtslib" ,r-rhtslib)
7991 ("r-zlibbioc" ,r-zlibbioc)))
7992 (inputs
7993 `(("zlib" ,zlib)))
5713bbf1 7994 (home-page "https://bioconductor.org/packages/bamsignals")
fe02c4c9
RW		 7995 (synopsis "Extract read count signals from bam files")
7996 (description
7997 "This package allows to efficiently obtain count vectors from indexed bam
7998files. It counts the number of nucleotide sequence reads in given genomic
7999ranges and it computes reads profiles and coverage profiles. It also handles
8000paired-end data.")
8001 (license license:gpl2+)))
8002
89984be4
RW		 8003(define-public r-rcas
8004 (package
8005 (name "r-rcas")
d82937fb 8006 (version "1.3.4")
89984be4
RW		 8007 (source (origin
8008 (method url-fetch)
8009 (uri (string-append "https://github.com/BIMSBbioinfo/RCAS/archive/v"
8010 version ".tar.gz"))
8011 (file-name (string-append name "-" version ".tar.gz"))
8012 (sha256
8013 (base32
d82937fb 8014 "1qgc7vi6fpzl440yg7jhiycg5q336kd4pxqzx10yx2zcq3bq3msg"))))
89984be4
RW		 8015 (build-system r-build-system)
8016 (native-inputs
8017 `(("r-knitr" ,r-knitr)
8018 ("r-testthat" ,r-testthat)
8019 ;; During vignette building knitr checks that "pandoc-citeproc"
8020 ;; is in the PATH.
8021 ("ghc-pandoc-citeproc" ,ghc-pandoc-citeproc)))
8022 (propagated-inputs
8023 `(("r-data-table" ,r-data-table)
8024 ("r-biomart" ,r-biomart)
8025 ("r-org-hs-eg-db" ,r-org-hs-eg-db)
8026 ("r-org-ce-eg-db" ,r-org-ce-eg-db)
8027 ("r-org-dm-eg-db" ,r-org-dm-eg-db)
8028 ("r-org-mm-eg-db" ,r-org-mm-eg-db)
8029 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
8030 ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
8031 ("r-bsgenome-celegans-ucsc-ce10" ,r-bsgenome-celegans-ucsc-ce10)
8032 ("r-bsgenome-dmelanogaster-ucsc-dm3" ,r-bsgenome-dmelanogaster-ucsc-dm3)
8033 ("r-topgo" ,r-topgo)
8034 ("r-dt" ,r-dt)
ebfd6a71 8035 ("r-pbapply" ,r-pbapply)
89984be4 8036 ("r-plotly" ,r-plotly)
2c8d6c0b 8037 ("r-plotrix" ,r-plotrix)
89984be4
RW		 8038 ("r-motifrg" ,r-motifrg)
8039 ("r-genomation" ,r-genomation)
8040 ("r-genomicfeatures" ,r-genomicfeatures)
8041 ("r-rtracklayer" ,r-rtracklayer)
8042 ("r-rmarkdown" ,r-rmarkdown)))
8043 (synopsis "RNA-centric annotation system")
8044 (description
8045 "RCAS aims to be a standalone RNA-centric annotation system that provides
8046intuitive reports and publication-ready graphics. This package provides the R
8047library implementing most of the pipeline's features.")
8048 (home-page "https://github.com/BIMSBbioinfo/RCAS")
75690c9f 8049 (license license:artistic2.0)))
89984be4 8050
50937297
RW		 8051(define-public rcas-web
8052 (package
8053 (name "rcas-web")
01d87d3c 8054 (version "0.0.4")
50937297
RW		 8055 (source
8056 (origin
8057 (method url-fetch)
8058 (uri (string-append "https://github.com/BIMSBbioinfo/rcas-web/"
8059 "releases/download/v" version
8060 "/rcas-web-" version ".tar.gz"))
8061 (sha256
8062 (base32
01d87d3c 8063 "1p16frfys41a8yaa4gkm457nzkqhqs2pc3lkac0ds457w9w5j1gm"))))
50937297
RW		 8064 (build-system gnu-build-system)
8065 (arguments
8066 `(#:phases
8067 (modify-phases %standard-phases
8068 (add-after 'install 'wrap-executable
8069 (lambda* (#:key inputs outputs #:allow-other-keys)
8070 (let* ((out (assoc-ref outputs "out"))
8071 (json (assoc-ref inputs "guile-json"))
8072 (redis (assoc-ref inputs "guile-redis"))
8073 (path (string-append
8074 json "/share/guile/site/2.2:"
8075 redis "/share/guile/site/2.2")))
8076 (wrap-program (string-append out "/bin/rcas-web")
8077 `("GUILE_LOAD_PATH" ":" = (,path))
8078 `("GUILE_LOAD_COMPILED_PATH" ":" = (,path))
8079 `("R_LIBS_SITE" ":" = (,(getenv "R_LIBS_SITE")))))
8080 #t)))))
8081 (inputs
2d7c4ae3 8082 `(("r-minimal" ,r-minimal)
50937297 8083 ("r-rcas" ,r-rcas)
f6396d86 8084 ("guile-next" ,guile-2.2)
2252f087 8085 ("guile-json" ,guile-json)
50937297
RW		 8086 ("guile-redis" ,guile2.2-redis)))
8087 (native-inputs
8088 `(("pkg-config" ,pkg-config)))
8089 (home-page "https://github.com/BIMSBbioinfo/rcas-web")
8090 (synopsis "Web interface for RNA-centric annotation system (RCAS)")
8091 (description "This package provides a simple web interface for the
8092@dfn{RNA-centric annotation system} (RCAS).")
8093 (license license:agpl3+)))
8094
7500e42b
RJ		 8095(define-public r-mutationalpatterns
8096 (package
8097 (name "r-mutationalpatterns")
b4584137 8098 (version "1.2.1")
7500e42b
RJ		 8099 (source
8100 (origin
8101 (method url-fetch)
8102 (uri (bioconductor-uri "MutationalPatterns" version))
8103 (sha256
8104 (base32
b4584137 8105 "1s50diwh1j6vg3mgahh6bczvq74mfdbmwjrad4d5lh723gnc5pjg"))))
7500e42b
RJ		 8106 (build-system r-build-system)
8107 (propagated-inputs
8108 `(("r-biocgenerics" ,r-biocgenerics)
8109 ("r-biostrings" ,r-biostrings)
cf4ac4e4
RJ		 8110 ("r-bsgenome-hsapiens-1000g" ,r-bsgenome-hsapiens-1000genomes-hs37d5)
8111 ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
7500e42b
RJ		 8112 ("r-genomicranges" ,r-genomicranges)
8113 ("r-genomeinfodb" ,r-genomeinfodb)
8114 ("r-ggplot2" ,r-ggplot2)
8115 ("r-gridextra" ,r-gridextra)
8116 ("r-iranges" ,r-iranges)
8117 ("r-nmf" ,r-nmf)
8118 ("r-plyr" ,r-plyr)
8119 ("r-pracma" ,r-pracma)
8120 ("r-reshape2" ,r-reshape2)
8121 ("r-summarizedexperiment" ,r-summarizedexperiment)
8122 ("r-variantannotation" ,r-variantannotation)))
5713bbf1 8123 (home-page "https://bioconductor.org/packages/MutationalPatterns/")
7500e42b
RJ		 8124 (synopsis "Extract and visualize mutational patterns in genomic data")
8125 (description "This package provides an extensive toolset for the
8126characterization and visualization of a wide range of mutational patterns
8127in SNV base substitution data.")
8128 (license license:expat)))
8129
d7160529
RW		 8130(define-public r-wgcna
8131 (package
8132 (name "r-wgcna")
1b22ecda 8133 (version "1.61")
d7160529
RW		 8134 (source
8135 (origin
8136 (method url-fetch)
8137 (uri (cran-uri "WGCNA" version))
8138 (sha256
8139 (base32
1b22ecda 8140 "1vrc2k33a196hrrl7k0z534fp96vv0shmigcr65ny1q0v6lq0h6i"))))
d7160529
RW		 8141 (properties `((upstream-name . "WGCNA")))
8142 (build-system r-build-system)
8143 (propagated-inputs
8144 `(("r-annotationdbi" ,r-annotationdbi)
8145 ("r-doparallel" ,r-doparallel)
8146 ("r-dynamictreecut" ,r-dynamictreecut)
8147 ("r-fastcluster" ,r-fastcluster)
8148 ("r-foreach" ,r-foreach)
8149 ("r-go-db" ,r-go-db)
8150 ("r-hmisc" ,r-hmisc)
8151 ("r-impute" ,r-impute)
1b22ecda
RW		 8152 ("r-rcpp" ,r-rcpp)
8153 ("r-robust" ,r-robust)
8154 ("r-survival" ,r-survival)
d7160529
RW		 8155 ("r-matrixstats" ,r-matrixstats)
8156 ("r-preprocesscore" ,r-preprocesscore)))
8157 (home-page
8158 "http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/Rpackages/WGCNA/")
8159 (synopsis "Weighted correlation network analysis")
8160 (description
8161 "This package provides functions necessary to perform Weighted
8162Correlation Network Analysis on high-dimensional data. It includes functions
8163for rudimentary data cleaning, construction and summarization of correlation
8164networks, module identification and functions for relating both variables and
8165modules to sample traits. It also includes a number of utility functions for
8166data manipulation and visualization.")
8167 (license license:gpl2+)))
8168
c827f202
RW		 8169(define-public r-chipkernels
8170 (let ((commit "c9cfcacb626b1221094fb3490ea7bac0fd625372")
8171 (revision "1"))
8172 (package
8173 (name "r-chipkernels")
8174 (version (string-append "1.1-" revision "." (string-take commit 9)))
8175 (source
8176 (origin
8177 (method git-fetch)
8178 (uri (git-reference
8179 (url "https://github.com/ManuSetty/ChIPKernels.git")
8180 (commit commit)))
8181 (file-name (string-append name "-" version))
8182 (sha256
8183 (base32
8184 "14bj5qhjm1hsm9ay561nfbqi9wxsa7y487df2idsaaf6z10nw4v0"))))
8185 (build-system r-build-system)
8186 (propagated-inputs
8187 `(("r-iranges" ,r-iranges)
8188 ("r-xvector" ,r-xvector)
8189 ("r-biostrings" ,r-biostrings)
8190 ("r-bsgenome" ,r-bsgenome)
8191 ("r-gtools" ,r-gtools)
8192 ("r-genomicranges" ,r-genomicranges)
8193 ("r-sfsmisc" ,r-sfsmisc)
8194 ("r-kernlab" ,r-kernlab)
8195 ("r-s4vectors" ,r-s4vectors)
8196 ("r-biocgenerics" ,r-biocgenerics)))
8197 (home-page "https://github.com/ManuSetty/ChIPKernels")
8198 (synopsis "Build string kernels for DNA Sequence analysis")
8199 (description "ChIPKernels is an R package for building different string
8200kernels used for DNA Sequence analysis. A dictionary of the desired kernel
8201must be built and this dictionary can be used for determining kernels for DNA
8202Sequences.")
8203 (license license:gpl2+))))
8204
2d9fb170
RW		 8205(define-public r-seqgl
8206 (package
8207 (name "r-seqgl")
8208 (version "1.1.4")
8209 (source
8210 (origin
8211 (method url-fetch)
8212 (uri (string-append "https://github.com/ManuSetty/SeqGL/"
8213 "archive/" version ".tar.gz"))
8214 (file-name (string-append name "-" version ".tar.gz"))
8215 (sha256
8216 (base32
8217 "0pnk1p3sci5yipyc8xnb6jbmydpl80fld927xgnbcv104hy8h8yh"))))
8218 (build-system r-build-system)
8219 (propagated-inputs
8220 `(("r-biostrings" ,r-biostrings)
8221 ("r-chipkernels" ,r-chipkernels)
8222 ("r-genomicranges" ,r-genomicranges)
8223 ("r-spams" ,r-spams)
8224 ("r-wgcna" ,r-wgcna)
8225 ("r-fastcluster" ,r-fastcluster)))
8226 (home-page "https://github.com/ManuSetty/SeqGL")
8227 (synopsis "Group lasso for Dnase/ChIP-seq data")
8228 (description "SeqGL is a group lasso based algorithm to extract
8229transcription factor sequence signals from ChIP, DNase and ATAC-seq profiles.
8230This package presents a method which uses group lasso to discriminate between
8231bound and non bound genomic regions to accurately identify transcription
8232factors bound at the specific regions.")
8233 (license license:gpl2+)))
8234
bd3be46e
RW		 8235(define-public r-gkmsvm
8236 (package
8237 (name "r-gkmsvm")
8238 (version "0.71.0")
8239 (source
8240 (origin
8241 (method url-fetch)
8242 (uri (cran-uri "gkmSVM" version))
8243 (sha256
8244 (base32
8245 "1zpxgxmf2nd5j5wn00ps6kfxr8wxh7d1swr1rr4spq7sj5z5z0k0"))))
8246 (properties `((upstream-name . "gkmSVM")))
8247 (build-system r-build-system)
8248 (propagated-inputs
8249 `(("r-biocgenerics" ,r-biocgenerics)
8250 ("r-biostrings" ,r-biostrings)
8251 ("r-genomeinfodb" ,r-genomeinfodb)
8252 ("r-genomicranges" ,r-genomicranges)
8253 ("r-iranges" ,r-iranges)
8254 ("r-kernlab" ,r-kernlab)
8255 ("r-rcpp" ,r-rcpp)
8256 ("r-rocr" ,r-rocr)
8257 ("r-rtracklayer" ,r-rtracklayer)
8258 ("r-s4vectors" ,r-s4vectors)
8259 ("r-seqinr" ,r-seqinr)))
8260 (home-page "http://cran.r-project.org/web/packages/gkmSVM")
8261 (synopsis "Gapped-kmer support vector machine")
8262 (description
8263 "This R package provides tools for training gapped-kmer SVM classifiers
8264for DNA and protein sequences. This package supports several sequence
8265kernels, including: gkmSVM, kmer-SVM, mismatch kernel and wildcard kernel.")
8266 (license license:gpl2+)))
8267
d4af25b5
RJPB		 8268(define-public r-tximport
8269 (package
8270 (name "r-tximport")
ff58a1af 8271 (version "1.4.0")
d4af25b5
RJPB		 8272 (source (origin
8273 (method url-fetch)
8274 (uri (bioconductor-uri "tximport" version))
8275 (sha256
8276 (base32
ff58a1af 8277 "1mklb02bj4gnbjlmb7vv6k4lr3w9fp3pzli9rddbrwd0y5n8fcpx"))))
d4af25b5 8278 (build-system r-build-system)
5713bbf1 8279 (home-page "https://bioconductor.org/packages/tximport")
d4af25b5
RJPB		 8280 (synopsis "Import and summarize transcript-level estimates for gene-level analysis")
8281 (description
8282 "This package provides tools to import transcript-level abundance,
8283estimated counts and transcript lengths, and to summarize them into matrices
8284for use with downstream gene-level analysis packages. Average transcript
8285length, weighted by sample-specific transcript abundance estimates, is
8286provided as a matrix which can be used as an offset for different expression
8287of gene-level counts.")
8288 (license license:gpl2+)))
8289
69f2b3bd
RJPB		 8290(define-public r-rhdf5
8291 (package
8292 (name "r-rhdf5")
04049a5f 8293 (version "2.20.0")
69f2b3bd
RJPB		 8294 (source (origin
8295 (method url-fetch)
8296 (uri (bioconductor-uri "rhdf5" version))
8297 (sha256
8298 (base32
04049a5f 8299 "1p6f5i6l44phl772a38x9cav2sya37bkqbkjzdc4pmyfzkv1j6hy"))))
69f2b3bd
RJPB		 8300 (build-system r-build-system)
8301 (arguments
8302 `(#:phases
8303 (modify-phases %standard-phases
8304 (add-after 'unpack 'unpack-smallhdf5
8305 (lambda* (#:key outputs #:allow-other-keys)
8306 (system* "tar" "-xzvf"
8307 "src/hdf5source/hdf5small.tgz" "-C" "src/" )
8308 (substitute* "src/Makevars"
8309 (("^.*cd hdf5source &&.*$") "")
8310 (("^.*gunzip -dc hdf5small.tgz.*$") "")
8311 (("^.*rm -rf hdf5.*$") "")
8312 (("^.*mv hdf5source/hdf5 ..*$") ""))
8313 (substitute* "src/hdf5/configure"
8314 (("/bin/mv") "mv"))
8315 #t)))))
8316 (propagated-inputs
8317 `(("r-zlibbioc" ,r-zlibbioc)))
8318 (inputs
8319 `(("perl" ,perl)
8320 ("zlib" ,zlib)))
5713bbf1 8321 (home-page "https://bioconductor.org/packages/rhdf5")
69f2b3bd
RJPB		 8322 (synopsis "HDF5 interface to R")
8323 (description
8324 "This R/Bioconductor package provides an interface between HDF5 and R.
8325HDF5's main features are the ability to store and access very large and/or
8326complex datasets and a wide variety of metadata on mass storage (disk) through
8327a completely portable file format. The rhdf5 package is thus suited for the
8328exchange of large and/or complex datasets between R and other software
8329package, and for letting R applications work on datasets that are larger than
8330the available RAM.")
8331 (license license:artistic2.0)))
8332
17cddc17
RW		 8333(define-public r-annotationfilter
8334 (package
8335 (name "r-annotationfilter")
8336 (version "1.0.0")
8337 (source (origin
8338 (method url-fetch)
8339 (uri (bioconductor-uri "AnnotationFilter" version))
8340 (sha256
8341 (base32
8342 "0pxvswjzwibdfmrkdragxmzcl844z73pmkn82z92wahwa6gjfyi7"))))
8343 (properties
8344 `((upstream-name . "AnnotationFilter")))
8345 (build-system r-build-system)
8346 (propagated-inputs
8347 `(("r-genomicranges" ,r-genomicranges)
8348 ("r-lazyeval" ,r-lazyeval)))
8349 (home-page "https://github.com/Bioconductor/AnnotationFilter")
8350 (synopsis "Facilities for filtering Bioconductor annotation resources")
8351 (description
8352 "This package provides classes and other infrastructure to implement
8353filters for manipulating Bioconductor annotation resources. The filters are
8354used by @code{ensembldb}, @code{Organism.dplyr}, and other packages.")
8355 (license license:artistic2.0)))
8356
66e40e00
RW		 8357(define-public emboss
8358 (package
8359 (name "emboss")
8360 (version "6.5.7")
8361 (source (origin
8362 (method url-fetch)
8363 (uri (string-append "ftp://emboss.open-bio.org/pub/EMBOSS/old/"
8364 (version-major+minor version) ".0/"
8365 "EMBOSS-" version ".tar.gz"))
8366 (sha256
8367 (base32
8368 "0vsmz96gc411yj2iyzdrsmg4l2n1nhgmp7vrgzlxx3xixv9xbf0q"))))
8369 (build-system gnu-build-system)
8370 (arguments
8371 `(#:configure-flags
8372 (list (string-append "--with-hpdf="
8373 (assoc-ref %build-inputs "libharu")))
8374 #:phases
8375 (modify-phases %standard-phases
8376 (add-after 'unpack 'fix-checks
8377 (lambda _
8378 ;; The PNGDRIVER tests check for the presence of libgd, libpng
8379 ;; and zlib, but assume that they are all found at the same
8380 ;; prefix.
8381 (substitute* "configure.in"
8382 (("CHECK_PNGDRIVER")
8383 "LIBS=\"$LIBS -lgd -lpng -lz -lm\"
8384AC_DEFINE([PLD_png], [1], [Define to 1 if PNG support is available])
8385AM_CONDITIONAL(AMPNG, true)"))
8386 #t))
d10092b8 8387 (add-after 'fix-checks 'disable-update-check
66e40e00
RW		 8388 (lambda _
8389 ;; At build time there is no connection to the Internet, so
8390 ;; looking for updates will not work.
8391 (substitute* "Makefile.am"
8392 (("\\$\\(bindir\\)/embossupdate") ""))
8393 #t))
d10092b8 8394 (add-after 'disable-update-check 'autogen
66e40e00
RW		 8395 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
8396 (inputs
8397 `(("perl" ,perl)
8398 ("libpng" ,libpng)
8399 ("gd" ,gd)
8400 ("libx11" ,libx11)
8401 ("libharu" ,libharu)
8402 ("zlib" ,zlib)))
8403 (native-inputs
8404 `(("autoconf" ,autoconf)
8405 ("automake" ,automake)
8406 ("libtool" ,libtool)
8407 ("pkg-config" ,pkg-config)))
8408 (home-page "http://emboss.sourceforge.net")
8409 (synopsis "Molecular biology analysis suite")
8410 (description "EMBOSS is the \"European Molecular Biology Open Software
8411Suite\". EMBOSS is an analysis package specially developed for the needs of
8412the molecular biology (e.g. EMBnet) user community. The software
8413automatically copes with data in a variety of formats and even allows
8414transparent retrieval of sequence data from the web. It also provides a
8415number of libraries for the development of software in the field of molecular
8416biology. EMBOSS also integrates a range of currently available packages and
8417tools for sequence analysis into a seamless whole.")
8418 (license license:gpl2+)))
8419
1f1b20b8
RW		 8420(define-public bits
8421 (let ((revision "1")
8422 (commit "3cc4567896d9d6442923da944beb704750a08d2d"))
8423 (package
8424 (name "bits")
8425 ;; The version is 2.13.0 even though no release archives have been
8426 ;; published as yet.
8427 (version (string-append "2.13.0-" revision "." (string-take commit 9)))
8428 (source (origin
8429 (method git-fetch)
8430 (uri (git-reference
8431 (url "https://github.com/arq5x/bits.git")
8432 (commit commit)))
8433 (file-name (string-append name "-" version "-checkout"))
8434 (sha256
8435 (base32
8436 "17n2kffk4kmhivd8c98g2vr6y1s23vbg4sxlxs689wni66797hbs"))))
8437 (build-system gnu-build-system)
8438 (arguments
8439 `(#:tests? #f ;no tests included
8440 #:phases
8441 (modify-phases %standard-phases
8442 (delete 'configure)
8443 (add-after 'unpack 'remove-cuda
8444 (lambda _
8445 (substitute* "Makefile"
8446 ((".*_cuda") "")
8447 (("(bits_test_intersections) \\\\" _ match) match))
8448 #t))
8449 (replace 'install
8450 (lambda* (#:key outputs #:allow-other-keys)
8451 (copy-recursively
8452 "bin" (string-append (assoc-ref outputs "out") "/bin"))
8453 #t)))))
8454 (inputs
8455 `(("gsl" ,gsl)
8456 ("zlib" ,zlib)))
8457 (home-page "https://github.com/arq5x/bits")
8458 (synopsis "Implementation of binary interval search algorithm")
8459 (description "This package provides an implementation of the
8460BITS (Binary Interval Search) algorithm, an approach to interval set
8461intersection. It is especially suited for the comparison of diverse genomic
8462datasets and the exploration of large datasets of genome
8463intervals (e.g. genes, sequence alignments).")
8464 (license license:gpl2))))
8465
e62ffce5 8466(define-public piranha
883302da
RW		 8467 ;; There is no release tarball for the latest version. The latest commit is
8468 ;; older than one year at the time of this writing.
8469 (let ((revision "1")
8470 (commit "0466d364b71117d01e4471b74c514436cc281233"))
8471 (package
8472 (name "piranha")
8473 (version (string-append "1.2.1-" revision "." (string-take commit 9)))
8474 (source (origin
8475 (method git-fetch)
8476 (uri (git-reference
8477 (url "https://github.com/smithlabcode/piranha.git")
8478 (commit commit)))
8479 (sha256
8480 (base32
8481 "117dc0zf20c61jam69sk4abl57ah6yi6i7qra7d7y5zrbgk12q5n"))))
8482 (build-system gnu-build-system)
8483 (arguments
8484 `(#:test-target "test"
8485 #:phases
8486 (modify-phases %standard-phases
8487 (add-after 'unpack 'copy-smithlab-cpp
8488 (lambda* (#:key inputs #:allow-other-keys)
e62ffce5 8489 (for-each (lambda (file)
883302da
RW		 8490 (install-file file "./src/smithlab_cpp/"))
8491 (find-files (assoc-ref inputs "smithlab-cpp")))
8492 #t))
8493 (add-after 'install 'install-to-store
8494 (lambda* (#:key outputs #:allow-other-keys)
8495 (let* ((out (assoc-ref outputs "out"))
8496 (bin (string-append out "/bin")))
883302da
RW		 8497 (for-each (lambda (file)
8498 (install-file file bin))
8499 (find-files "bin" ".*")))
8500 #t)))
8501 #:configure-flags
8502 (list (string-append "--with-bam_tools_headers="
8503 (assoc-ref %build-inputs "bamtools") "/include/bamtools")
8504 (string-append "--with-bam_tools_library="
8505 (assoc-ref %build-inputs "bamtools") "/lib/bamtools"))))
8506 (inputs
8507 `(("bamtools" ,bamtools)
8508 ("samtools" ,samtools-0.1)
8509 ("gsl" ,gsl)
8510 ("smithlab-cpp"
8511 ,(let ((commit "3723e2db438c51501d0423429ff396c3035ba46a"))
8512 (origin
8513 (method git-fetch)
8514 (uri (git-reference
8515 (url "https://github.com/smithlabcode/smithlab_cpp.git")
8516 (commit commit)))
8517 (file-name (string-append "smithlab_cpp-" commit "-checkout"))
8518 (sha256
8519 (base32
8520 "0l4gvbwslw5ngziskja41c00x1r06l3yidv7y0xw9djibhykzy0g")))))))
8521 (native-inputs
8522 `(("python" ,python-2)))
8523 (home-page "https://github.com/smithlabcode/piranha")
8524 (synopsis "Peak-caller for CLIP-seq and RIP-seq data")
8525 (description
8526 "Piranha is a peak-caller for genomic data produced by CLIP-seq and
e62ffce5
RW		 8527RIP-seq experiments. It takes input in BED or BAM format and identifies
8528regions of statistically significant read enrichment. Additional covariates
8529may optionally be provided to further inform the peak-calling process.")
883302da 8530 (license license:gpl3+))))
e62ffce5 8531
d1e32822
RW		 8532(define-public pepr
8533 (package
8534 (name "pepr")
8535 (version "1.0.9")
8536 (source (origin
8537 (method url-fetch)
8538 (uri (string-append "https://pypi.python.org/packages/source/P"
8539 "/PePr/PePr-" version ".tar.gz"))
8540 (sha256
8541 (base32
8542 "0qxjfdpl1b1y53nccws2d85f6k74zwmx8y8sd9rszcqhfayx6gdx"))))
8543 (build-system python-build-system)
8544 (arguments
8545 `(#:python ,python-2 ; python2 only
b41a05ce 8546 #:tests? #f)) ; no tests included
d1e32822
RW		 8547 (propagated-inputs
8548 `(("python2-numpy" ,python2-numpy)
8549 ("python2-scipy" ,python2-scipy)
8550 ("python2-pysam" ,python2-pysam)))
0c6c9c00 8551 (home-page "https://github.com/shawnzhangyx/PePr")
d1e32822
RW		 8552 (synopsis "Peak-calling and prioritization pipeline for ChIP-Seq data")
8553 (description
8554 "PePr is a ChIP-Seq peak calling or differential binding analysis tool
8555that is primarily designed for data with biological replicates. It uses a
8556negative binomial distribution to model the read counts among the samples in
8557the same group, and look for consistent differences between ChIP and control
8558group or two ChIP groups run under different conditions.")
8559 (license license:gpl3+)))
6b49a37e
RJ		 8560
8561(define-public filevercmp
8562 (let ((commit "1a9b779b93d0b244040274794d402106907b71b7"))
8563 (package
8564 (name "filevercmp")
8565 (version (string-append "0-1." (string-take commit 7)))
8566 (source (origin
8567 (method url-fetch)
8568 (uri (string-append "https://github.com/ekg/filevercmp/archive/"
8569 commit ".tar.gz"))
8570 (file-name (string-append name "-" version ".tar.gz"))
8571 (sha256
8572 (base32 "0yp5jswf5j2pqc6517x277s4s6h1ss99v57kxw9gy0jkfl3yh450"))))
8573 (build-system gnu-build-system)
8574 (arguments
8575 `(#:tests? #f ; There are no tests to run.
8576 #:phases
8577 (modify-phases %standard-phases
8578 (delete 'configure) ; There is no configure phase.
8579 (replace 'install
8580 (lambda* (#:key outputs #:allow-other-keys)
8581 (let ((bin (string-append (assoc-ref outputs "out") "/bin")))
8582 (install-file "filevercmp" bin)))))))
8583 (home-page "https://github.com/ekg/filevercmp")
8584 (synopsis "This program compares version strings")
8585 (description "This program compares version strings. It intends to be a
8586replacement for strverscmp.")
8587 (license license:gpl3+))))
5fb5dffb
RW		 8588
8589(define-public multiqc
8590 (package
8591 (name "multiqc")
66865ab8 8592 (version "1.3")
5fb5dffb
RW		 8593 (source
8594 (origin
8595 (method url-fetch)
8596 (uri (pypi-uri "multiqc" version))
8597 (sha256
8598 (base32
66865ab8 8599 "0fx1sx53znbgzfhbbiyd8j6cg5llpcsl5q5c45jy2c81d12piqfd"))))
5fb5dffb
RW		 8600 (build-system python-build-system)
8601 (propagated-inputs
66865ab8
RW		 8602 `(("python-enum34" ,python-enum34)
8603 ("python-jinja2" ,python-jinja2)
5fb5dffb
RW		 8604 ("python-simplejson" ,python-simplejson)
8605 ("python-pyyaml" ,python-pyyaml)
8606 ("python-click" ,python-click)
799247d2
RW		 8607 ("python-spectra" ,python-spectra)
8608 ("python-requests" ,python-requests)
8609 ("python-markdown" ,python-markdown)
8610 ("python-lzstring" ,python-lzstring)
5fb5dffb 8611 ("python-matplotlib" ,python-matplotlib)
1778ea74
RW		 8612 ("python-numpy" ,python-numpy)
8613 ;; MultQC checks for the presence of nose at runtime.
8614 ("python-nose" ,python-nose)))
5fb5dffb
RW		 8615 (home-page "http://multiqc.info")
8616 (synopsis "Aggregate bioinformatics analysis reports")
8617 (description
8618 "MultiQC is a tool to aggregate bioinformatics results across many
8619samples into a single report. It contains modules for a large number of
8620common bioinformatics tools.")
66865ab8 8621 (license license:gpl3+)))
6193563a
RW		 8622
8623(define-public r-chipseq
8624 (package
8625 (name "r-chipseq")
7d72f2a8 8626 (version "1.26.0")
6193563a
RW		 8627 (source
8628 (origin
8629 (method url-fetch)
8630 (uri (bioconductor-uri "chipseq" version))
8631 (sha256
8632 (base32
7d72f2a8 8633 "1hahyqiwb2ch8214xqpw0c3jpiwkmyf3dwz0xc87jx6cdnzipj3i"))))
6193563a
RW		 8634 (build-system r-build-system)
8635 (propagated-inputs
8636 `(("r-biocgenerics" ,r-biocgenerics)
8637 ("r-genomicranges" ,r-genomicranges)
8638 ("r-iranges" ,r-iranges)
8639 ("r-s4vectors" ,r-s4vectors)
8640 ("r-shortread" ,r-shortread)))
5713bbf1 8641 (home-page "https://bioconductor.org/packages/chipseq")
6193563a
RW		 8642 (synopsis "Package for analyzing ChIPseq data")
8643 (description
8644 "This package provides tools for processing short read data from ChIPseq
8645experiments.")
8646 (license license:artistic2.0)))
d407bdb9
RW		 8647
8648(define-public r-copyhelper
8649 (package
8650 (name "r-copyhelper")
8651 (version "1.6.0")
8652 (source
8653 (origin
8654 (method url-fetch)
5713bbf1 8655 (uri (string-append "https://bioconductor.org/packages/release/"
d407bdb9
RW		 8656 "data/experiment/src/contrib/CopyhelpeR_"
8657 version ".tar.gz"))
8658 (sha256
8659 (base32
8660 "0x7cyynjmxls9as2gg0iyp9x5fpalxmdjq914ss7i84i9zyk5bhq"))))
8661 (properties `((upstream-name . "CopyhelpeR")))
8662 (build-system r-build-system)
5713bbf1 8663 (home-page "https://bioconductor.org/packages/CopyhelpeR/")
d407bdb9
RW		 8664 (synopsis "Helper files for CopywriteR")
8665 (description
8666 "This package contains the helper files that are required to run the
8667Bioconductor package CopywriteR. It contains pre-assembled 1kb bin GC-content
8668and mappability files for the reference genomes hg18, hg19, hg38, mm9 and
8669mm10. In addition, it contains a blacklist filter to remove regions that
8670display copy number variation. Files are stored as GRanges objects from the
8671GenomicRanges Bioconductor package.")
8672 (license license:gpl2)))
3a3bf2f8
RW		 8673
8674(define-public r-copywriter
8675 (package
8676 (name "r-copywriter")
9eb27146 8677 (version "2.8.1")
3a3bf2f8
RW		 8678 (source
8679 (origin
8680 (method url-fetch)
8681 (uri (bioconductor-uri "CopywriteR" version))
8682 (sha256
8683 (base32
9eb27146 8684 "0xgqnq5v5213b3nzvlmjysjb7w1bc0iblqpmzbjqn7n0ib0qyhbm"))))
3a3bf2f8
RW		 8685 (properties `((upstream-name . "CopywriteR")))
8686 (build-system r-build-system)
8687 (propagated-inputs
8688 `(("r-biocparallel" ,r-biocparallel)
8689 ("r-chipseq" ,r-chipseq)
8690 ("r-copyhelper" ,r-copyhelper)
8691 ("r-data-table" ,r-data-table)
8692 ("r-dnacopy" ,r-dnacopy)
8693 ("r-futile-logger" ,r-futile-logger)
8694 ("r-genomeinfodb" ,r-genomeinfodb)
8695 ("r-genomicalignments" ,r-genomicalignments)
8696 ("r-genomicranges" ,r-genomicranges)
8697 ("r-gtools" ,r-gtools)
8698 ("r-iranges" ,r-iranges)
8699 ("r-matrixstats" ,r-matrixstats)
8700 ("r-rsamtools" ,r-rsamtools)
8701 ("r-s4vectors" ,r-s4vectors)))
8702 (home-page "https://github.com/PeeperLab/CopywriteR")
8703 (synopsis "Copy number information from targeted sequencing")
8704 (description
8705 "CopywriteR extracts DNA copy number information from targeted sequencing
8706by utilizing off-target reads. It allows for extracting uniformly distributed
8707copy number information, can be used without reference, and can be applied to
8708sequencing data obtained from various techniques including chromatin
8709immunoprecipitation and target enrichment on small gene panels. Thereby,
8710CopywriteR constitutes a widely applicable alternative to available copy
8711number detection tools.")
8712 (license license:gpl2)))
dd42a330 8713
bf3fa996
RW		 8714(define-public r-methylkit
8715 (package
8716 (name "r-methylkit")
8717 (version "1.2.0")
8718 (source (origin
8719 (method url-fetch)
8720 (uri (bioconductor-uri "methylKit" version))
8721 (sha256
8722 (base32
8723 "02acdjf6jl0c1glymin84pdna4farn4vv0gb6107d9iqz3y3gkmm"))))
8724 (properties `((upstream-name . "methylKit")))
8725 (build-system r-build-system)
8726 (propagated-inputs
8727 `(("r-data-table" ,r-data-table)
8728 ("r-emdbook" ,r-emdbook)
8729 ("r-fastseg" ,r-fastseg)
8730 ("r-genomeinfodb" ,r-genomeinfodb)
8731 ("r-genomicranges" ,r-genomicranges)
8732 ("r-gtools" ,r-gtools)
8733 ("r-iranges" ,r-iranges)
8734 ("r-kernsmooth" ,r-kernsmooth)
8735 ("r-limma" ,r-limma)
8736 ("r-mclust" ,r-mclust)
8737 ("r-qvalue" ,r-qvalue)
8738 ("r-r-utils" ,r-r-utils)
8739 ("r-rcpp" ,r-rcpp)
8740 ("r-rhtslib" ,r-rhtslib)
8741 ("r-rsamtools" ,r-rsamtools)
8742 ("r-rtracklayer" ,r-rtracklayer)
8743 ("r-s4vectors" ,r-s4vectors)
8744 ("r-zlibbioc" ,r-zlibbioc)))
8745 (inputs
8746 `(("zlib" ,zlib)))
8747 (home-page "http://code.google.com/p/methylkit/")
8748 (synopsis
8749 "DNA methylation analysis from high-throughput bisulfite sequencing results")
8750 (description
8751 "MethylKit is an R package for DNA methylation analysis and annotation
8752from high-throughput bisulfite sequencing. The package is designed to deal
8753with sequencing data from @dfn{Reduced representation bisulfite
8754sequencing} (RRBS) and its variants, but also target-capture methods and whole
8755genome bisulfite sequencing. It also has functions to analyze base-pair
8756resolution 5hmC data from experimental protocols such as oxBS-Seq and
8757TAB-Seq.")
8758 (license license:artistic2.0)))
8759
dd42a330
RJPB		 8760(define-public r-sva
8761 (package
8762 (name "r-sva")
bfa0d1e7 8763 (version "3.24.4")
dd42a330
RJPB		 8764 (source
8765 (origin
8766 (method url-fetch)
8767 (uri (bioconductor-uri "sva" version))
8768 (sha256
8769 (base32
bfa0d1e7 8770 "0wcway4ai9im81xnrzb1vij2iidq5pw24qhjfgacmhxvx3dzhbsc"))))
dd42a330
RJPB		 8771 (build-system r-build-system)
8772 (propagated-inputs
aeb64f3c 8773 `(("r-genefilter" ,r-genefilter)
bfa0d1e7
RJ		 8774 ("r-mgcv" ,r-mgcv)
8775 ("r-biocparallel" ,r-biocparallel)
8776 ("r-matrixstats" ,r-matrixstats)
8777 ("r-limma" ,r-limma)))
5713bbf1 8778 (home-page "https://bioconductor.org/packages/sva")
dd42a330
RJPB		 8779 (synopsis "Surrogate variable analysis")
8780 (description
8781 "This package contains functions for removing batch effects and other
8782unwanted variation in high-throughput experiment. It also contains functions
8783for identifying and building surrogate variables for high-dimensional data
8784sets. Surrogate variables are covariates constructed directly from
8785high-dimensional data like gene expression/RNA sequencing/methylation/brain
8786imaging data that can be used in subsequent analyses to adjust for unknown,
8787unmodeled, or latent sources of noise.")
8788 (license license:artistic2.0)))
fb1e528e
RW		 8789
8790(define-public r-seqminer
8791 (package
8792 (name "r-seqminer")
5e0de298 8793 (version "6.0")
fb1e528e
RW		 8794 (source
8795 (origin
8796 (method url-fetch)
8797 (uri (cran-uri "seqminer" version))
8798 (sha256
8799 (base32
5e0de298 8800 "057j1l6dip35l1aivilapl2zv9db677b3di2pb3sfgq2sxg0ps3l"))))
fb1e528e
RW		 8801 (build-system r-build-system)
8802 (inputs
8803 `(("zlib" ,zlib)))
8804 (home-page "http://seqminer.genomic.codes")
8805 (synopsis "Read nucleotide sequence data (VCF, BCF, and METAL formats)")
8806 (description
8807 "This package provides tools to integrate nucleotide sequencing
8808data (variant call format, e.g. VCF or BCF) or meta-analysis results in R.")
8809 ;; Any version of the GPL is acceptable
8810 (license (list license:gpl2+ license:gpl3+))))
1f40e73c
RW		 8811
8812(define-public r-raremetals2
8813 (package
8814 (name "r-raremetals2")
8815 (version "0.1")
8816 (source
8817 (origin
8818 (method url-fetch)
8819 (uri (string-append "http://genome.sph.umich.edu/w/images/"
8820 "b/b7/RareMETALS2_" version ".tar.gz"))
8821 (sha256
8822 (base32
8823 "0z5ljcgvnm06ja9lm85a3cniq7slxcy37aqqkxrdidr79an5fs4s"))))
8824 (properties `((upstream-name . "RareMETALS2")))
8825 (build-system r-build-system)
8826 (propagated-inputs
8827 `(("r-seqminer" ,r-seqminer)
8828 ("r-mvtnorm" ,r-mvtnorm)
aeb64f3c 8829 ("r-mass" ,r-mass)
1f40e73c
RW		 8830 ("r-compquadform" ,r-compquadform)
8831 ("r-getopt" ,r-getopt)))
8832 (home-page "http://genome.sph.umich.edu/wiki/RareMETALS2")
8833 (synopsis "Analyze gene-level association tests for binary trait")
8834 (description
8835 "The R package rareMETALS2 is an extension of the R package rareMETALS.
8836It was designed to meta-analyze gene-level association tests for binary trait.
8837While rareMETALS offers a near-complete solution for meta-analysis of
8838gene-level tests for quantitative trait, it does not offer the optimal
8839solution for binary trait. The package rareMETALS2 offers improved features
8840for analyzing gene-level association tests in meta-analyses for binary
8841trait.")
8842 (license license:gpl3)))
0e7d058e
RW		 8843
8844(define-public r-maldiquant
8845 (package
8846 (name "r-maldiquant")
7b92acdf 8847 (version "1.16.4")
0e7d058e
RW		 8848 (source
8849 (origin
8850 (method url-fetch)
8851 (uri (cran-uri "MALDIquant" version))
8852 (sha256
8853 (base32
7b92acdf 8854 "1pmhsfvd45a44xdiml4zx3zd5fhygqyziqvygahkk9yibnyhv4cv"))))
0e7d058e
RW		 8855 (properties `((upstream-name . "MALDIquant")))
8856 (build-system r-build-system)
8857 (home-page "http://cran.r-project.org/web/packages/MALDIquant")
8858 (synopsis "Quantitative analysis of mass spectrometry data")
8859 (description
8860 "This package provides a complete analysis pipeline for matrix-assisted
8861laser desorption/ionization-time-of-flight (MALDI-TOF) and other
8862two-dimensional mass spectrometry data. In addition to commonly used plotting
8863and processing methods it includes distinctive features, namely baseline
8864subtraction methods such as morphological filters (TopHat) or the
8865statistics-sensitive non-linear iterative peak-clipping algorithm (SNIP), peak
8866alignment using warping functions, handling of replicated measurements as well
8867as allowing spectra with different resolutions.")
8868 (license license:gpl3+)))
ae262174
RW		 8869
8870(define-public r-protgenerics
8871 (package
8872 (name "r-protgenerics")
5b1337cc 8873 (version "1.8.0")
ae262174
RW		 8874 (source
8875 (origin
8876 (method url-fetch)
8877 (uri (bioconductor-uri "ProtGenerics" version))
8878 (sha256
8879 (base32
5b1337cc 8880 "08idb3rvxn4fl7rd66jasyqz47cb76dbc3968r1g26jr2ci3w1pl"))))
ae262174
RW		 8881 (properties `((upstream-name . "ProtGenerics")))
8882 (build-system r-build-system)
8883 (home-page "https://github.com/lgatto/ProtGenerics")
8884 (synopsis "S4 generic functions for proteomics infrastructure")
8885 (description
8886 "This package provides S4 generic functions needed by Bioconductor
8887proteomics packages.")
8888 (license license:artistic2.0)))
71676a1f
RW		 8889
8890(define-public r-mzr
8891 (package
8892 (name "r-mzr")
e440ad6d 8893 (version "2.10.0")
71676a1f
RW		 8894 (source
8895 (origin
8896 (method url-fetch)
8897 (uri (bioconductor-uri "mzR" version))
8898 (sha256
8899 (base32
e440ad6d 8900 "1zir46h320n2vbrky6q3m8l221f3wdjlfsnx4ak9xca5min24xm7"))))
71676a1f
RW		 8901 (properties `((upstream-name . "mzR")))
8902 (build-system r-build-system)
8903 (inputs
d534f3a4
EF		 8904 `(("boost" ,boost)
8905 ("netcdf" ,netcdf)))
71676a1f
RW		 8906 (propagated-inputs
8907 `(("r-biobase" ,r-biobase)
8908 ("r-biocgenerics" ,r-biocgenerics)
8909 ("r-protgenerics" ,r-protgenerics)
8910 ("r-rcpp" ,r-rcpp)
8911 ("r-zlibbioc" ,r-zlibbioc)))
8912 (home-page "https://github.com/sneumann/mzR/")
8913 (synopsis "Parser for mass spectrometry data files")
8914 (description
8915 "The mzR package provides a unified API to the common file formats and
8916parsers available for mass spectrometry data. It comes with a wrapper for the
8917ISB random access parser for mass spectrometry mzXML, mzData and mzML files.
8918The package contains the original code written by the ISB, and a subset of the
8919proteowizard library for mzML and mzIdentML. The netCDF reading code has
8920previously been used in XCMS.")
8921 (license license:artistic2.0)))
5c9d1505
RW		 8922
8923(define-public r-affyio
8924 (package
8925 (name "r-affyio")
71abcc4c 8926 (version "1.46.0")
5c9d1505
RW		 8927 (source
8928 (origin
8929 (method url-fetch)
8930 (uri (bioconductor-uri "affyio" version))
8931 (sha256
8932 (base32
71abcc4c 8933 "13w6al9296q916w0w6ngbsk25b21ahla1b6n40fcqhbvhyfii6sj"))))
5c9d1505
RW		 8934 (build-system r-build-system)
8935 (propagated-inputs
8936 `(("r-zlibbioc" ,r-zlibbioc)))
8937 (inputs
8938 `(("zlib" ,zlib)))
8939 (home-page "https://github.com/bmbolstad/affyio")
8940 (synopsis "Tools for parsing Affymetrix data files")
8941 (description
8942 "This package provides routines for parsing Affymetrix data files based
8943upon file format information. The primary focus is on accessing the CEL and
8944CDF file formats.")
8945 (license license:lgpl2.0+)))
05c7e5fb
RW		 8946
8947(define-public r-affy
8948 (package
8949 (name "r-affy")
bd3cc4cc 8950 (version "1.54.0")
05c7e5fb
RW		 8951 (source
8952 (origin
8953 (method url-fetch)
8954 (uri (bioconductor-uri "affy" version))
8955 (sha256
8956 (base32
bd3cc4cc 8957 "0azwg2qxzgflr1rjvbdln5i5rbcr9gs36kqlacd9cwl1szb9ad3m"))))
05c7e5fb
RW		 8958 (build-system r-build-system)
8959 (propagated-inputs
8960 `(("r-affyio" ,r-affyio)
8961 ("r-biobase" ,r-biobase)
8962 ("r-biocgenerics" ,r-biocgenerics)
8963 ("r-biocinstaller" ,r-biocinstaller)
8964 ("r-preprocesscore" ,r-preprocesscore)
8965 ("r-zlibbioc" ,r-zlibbioc)))
5713bbf1 8966 (home-page "https://bioconductor.org/packages/affy")
05c7e5fb
RW		 8967 (synopsis "Methods for affymetrix oligonucleotide arrays")
8968 (description
8969 "This package contains functions for exploratory oligonucleotide array
8970analysis.")
8971 (license license:lgpl2.0+)))
7c08afaf
RW		 8972
8973(define-public r-vsn
8974 (package
8975 (name "r-vsn")
60c08b2f 8976 (version "3.44.0")
7c08afaf
RW		 8977 (source
8978 (origin
8979 (method url-fetch)
8980 (uri (bioconductor-uri "vsn" version))
8981 (sha256
8982 (base32
60c08b2f 8983 "0qhg3a4sc62pfdxcpvmk831rk138xh4zx4f1s39jhxpqqhmr7jvk"))))
7c08afaf
RW		 8984 (build-system r-build-system)
8985 (propagated-inputs
8986 `(("r-affy" ,r-affy)
8987 ("r-biobase" ,r-biobase)
8988 ("r-ggplot2" ,r-ggplot2)
60c08b2f 8989 ("r-hexbin" ,r-hexbin)
aeb64f3c 8990 ("r-lattice" ,r-lattice)
7c08afaf 8991 ("r-limma" ,r-limma)))
5713bbf1 8992 (home-page "https://bioconductor.org/packages/release/bioc/html/vsn.html")
7c08afaf
RW		 8993 (synopsis "Variance stabilization and calibration for microarray data")
8994 (description
8995 "The package implements a method for normalising microarray intensities,
8996and works for single- and multiple-color arrays. It can also be used for data
8997from other technologies, as long as they have similar format. The method uses
8998a robust variant of the maximum-likelihood estimator for an
8999additive-multiplicative error model and affine calibration. The model
9000incorporates data calibration step (a.k.a. normalization), a model for the
9001dependence of the variance on the mean intensity and a variance stabilizing
9002data transformation. Differences between transformed intensities are
9003analogous to \"normalized log-ratios\". However, in contrast to the latter,
9004their variance is independent of the mean, and they are usually more sensitive
9005and specific in detecting differential transcription.")
9006 (license license:artistic2.0)))
4aa7d592
RW		 9007
9008(define-public r-mzid
9009 (package
9010 (name "r-mzid")
63bf2102 9011 (version "1.14.0")
4aa7d592
RW		 9012 (source
9013 (origin
9014 (method url-fetch)
9015 (uri (bioconductor-uri "mzID" version))
9016 (sha256
9017 (base32
63bf2102 9018 "11xnild02jz24vbsfy92lb7jlqqwnrswg66a7r4rsw8d2ibrbk33"))))
4aa7d592
RW		 9019 (properties `((upstream-name . "mzID")))
9020 (build-system r-build-system)
9021 (propagated-inputs
9022 `(("r-doparallel" ,r-doparallel)
9023 ("r-foreach" ,r-foreach)
9024 ("r-iterators" ,r-iterators)
9025 ("r-plyr" ,r-plyr)
9026 ("r-protgenerics" ,r-protgenerics)
9027 ("r-rcpp" ,r-rcpp)
9028 ("r-xml" ,r-xml)))
5713bbf1 9029 (home-page "https://bioconductor.org/packages/mzID")
4aa7d592
RW		 9030 (synopsis "Parser for mzIdentML files")
9031 (description
9032 "This package provides a parser for mzIdentML files implemented using the
9033XML package. The parser tries to be general and able to handle all types of
9034mzIdentML files with the drawback of having less pretty output than a vendor
9035specific parser.")
9036 (license license:gpl2+)))
6a67e181
RW		 9037
9038(define-public r-pcamethods
9039 (package
9040 (name "r-pcamethods")
5bfa6e09 9041 (version "1.68.0")
6a67e181
RW		 9042 (source
9043 (origin
9044 (method url-fetch)
9045 (uri (bioconductor-uri "pcaMethods" version))
9046 (sha256
9047 (base32
5bfa6e09 9048 "0c4lphqyzj577ws4s172391cgv00s5nhy152zp18k2k4diyhq6n0"))))
6a67e181
RW		 9049 (properties `((upstream-name . "pcaMethods")))
9050 (build-system r-build-system)
9051 (propagated-inputs
9052 `(("r-biobase" ,r-biobase)
9053 ("r-biocgenerics" ,r-biocgenerics)
aeb64f3c 9054 ("r-mass" ,r-mass)
6a67e181
RW		 9055 ("r-rcpp" ,r-rcpp)))
9056 (home-page "https://github.com/hredestig/pcamethods")
9057 (synopsis "Collection of PCA methods")
9058 (description
9059 "This package provides Bayesian PCA, Probabilistic PCA, Nipals PCA,
9060Inverse Non-Linear PCA and the conventional SVD PCA. A cluster based method
9061for missing value estimation is included for comparison. BPCA, PPCA and
9062NipalsPCA may be used to perform PCA on incomplete data as well as for
9063accurate missing value estimation. A set of methods for printing and plotting
9064the results is also provided. All PCA methods make use of the same data
9065structure (pcaRes) to provide a common interface to the PCA results.")
9066 (license license:gpl3+)))
11879284
RW		 9067
9068(define-public r-msnbase
9069 (package
9070 (name "r-msnbase")
82359c72 9071 (version "2.2.0")
11879284
RW		 9072 (source
9073 (origin
9074 (method url-fetch)
9075 (uri (bioconductor-uri "MSnbase" version))
9076 (sha256
9077 (base32
82359c72 9078 "1n9bbdlp8d8bx2mqby0c4yylz1yil42scbkxqgyrfr9s5sn6dqff"))))
11879284
RW		 9079 (properties `((upstream-name . "MSnbase")))
9080 (build-system r-build-system)
9081 (propagated-inputs
9082 `(("r-affy" ,r-affy)
9083 ("r-biobase" ,r-biobase)
9084 ("r-biocgenerics" ,r-biocgenerics)
9085 ("r-biocparallel" ,r-biocparallel)
9086 ("r-digest" ,r-digest)
9087 ("r-ggplot2" ,r-ggplot2)
9088 ("r-impute" ,r-impute)
9089 ("r-iranges" ,r-iranges)
9090 ("r-maldiquant" ,r-maldiquant)
9091 ("r-mzid" ,r-mzid)
9092 ("r-mzr" ,r-mzr)
9093 ("r-pcamethods" ,r-pcamethods)
9094 ("r-plyr" ,r-plyr)
9095 ("r-preprocesscore" ,r-preprocesscore)
9096 ("r-protgenerics" ,r-protgenerics)
9097 ("r-rcpp" ,r-rcpp)
9098 ("r-reshape2" ,r-reshape2)
9099 ("r-s4vectors" ,r-s4vectors)
9100 ("r-vsn" ,r-vsn)
9101 ("r-xml" ,r-xml)))
9102 (home-page "https://github.com/lgatto/MSnbase")
9103 (synopsis "Base functions and classes for MS-based proteomics")
9104 (description
9105 "This package provides basic plotting, data manipulation and processing
e614d639 9106of mass spectrometry based proteomics data.")
11879284 9107 (license license:artistic2.0)))
66da3a9f
RW		 9108
9109(define-public r-msnid
9110 (package
9111 (name "r-msnid")
524bfe2d 9112 (version "1.10.0")
66da3a9f
RW		 9113 (source
9114 (origin
9115 (method url-fetch)
9116 (uri (bioconductor-uri "MSnID" version))
9117 (sha256
9118 (base32
524bfe2d 9119 "0pjwargi5lif8q53fd43ql67p3yk9w10jychafd9qgbaw5k3f68k"))))
66da3a9f
RW		 9120 (properties `((upstream-name . "MSnID")))
9121 (build-system r-build-system)
9122 (propagated-inputs
9123 `(("r-biobase" ,r-biobase)
9124 ("r-data-table" ,r-data-table)
9125 ("r-doparallel" ,r-doparallel)
9126 ("r-dplyr" ,r-dplyr)
9127 ("r-foreach" ,r-foreach)
9128 ("r-iterators" ,r-iterators)
9129 ("r-msnbase" ,r-msnbase)
9130 ("r-mzid" ,r-mzid)
9131 ("r-mzr" ,r-mzr)
9132 ("r-protgenerics" ,r-protgenerics)
9133 ("r-r-cache" ,r-r-cache)
9134 ("r-rcpp" ,r-rcpp)
9135 ("r-reshape2" ,r-reshape2)))
5713bbf1 9136 (home-page "https://bioconductor.org/packages/MSnID")
66da3a9f
RW		 9137 (synopsis "Utilities for LC-MSn proteomics identifications")
9138 (description
2923f3e5
RW		 9139 "This package extracts @dfn{tandem mass spectrometry} (MS/MS) ID data
9140from mzIdentML (leveraging the mzID package) or text files. After collating
9141the search results from multiple datasets it assesses their identification
9142quality and optimize filtering criteria to achieve the maximum number of
9143identifications while not exceeding a specified false discovery rate. It also
9144contains a number of utilities to explore the MS/MS results and assess missed
9145and irregular enzymatic cleavages, mass measurement accuracy, etc.")
66da3a9f 9146 (license license:artistic2.0)))
5ded35d8 9147
2441c284
RJPB		 9148(define-public r-seurat
9149 ;; Source releases are only made for new x.0 versions. All newer versions
9150 ;; are only released as pre-built binaries. At the time of this writing the
9151 ;; latest binary release is 1.4.0.12, which is equivalent to this commit.
9152 (let ((commit "fccb77d1452c35ee47e47ebf8e87bddb59f3b08d")
9153 (revision "1"))
9154 (package
9155 (name "r-seurat")
9156 (version (string-append "1.4.0.12-" revision "." (string-take commit 7)))
9157 (source (origin
9158 (method git-fetch)
9159 (uri (git-reference
9160 (url "https://github.com/satijalab/seurat")
9161 (commit commit)))
9162 (file-name (string-append name "-" version "-checkout"))
9163 (sha256
9164 (base32
9165 "101wq3aqrdmbfi3lqmq4iivk9iwbf10d4z216ss25hf7n9091cyl"))
9166 ;; Delete pre-built jar.
9167 (snippet
9168 '(begin (delete-file "inst/java/ModularityOptimizer.jar")
9169 #t))))
9170 (build-system r-build-system)
9171 (arguments
9172 `(#:phases
9173 (modify-phases %standard-phases
9174 (add-after 'unpack 'build-jar
9175 (lambda* (#:key inputs #:allow-other-keys)
9176 (let ((classesdir "tmp-classes"))
9177 (setenv "JAVA_HOME" (assoc-ref inputs "jdk"))
9178 (mkdir classesdir)
9179 (and (zero? (apply system* `("javac" "-d" ,classesdir
9180 ,@(find-files "java" "\\.java$"))))
9181 (zero? (system* "jar"
9182 "-cf" "inst/java/ModularityOptimizer.jar"
9183 "-C" classesdir ".")))))))))
9184 (native-inputs
9185 `(("jdk" ,icedtea "jdk")))
9186 (propagated-inputs
9187 `(("r-ape" ,r-ape)
9188 ("r-caret" ,r-caret)
9189 ("r-cowplot" ,r-cowplot)
9190 ("r-dplyr" ,r-dplyr)
9191 ("r-fastica" ,r-fastica)
9192 ("r-fnn" ,r-fnn)
9193 ("r-fpc" ,r-fpc)
9194 ("r-gdata" ,r-gdata)
9195 ("r-ggplot2" ,r-ggplot2)
9196 ("r-gplots" ,r-gplots)
9197 ("r-gridextra" ,r-gridextra)
9198 ("r-igraph" ,r-igraph)
9199 ("r-irlba" ,r-irlba)
9200 ("r-lars" ,r-lars)
9201 ("r-mixtools" ,r-mixtools)
9202 ("r-pbapply" ,r-pbapply)
9203 ("r-plyr" ,r-plyr)
9204 ("r-ranger" ,r-ranger)
9205 ("r-rcolorbrewer" ,r-rcolorbrewer)
9206 ("r-rcpp" ,r-rcpp)
9207 ("r-rcppeigen" ,r-rcppeigen)
9208 ("r-rcppprogress" ,r-rcppprogress)
9209 ("r-reshape2" ,r-reshape2)
9210 ("r-rocr" ,r-rocr)
9211 ("r-rtsne" ,r-rtsne)
9212 ("r-stringr" ,r-stringr)
9213 ("r-tclust" ,r-tclust)
9214 ("r-tsne" ,r-tsne)
9215 ("r-vgam" ,r-vgam)))
9216 (home-page "http://www.satijalab.org/seurat")
9217 (synopsis "Seurat is an R toolkit for single cell genomics")
9218 (description
9219 "This package is an R package designed for QC, analysis, and
9220exploration of single cell RNA-seq data. It easily enables widely-used
9221analytical techniques, including the identification of highly variable genes,
9222dimensionality reduction; PCA, ICA, t-SNE, standard unsupervised clustering
9223algorithms; density clustering, hierarchical clustering, k-means, and the
9224discovery of differentially expressed genes and markers.")
9225 (license license:gpl3))))
9226
d4b83651
RW		 9227(define-public r-aroma-light
9228 (package
9229 (name "r-aroma-light")
9230 (version "3.6.0")
9231 (source
9232 (origin
9233 (method url-fetch)
9234 (uri (bioconductor-uri "aroma.light" version))
9235 (sha256
9236 (base32
9237 "10snykmmx36qaymyf5s1n1km8hsscyzpykcpf0mzsrcv8ml9rp8a"))))
9238 (properties `((upstream-name . "aroma.light")))
9239 (build-system r-build-system)
9240 (propagated-inputs
9241 `(("r-matrixstats" ,r-matrixstats)
9242 ("r-r-methodss3" ,r-r-methodss3)
9243 ("r-r-oo" ,r-r-oo)
9244 ("r-r-utils" ,r-r-utils)))
9245 (home-page "https://github.com/HenrikBengtsson/aroma.light")
9246 (synopsis "Methods for normalization and visualization of microarray data")
9247 (description
9248 "This package provides methods for microarray analysis that take basic
9249data types such as matrices and lists of vectors. These methods can be used
9250standalone, be utilized in other packages, or be wrapped up in higher-level
9251classes.")
9252 (license license:gpl2+)))
9253
7b465327
RW		 9254(define-public r-deseq
9255 (package
9256 (name "r-deseq")
9257 (version "1.28.0")
9258 (source
9259 (origin
9260 (method url-fetch)
9261 (uri (bioconductor-uri "DESeq" version))
9262 (sha256
9263 (base32
9264 "0j3dgcxd64m9qknmlcbdzvg4xhp981xd6nbwsvnqjfn6yypslgyw"))))
9265 (properties `((upstream-name . "DESeq")))
9266 (build-system r-build-system)
9267 (propagated-inputs
9268 `(("r-biobase" ,r-biobase)
9269 ("r-biocgenerics" ,r-biocgenerics)
9270 ("r-genefilter" ,r-genefilter)
9271 ("r-geneplotter" ,r-geneplotter)
9272 ("r-lattice" ,r-lattice)
9273 ("r-locfit" ,r-locfit)
9274 ("r-mass" ,r-mass)
9275 ("r-rcolorbrewer" ,r-rcolorbrewer)))
9276 (home-page "http://www-huber.embl.de/users/anders/DESeq")
9277 (synopsis "Differential gene expression analysis")
9278 (description
9279 "This package provides tools for estimating variance-mean dependence in
9280count data from high-throughput genetic sequencing assays and for testing for
9281differential expression based on a model using the negative binomial
9282distribution.")
9283 (license license:gpl3+)))
9284
296b3c63
RW		 9285(define-public r-edaseq
9286 (package
9287 (name "r-edaseq")
9288 (version "2.10.0")
9289 (source
9290 (origin
9291 (method url-fetch)
9292 (uri (bioconductor-uri "EDASeq" version))
9293 (sha256
9294 (base32
9295 "0f25dfc8hdii9fjm3bf89vy9jkxv23sa62fkcga5b4gkipwrvm9a"))))
9296 (properties `((upstream-name . "EDASeq")))
9297 (build-system r-build-system)
9298 (propagated-inputs
9299 `(("r-annotationdbi" ,r-annotationdbi)
9300 ("r-aroma-light" ,r-aroma-light)
9301 ("r-biobase" ,r-biobase)
9302 ("r-biocgenerics" ,r-biocgenerics)
9303 ("r-biomart" ,r-biomart)
9304 ("r-biostrings" ,r-biostrings)
9305 ("r-deseq" ,r-deseq)
9306 ("r-genomicfeatures" ,r-genomicfeatures)
9307 ("r-genomicranges" ,r-genomicranges)
9308 ("r-iranges" ,r-iranges)
9309 ("r-rsamtools" ,r-rsamtools)
9310 ("r-shortread" ,r-shortread)))
9311 (home-page "https://github.com/drisso/EDASeq")
9312 (synopsis "Exploratory data analysis and normalization for RNA-Seq")
9313 (description
9314 "This package provides support for numerical and graphical summaries of
9315RNA-Seq genomic read data. Provided within-lane normalization procedures to
9316adjust for GC-content effect (or other gene-level effects) on read counts:
9317loess robust local regression, global-scaling, and full-quantile
9318normalization. Between-lane normalization procedures to adjust for
9319distributional differences between lanes (e.g., sequencing depth):
9320global-scaling and full-quantile normalization.")
9321 (license license:artistic2.0)))
9322
5f96f30f
RW		 9323(define-public r-interactivedisplaybase
9324 (package
9325 (name "r-interactivedisplaybase")
9326 (version "1.14.0")
9327 (source
9328 (origin
9329 (method url-fetch)
9330 (uri (bioconductor-uri "interactiveDisplayBase" version))
9331 (sha256
9332 (base32
9333 "12f6ap4bl3h2iwwhg8i3r9a7yyd28d8i5lb3fj1vnfvjs762r7r7"))))
9334 (properties
9335 `((upstream-name . "interactiveDisplayBase")))
9336 (build-system r-build-system)
9337 (propagated-inputs
9338 `(("r-biocgenerics" ,r-biocgenerics)
9339 ("r-shiny" ,r-shiny)))
5713bbf1 9340 (home-page "https://bioconductor.org/packages/interactiveDisplayBase")
5f96f30f
RW		 9341 (synopsis "Base package for web displays of Bioconductor objects")
9342 (description
9343 "This package contains the basic methods needed to generate interactive
9344Shiny-based display methods for Bioconductor objects.")
9345 (license license:artistic2.0)))
9346
4a007ffa
RW		 9347(define-public r-annotationhub
9348 (package
9349 (name "r-annotationhub")
9350 (version "2.8.2")
9351 (source
9352 (origin
9353 (method url-fetch)
9354 (uri (bioconductor-uri "AnnotationHub" version))
9355 (sha256
9356 (base32
9357 "1nh5si3j1nv37jcg4260582ayjg18851np47cskrm54prnvhwd9r"))))
9358 (properties `((upstream-name . "AnnotationHub")))
9359 (build-system r-build-system)
9360 (propagated-inputs
9361 `(("r-annotationdbi" ,r-annotationdbi)
9362 ("r-biocgenerics" ,r-biocgenerics)
9363 ("r-biocinstaller" ,r-biocinstaller)
9364 ("r-httr" ,r-httr)
9365 ("r-interactivedisplaybase" ,r-interactivedisplaybase)
9366 ("r-rsqlite" ,r-rsqlite)
9367 ("r-s4vectors" ,r-s4vectors)
9368 ("r-yaml" ,r-yaml)))
5713bbf1 9369 (home-page "https://bioconductor.org/packages/AnnotationHub")
4a007ffa
RW		 9370 (synopsis "Client to access AnnotationHub resources")
9371 (description
9372 "This package provides a client for the Bioconductor AnnotationHub web
9373resource. The AnnotationHub web resource provides a central location where
9374genomic files (e.g. VCF, bed, wig) and other resources from standard
9375locations (e.g. UCSC, Ensembl) can be discovered. The resource includes
9376metadata about each resource, e.g., a textual description, tags, and date of
9377modification. The client creates and manages a local cache of files retrieved
9378by the user, helping with quick and reproducible access.")
9379 (license license:artistic2.0)))
9380
d4a1e918
RW		 9381(define-public r-fastseg
9382 (package
9383 (name "r-fastseg")
9384 (version "1.22.0")
9385 (source
9386 (origin
9387 (method url-fetch)
9388 (uri (bioconductor-uri "fastseg" version))
9389 (sha256
9390 (base32
9391 "083wiz03q9mynwchs9frlpp6c84dncri5ncibx6h82p228cpja6h"))))
9392 (build-system r-build-system)
9393 (propagated-inputs
9394 `(("r-biobase" ,r-biobase)
9395 ("r-biocgenerics" ,r-biocgenerics)
9396 ("r-genomicranges" ,r-genomicranges)
9397 ("r-iranges" ,r-iranges)
9398 ("r-s4vectors" ,r-s4vectors)))
9399 (home-page "http://www.bioinf.jku.at/software/fastseg/index.html")
9400 (synopsis "Fast segmentation algorithm for genetic sequencing data")
9401 (description
9402 "Fastseg implements a very fast and efficient segmentation algorithm.
9403It can segment data from DNA microarrays and data from next generation
9404sequencing for example to detect copy number segments. Further it can segment
9405data from RNA microarrays like tiling arrays to identify transcripts. Most
9406generally, it can segment data given as a matrix or as a vector. Various data
9407formats can be used as input to fastseg like expression set objects for
9408microarrays or GRanges for sequencing data.")
9409 (license license:lgpl2.0+)))
9410
212f6120
RW		 9411(define-public r-qvalue
9412 (package
9413 (name "r-qvalue")
9414 (version "2.8.0")
9415 (source
9416 (origin
9417 (method url-fetch)
9418 (uri (bioconductor-uri "qvalue" version))
9419 (sha256
9420 (base32
9421 "1dxdwa767a9r8n61r272ypi09qblcdfpzzwkmri74y5mbp1r3y4i"))))
9422 (build-system r-build-system)
9423 (propagated-inputs
9424 `(("r-ggplot2" ,r-ggplot2)
9425 ("r-reshape2" ,r-reshape2)))
9426 (home-page "http://github.com/jdstorey/qvalue")
9427 (synopsis "Q-value estimation for false discovery rate control")
9428 (description
9429 "This package takes a list of p-values resulting from the simultaneous
9430testing of many hypotheses and estimates their q-values and local @dfn{false
9431discovery rate} (FDR) values. The q-value of a test measures the proportion
9432of false positives incurred when that particular test is called significant.
9433The local FDR measures the posterior probability the null hypothesis is true
9434given the test's p-value. Various plots are automatically generated, allowing
9435one to make sensible significance cut-offs. The software can be applied to
9436problems in genomics, brain imaging, astrophysics, and data mining.")
9437 ;; Any version of the LGPL.
9438 (license license:lgpl3+)))
9439
5ded35d8
RW		 9440(define htslib-for-sambamba
9441 (let ((commit "2f3c3ea7b301f9b45737a793c0b2dcf0240e5ee5"))
9442 (package
9443 (inherit htslib)
9444 (name "htslib-for-sambamba")
9445 (version (string-append "1.3.1-1." (string-take commit 9)))
9446 (source
9447 (origin
9448 (method git-fetch)
9449 (uri (git-reference
9450 (url "https://github.com/lomereiter/htslib.git")
9451 (commit commit)))
9452 (file-name (string-append "htslib-" version "-checkout"))
9453 (sha256
9454 (base32
9455 "0g38g8s3npr0gjm9fahlbhiskyfws9l5i0x1ml3rakzj7az5l9c9"))))
9456 (arguments
9457 (substitute-keyword-arguments (package-arguments htslib)
9458 ((#:phases phases)
9459 `(modify-phases ,phases
d10092b8 9460 (add-after 'unpack 'bootstrap
5ded35d8
RW		 9461 (lambda _
9462 (zero? (system* "autoreconf" "-vif"))))))))
9463 (native-inputs
9464 `(("autoconf" ,autoconf)
9465 ("automake" ,automake)
9466 ,@(package-native-inputs htslib))))))
9467
9468(define-public sambamba
9469 (package
9470 (name "sambamba")
9471 (version "0.6.5")
9472 (source
9473 (origin
9474 (method url-fetch)
9475 (uri (string-append "https://github.com/lomereiter/sambamba/"
9476 "archive/v" version ".tar.gz"))
9477 (file-name (string-append name "-" version ".tar.gz"))
9478 (sha256
9479 (base32
9480 "17076gijd65a3f07zns2gvbgahiz5lriwsa6dq353ss3jl85d8vy"))))
9481 (build-system gnu-build-system)
9482 (arguments
9483 `(#:tests? #f ; there is no test target
9484 #:make-flags
9485 '("D_COMPILER=ldc2"
9486 ;; Override "--compiler" flag only.
9487 "D_FLAGS=--compiler=ldc2 -IBioD -g -d"
9488 "sambamba-ldmd2-64")
9489 #:phases
9490 (modify-phases %standard-phases
9491 (delete 'configure)
9492 (add-after 'unpack 'place-biod
9493 (lambda* (#:key inputs #:allow-other-keys)
9494 (copy-recursively (assoc-ref inputs "biod") "BioD")
9495 #t))
9496 (add-after 'unpack 'unbundle-prerequisites
9497 (lambda _
9498 (substitute* "Makefile"
9499 ((" htslib-static lz4-static") ""))
9500 #t))
9501 (replace 'install
9502 (lambda* (#:key outputs #:allow-other-keys)
9503 (let* ((out (assoc-ref outputs "out"))
9504 (bin (string-append out "/bin")))
9505 (mkdir-p bin)
9506 (install-file "build/sambamba" bin)
9507 #t))))))
9508 (native-inputs
9509 `(("ldc" ,ldc)
9510 ("rdmd" ,rdmd)
9511 ("biod"
9512 ,(let ((commit "1248586b54af4bd4dfb28ebfebfc6bf012e7a587"))
9513 (origin
9514 (method git-fetch)
9515 (uri (git-reference
9516 (url "https://github.com/biod/BioD.git")
9517 (commit commit)))
9518 (file-name (string-append "biod-"
9519 (string-take commit 9)
9520 "-checkout"))
9521 (sha256
9522 (base32
9523 "1m8hi1n7x0ri4l6s9i0x6jg4z4v94xrfdzp7mbizdipfag0m17g3")))))))
9524 (inputs
9525 `(("lz4" ,lz4)
9526 ("htslib" ,htslib-for-sambamba)))
9527 (home-page "http://lomereiter.github.io/sambamba")
9528 (synopsis "Tools for working with SAM/BAM data")
9529 (description "Sambamba is a high performance modern robust and
9530fast tool (and library), written in the D programming language, for
9531working with SAM and BAM files. Current parallelised functionality is
9532an important subset of samtools functionality, including view, index,
9533sort, markdup, and depth.")
9534 (license license:gpl2+)))
b8fffd3a
RW		 9535
9536(define-public ritornello
9537 (package
9538 (name "ritornello")
9539 (version "1.0.0")
9540 (source (origin
9541 (method url-fetch)
9542 (uri (string-append "https://github.com/KlugerLab/"
9543 "Ritornello/archive/v"
9544 version ".tar.gz"))
9545 (file-name (string-append name "-" version ".tar.gz"))
9546 (sha256
9547 (base32
9548 "02nik86gq9ljjriv6pamwlmqnfky3ads1fpklx6mc3hx6k40pg38"))))
9549 (build-system gnu-build-system)
9550 (arguments
9551 `(#:tests? #f ; there are no tests
9552 #:phases
9553 (modify-phases %standard-phases
9554 (add-after 'unpack 'patch-samtools-references
9555 (lambda* (#:key inputs #:allow-other-keys)
9556 (substitute* '("src/SamStream.h"
9557 "src/BufferedGenomeReader.h")
9558 (("<sam.h>") "<samtools/sam.h>"))
9559 #t))
9560 (delete 'configure)
9561 (replace 'install
9562 (lambda* (#:key inputs outputs #:allow-other-keys)
9563 (let* ((out (assoc-ref outputs "out"))
9564 (bin (string-append out "/bin/")))
9565 (mkdir-p bin)
9566 (install-file "bin/Ritornello" bin)
9567 #t))))))
9568 (inputs
9569 `(("samtools" ,samtools-0.1)
9570 ("fftw" ,fftw)
9571 ("boost" ,boost)
9572 ("zlib" ,zlib)))
9573 (home-page "https://github.com/KlugerLab/Ritornello")
9574 (synopsis "Control-free peak caller for ChIP-seq data")
9575 (description "Ritornello is a ChIP-seq peak calling algorithm based on
9576signal processing that can accurately call binding events without the need to
9577do a pair total DNA input or IgG control sample. It has been tested for use
9578with narrow binding events such as transcription factor ChIP-seq.")
9579 (license license:gpl3+)))
44f6c889
RW		 9580
9581(define-public trim-galore
9582 (package
9583 (name "trim-galore")
9584 (version "0.4.2")
9585 (source
9586 (origin
9587 (method url-fetch)
9588 (uri (string-append "http://www.bioinformatics.babraham.ac.uk/"
9589 "projects/trim_galore/trim_galore_v"
9590 version ".zip"))
9591 (sha256
9592 (base32
9593 "0b9qdxi4521gsrjvbhgky8g7kry9b5nx3byzaxkgxz7p4k8bn1mn"))))
9594 (build-system gnu-build-system)
9595 (arguments
9596 `(#:tests? #f ; no tests
9597 #:phases
9598 (modify-phases %standard-phases
9599 ;; The archive contains plain files.
9600 (replace 'unpack
9601 (lambda* (#:key source #:allow-other-keys)
9602 (zero? (system* "unzip" source))))
9603 (delete 'configure)
9604 (delete 'build)
9605 (add-after 'unpack 'hardcode-tool-references
9606 (lambda* (#:key inputs #:allow-other-keys)
9607 (substitute* "trim_galore"
9608 (("\\$path_to_cutadapt = 'cutadapt'")
9609 (string-append "$path_to_cutadapt = '"
9610 (assoc-ref inputs "cutadapt")
9611 "/bin/cutadapt'"))
9612 (("\\| gzip")
9613 (string-append "| "
9614 (assoc-ref inputs "gzip")
9615 "/bin/gzip"))
9616 (("\"gunzip")
9617 (string-append "\""
9618 (assoc-ref inputs "gzip")
9619 "/bin/gunzip")))
9620 #t))
9621 (replace 'install
9622 (lambda* (#:key outputs #:allow-other-keys)
9623 (let ((bin (string-append (assoc-ref outputs "out")
9624 "/bin")))
9625 (mkdir-p bin)
9626 (install-file "trim_galore" bin)
9627 #t))))))
9628 (inputs
9629 `(("gzip" ,gzip)
9630 ("perl" ,perl)
9631 ("cutadapt" ,cutadapt)))
9632 (native-inputs
9633 `(("unzip" ,unzip)))
9634 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/")
9635 (synopsis "Wrapper around Cutadapt and FastQC")
9636 (description "Trim Galore! is a wrapper script to automate quality and
9637adapter trimming as well as quality control, with some added functionality to
9638remove biased methylation positions for RRBS sequence files.")
9639 (license license:gpl3+)))
3420c905
RW		 9640
9641(define-public gess
9642 (package
9643 (name "gess")
9644 (version "1.0")
9645 (source (origin
9646 (method url-fetch)
9647 (uri (string-append "http://compbio.uthscsa.edu/"
9648 "GESS_Web/files/"
9649 "gess-" version ".src.tar.gz"))
9650 (sha256
9651 (base32
9652 "0hyk403kxscclzfs24pvdgiv0wm03kjcziqdrp5w46cb049gz0d7"))))
9653 (build-system gnu-build-system)
9654 (arguments
9655 `(#:tests? #f ; no tests
9656 #:phases
9657 (modify-phases %standard-phases
9658 (delete 'configure)
9659 (delete 'build)
9660 (replace 'install
9661 (lambda* (#:key inputs outputs #:allow-other-keys)
9662 (let* ((python (assoc-ref inputs "python"))
9663 (out (assoc-ref outputs "out"))
9664 (bin (string-append out "/bin/"))
9665 (target (string-append
9666 out "/lib/python2.7/site-packages/gess/")))
9667 (mkdir-p target)
9668 (copy-recursively "." target)
9669 ;; Make GESS.py executable
9670 (chmod (string-append target "GESS.py") #o555)
9671 ;; Add Python shebang to the top and make Matplotlib
9672 ;; usable.
9673 (substitute* (string-append target "GESS.py")
9674 (("\"\"\"Description:" line)
9675 (string-append "#!" (which "python") "
9676import matplotlib
9677matplotlib.use('Agg')
9678" line)))
9679 ;; Make sure GESS has all modules in its path
9680 (wrap-program (string-append target "GESS.py")
9681 `("PYTHONPATH" ":" prefix (,target ,(getenv "PYTHONPATH"))))
9682 (mkdir-p bin)
9683 (symlink (string-append target "GESS.py")
9684 (string-append bin "GESS.py"))
9685 #t))))))
9686 (inputs
9687 `(("python" ,python-2)
9688 ("python2-pysam" ,python2-pysam)
9689 ("python2-scipy" ,python2-scipy)
9690 ("python2-numpy" ,python2-numpy)
9691 ("python2-networkx" ,python2-networkx)
9692 ("python2-biopython" ,python2-biopython)))
9693 (home-page "http://compbio.uthscsa.edu/GESS_Web/")
9694 (synopsis "Detect exon-skipping events from raw RNA-seq data")
9695 (description
9696 "GESS is an implementation of a novel computational method to detect de
9697novo exon-skipping events directly from raw RNA-seq data without the prior
9698knowledge of gene annotation information. GESS stands for the graph-based
9699exon-skipping scanner detection scheme.")
9700 (license license:bsd-3)))
282b0151
RW		 9701
9702(define-public phylip
9703 (package
9704 (name "phylip")
9705 (version "3.696")
9706 (source
9707 (origin
9708 (method url-fetch)
9709 (uri (string-append "http://evolution.gs.washington.edu/phylip/"
9710 "download/phylip-" version ".tar.gz"))
9711 (sha256
9712 (base32
9713 "01jar1rayhr2gba2pgbw49m56rc5z4p5wn3ds0m188hrlln4a2nd"))))
9714 (build-system gnu-build-system)
9715 (arguments
9716 `(#:tests? #f ; no check target
9717 #:make-flags (list "-f" "Makefile.unx" "install")
9718 #:parallel-build? #f ; not supported
9719 #:phases
9720 (modify-phases %standard-phases
9721 (add-after 'unpack 'enter-dir
9722 (lambda _ (chdir "src") #t))
9723 (delete 'configure)
9724 (replace 'install
9725 (lambda* (#:key inputs outputs #:allow-other-keys)
9726 (let ((target (string-append (assoc-ref outputs "out")
9727 "/bin")))
9728 (mkdir-p target)
9729 (for-each (lambda (file)
9730 (install-file file target))
9731 (find-files "../exe" ".*")))
9732 #t)))))
9733 (home-page "http://evolution.genetics.washington.edu/phylip/")
9734 (synopsis "Tools for inferring phylogenies")
9735 (description "PHYLIP (the PHYLogeny Inference Package) is a package of
9736programs for inferring phylogenies (evolutionary trees).")
9737 (license license:bsd-2)))
aa163424
RW		 9738
9739(define-public imp
9740 (package
9741 (name "imp")
9742 (version "2.6.2")
9743 (source
9744 (origin
9745 (method url-fetch)
9746 (uri (string-append "https://integrativemodeling.org/"
9747 version "/download/imp-" version ".tar.gz"))
9748 (sha256
9749 (base32
9750 "0lxqx7vh79d771svr611dkilp6sn30qrbw8zvscbrm37v38d2j6h"))))
9751 (build-system cmake-build-system)
9752 (arguments
9753 `(;; FIXME: Some tests fail because they produce warnings, others fail
9754 ;; because the PYTHONPATH does not include the modeller's directory.
9755 #:tests? #f
9756 ;; Do not place libraries in an architecture-specific directory.
9757 #:configure-flags
9758 (list "-DCMAKE_INSTALL_LIBDIR=lib")))
9759 (inputs
9760 `(("boost" ,boost)
9761 ("gsl" ,gsl)
9762 ("swig" ,swig)
9763 ("hdf5" ,hdf5)
9764 ("fftw" ,fftw)
9765 ("python" ,python-2)))
9766 (propagated-inputs
9767 `(("python2-numpy" ,python2-numpy)
9768 ("python2-scipy" ,python2-scipy)
9769 ("python2-pandas" ,python2-pandas)
9770 ("python2-scikit-learn" ,python2-scikit-learn)
9771 ("python2-networkx" ,python2-networkx)))
9772 (home-page "https://integrativemodeling.org")
9773 (synopsis "Integrative modeling platform")
9774 (description "IMP's broad goal is to contribute to a comprehensive
9775structural characterization of biomolecules ranging in size and complexity
9776from small peptides to large macromolecular assemblies, by integrating data
9777from diverse biochemical and biophysical experiments. IMP provides a C++ and
9778Python toolbox for solving complex modeling problems, and a number of
9779applications for tackling some common problems in a user-friendly way.")
9780 ;; IMP is largely available under the GNU Lesser GPL; see the file
9781 ;; COPYING.LGPL for the full text of this license. Some IMP modules are
9782 ;; available under the GNU GPL (see the file COPYING.GPL).
9783 (license (list license:lgpl2.1+
9784 license:gpl3+))))
8befd094
RW		 9785
9786(define-public tadbit
9787 (package
9788 (name "tadbit")
9789 (version "0.2")
9790 (source (origin
9791 (method url-fetch)
9792 (uri (string-append "https://github.com/3DGenomes/TADbit/"
9793 "archive/v" version ".tar.gz"))
9794 (file-name (string-append name "-" version ".tar.gz"))
9795 (sha256
9796 (base32
9797 "1cnfqrl4685zar4nnw94j94nhvl2h29jm448nadqi1h05z6fdk4f"))))
9798 (build-system python-build-system)
9799 (arguments
9800 `(;; Tests are included and must be run after installation, but
9801 ;; they are incomplete and thus cannot be run.
9802 #:tests? #f
9803 #:python ,python-2
9804 #:phases
9805 (modify-phases %standard-phases
9806 (add-after 'unpack 'fix-problems-with-setup.py
9807 (lambda* (#:key outputs #:allow-other-keys)
9808 ;; setup.py opens these files for writing
9809 (chmod "_pytadbit/_version.py" #o664)
9810 (chmod "README.rst" #o664)
9811
9812 ;; Don't attempt to install the bash completions to
9813 ;; the home directory.
9814 (rename-file "extras/.bash_completion"
9815 "extras/tadbit")
9816 (substitute* "setup.py"
9817 (("\\(path.expanduser\\('~'\\)")
9818 (string-append "(\""
9819 (assoc-ref outputs "out")
9820 "/etc/bash_completion.d\""))
9821 (("extras/\\.bash_completion")
9822 "extras/tadbit"))
9823 #t)))))
9824 (inputs
9825 ;; TODO: add Chimera for visualization
9826 `(("imp" ,imp)
9827 ("mcl" ,mcl)
9828 ("python2-scipy" ,python2-scipy)
9829 ("python2-numpy" ,python2-numpy)
9830 ("python2-matplotlib" ,python2-matplotlib)
9831 ("python2-pysam" ,python2-pysam)))
9832 (home-page "http://3dgenomes.github.io/TADbit/")
9833 (synopsis "Analyze, model, and explore 3C-based data")
9834 (description
9835 "TADbit is a complete Python library to deal with all steps to analyze,
9836model, and explore 3C-based data. With TADbit the user can map FASTQ files to
9837obtain raw interaction binned matrices (Hi-C like matrices), normalize and
a147fadd 9838correct interaction matrices, identify and compare the so-called
8befd094
RW		 9839@dfn{Topologically Associating Domains} (TADs), build 3D models from the
9840interaction matrices, and finally, extract structural properties from the
9841models. TADbit is complemented by TADkit for visualizing 3D models.")
9842 (license license:gpl3+)))
94820951
RW		 9843
9844(define-public kentutils
9845 (package
9846 (name "kentutils")
9847 ;; 302.1.0 is out, but the only difference is the inclusion of
9848 ;; pre-built binaries.
9849 (version "302.0.0")
9850 (source
9851 (origin
9852 (method url-fetch)
9853 (uri (string-append "https://github.com/ENCODE-DCC/kentUtils/"
9854 "archive/v" version ".tar.gz"))
9855 (file-name (string-append name "-" version ".tar.gz"))
9856 (sha256
9857 (base32
9858 "134aja3k1cj32kbk1nnw0q9gxjb2krr15q6sga8qldzvc0585rmm"))
9859 (modules '((guix build utils)
9860 (srfi srfi-26)
9861 (ice-9 ftw)))
9862 (snippet
9863 '(begin
9864 ;; Only the contents of the specified directories are free
9865 ;; for all uses, so we remove the rest. "hg/autoSql" and
9866 ;; "hg/autoXml" are nominally free, but they depend on a
9867 ;; library that is built from the sources in "hg/lib",
9868 ;; which is nonfree.
9869 (let ((free (list "." ".."
9870 "utils" "lib" "inc" "tagStorm"
9871 "parasol" "htslib"))
9872 (directory? (lambda (file)
9873 (eq? 'directory (stat:type (stat file))))))
9874 (for-each (lambda (file)
9875 (and (directory? file)
9876 (delete-file-recursively file)))
9877 (map (cut string-append "src/" <>)
9878 (scandir "src"
9879 (lambda (file)
9880 (not (member file free)))))))
9881 ;; Only make the utils target, not the userApps target,
9882 ;; because that requires libraries we won't build.
9883 (substitute* "Makefile"
9884 ((" userApps") " utils"))
9885 ;; Only build libraries that are free.
9886 (substitute* "src/makefile"
9887 (("DIRS =.*") "DIRS =\n")
9888 (("cd jkOwnLib.*") "")
9889 ((" hgLib") "")
9890 (("cd hg.*") ""))
9891 (substitute* "src/utils/makefile"
9892 ;; These tools depend on "jkhgap.a", which is part of the
9893 ;; nonfree "src/hg/lib" directory.
9894 (("raSqlQuery") "")
9895 (("pslLiftSubrangeBlat") "")
9896
9897 ;; Do not build UCSC tools, which may require nonfree
9898 ;; components.
9899 (("ALL_APPS =.*") "ALL_APPS = $(UTILS_APPLIST)\n"))
9900 #t))))
9901 (build-system gnu-build-system)
9902 (arguments
9903 `( ;; There is no global test target and the test target for
9904 ;; individual tools depends on input files that are not
9905 ;; included.
9906 #:tests? #f
9907 #:phases
9908 (modify-phases %standard-phases
9909 (add-after 'unpack 'fix-paths
9910 (lambda _
9911 (substitute* "Makefile"
9912 (("/bin/echo") (which "echo")))
9913 #t))
9914 (add-after 'unpack 'prepare-samtabix
9915 (lambda* (#:key inputs #:allow-other-keys)
9916 (copy-recursively (assoc-ref inputs "samtabix")
9917 "samtabix")
9918 #t))
9919 (delete 'configure)
9920 (replace 'install
9921 (lambda* (#:key outputs #:allow-other-keys)
9922 (let ((bin (string-append (assoc-ref outputs "out")
9923 "/bin")))
9924 (copy-recursively "bin" bin))
9925 #t)))))
9926 (native-inputs
9927 `(("samtabix"
9928 ,(origin
9929 (method git-fetch)
9930 (uri (git-reference
9931 (url "http://genome-source.cse.ucsc.edu/samtabix.git")
9932 (commit "10fd107909c1ac4d679299908be4262a012965ba")))
9933 (sha256
9934 (base32
9935 "0c1nj64l42v395sa84n7az43xiap4i6f9n9dfz4058aqiwkhkmma"))))))
9936 (inputs
9937 `(("zlib" ,zlib)
9938 ("tcsh" ,tcsh)
9939 ("perl" ,perl)
9940 ("libpng" ,libpng)
9941 ("mysql" ,mysql)
9942 ("openssl" ,openssl)))
9943 (home-page "http://genome.cse.ucsc.edu/index.html")
9944 (synopsis "Assorted bioinformatics utilities")
9945 (description "This package provides the kentUtils, a selection of
9946bioinformatics utilities used in combination with the UCSC genome
9947browser.")
9948 ;; Only a subset of the sources are released under a non-copyleft
9949 ;; free software license. All other sources are removed in a
9950 ;; snippet. See this bug report for an explanation of how the
9951 ;; license statements apply:
9952 ;; https://github.com/ENCODE-DCC/kentUtils/issues/12
9953 (license (license:non-copyleft
9954 "http://genome.ucsc.edu/license/"
9955 "The contents of this package are free for all uses."))))
7ceb0a83
RW		 9956
9957(define-public f-seq
9958 (let ((commit "6ccded34cff38cf432deed8503648b4a66953f9b")
9959 (revision "1"))
9960 (package
9961 (name "f-seq")
c6643f2d 9962 (version (string-append "1.1-" revision "." (string-take commit 7)))
7ceb0a83
RW		 9963 (source (origin
9964 (method git-fetch)
9965 (uri (git-reference
9966 (url "https://github.com/aboyle/F-seq.git")
9967 (commit commit)))
9968 (file-name (string-append name "-" version))
9969 (sha256
9970 (base32
9971 "1nk33k0yajg2id4g59bc4szr58r2q6pdq42vgcw054m8ip9wv26h"))
9972 (modules '((guix build utils)))
9973 ;; Remove bundled Java library archives.
9974 (snippet
9975 '(begin
9976 (for-each delete-file (find-files "lib" ".*"))
9977 #t))))
9978 (build-system ant-build-system)
9979 (arguments
9980 `(#:tests? #f ; no tests included
9981 #:phases
9982 (modify-phases %standard-phases
9983 (replace 'install
c0770fee 9984 (lambda* (#:key inputs outputs #:allow-other-keys)
7ceb0a83
RW		 9985 (let* ((target (assoc-ref outputs "out"))
9986 (doc (string-append target "/share/doc/f-seq/")))
9987 (mkdir-p target)
9988 (mkdir-p doc)
9989 (substitute* "bin/linux/fseq"
c0770fee
RW		 9990 (("java") (which "java"))
9991 (("\\$REALDIR/../lib/commons-cli-1.1.jar")
9992 (string-append (assoc-ref inputs "java-commons-cli")
9993 "/share/java/commons-cli.jar"))
9994 (("REALDIR=.*")
9995 (string-append "REALDIR=" target "/bin\n")))
7ceb0a83
RW		 9996 (install-file "README.txt" doc)
9997 (install-file "bin/linux/fseq" (string-append target "/bin"))
9998 (install-file "build~/fseq.jar" (string-append target "/lib"))
9999 (copy-recursively "lib" (string-append target "/lib"))
10000 #t))))))
10001 (inputs
10002 `(("perl" ,perl)
10003 ("java-commons-cli" ,java-commons-cli)))
10004 (home-page "http://fureylab.web.unc.edu/software/fseq/")
10005 (synopsis "Feature density estimator for high-throughput sequence tags")
10006 (description
10007 "F-Seq is a software package that generates a continuous tag sequence
10008density estimation allowing identification of biologically meaningful sites
10009such as transcription factor binding sites (ChIP-seq) or regions of open
10010chromatin (DNase-seq). Output can be displayed directly in the UCSC Genome
10011Browser.")
10012 (license license:gpl3+))))
96348048
RW		 10013
10014(define-public bismark
10015 (package
10016 (name "bismark")
10017 (version "0.16.3")
10018 (source
10019 (origin
10020 (method url-fetch)
10021 (uri (string-append "https://github.com/FelixKrueger/Bismark/"
10022 "archive/" version ".tar.gz"))
10023 (file-name (string-append name "-" version ".tar.gz"))
10024 (sha256
10025 (base32
10026 "1204i0pa02ll2jn5pnxypkclnskvv7a2nwh5nxhagmhxk9wfv9sq"))))
10027 (build-system perl-build-system)
10028 (arguments
10029 `(#:tests? #f ; there are no tests
10030 #:phases
10031 (modify-phases %standard-phases
10032 (delete 'configure)
10033 (delete 'build)
10034 (replace 'install
10035 (lambda* (#:key outputs #:allow-other-keys)
10036 (let ((bin (string-append (assoc-ref outputs "out")
10037 "/bin"))
10038 (docdir (string-append (assoc-ref outputs "out")
10039 "/share/doc/bismark"))
10040 (docs '("Bismark_User_Guide.pdf"
10041 "RELEASE_NOTES.txt"))
10042 (scripts '("bismark"
10043 "bismark_genome_preparation"
10044 "bismark_methylation_extractor"
10045 "bismark2bedGraph"
10046 "bismark2report"
10047 "coverage2cytosine"
10048 "deduplicate_bismark"
10049 "bismark_sitrep.tpl"
10050 "bam2nuc"
10051 "bismark2summary")))
10052 (mkdir-p docdir)
10053 (mkdir-p bin)
10054 (for-each (lambda (file) (install-file file bin))
10055 scripts)
10056 (for-each (lambda (file) (install-file file docdir))
10057 docs)
10058 #t))))))
10059 (home-page "http://www.bioinformatics.babraham.ac.uk/projects/bismark/")
10060 (synopsis "Map bisulfite treated sequence reads and analyze methylation")
10061 (description "Bismark is a program to map bisulfite treated sequencing
10062reads to a genome of interest and perform methylation calls in a single step.
10063The output can be easily imported into a genome viewer, such as SeqMonk, and
10064enables a researcher to analyse the methylation levels of their samples
10065straight away. Its main features are:
10066
10067@itemize
10068@item Bisulfite mapping and methylation calling in one single step
10069@item Supports single-end and paired-end read alignments
10070@item Supports ungapped and gapped alignments
10071@item Alignment seed length, number of mismatches etc are adjustable
10072@item Output discriminates between cytosine methylation in CpG, CHG
10073 and CHH context
10074@end itemize\n")
10075 (license license:gpl3+)))
dc8ddbfb
RW		 10076
10077(define-public paml
10078 (package
10079 (name "paml")
10080 (version "4.9e")
10081 (source (origin
10082 (method url-fetch)
10083 (uri (string-append "http://abacus.gene.ucl.ac.uk/software/"
10084 "paml" version ".tgz"))
10085 (sha256
10086 (base32
10087 "13zf6h9fiqghwhch2h06x1zdr6s42plsnqahflp5g7myr3han3s6"))
10088 (modules '((guix build utils)))
10089 ;; Remove Windows binaries
10090 (snippet
10091 '(begin
10092 (for-each delete-file (find-files "." "\\.exe$"))
10093 #t))))
10094 (build-system gnu-build-system)
10095 (arguments
10096 `(#:tests? #f ; there are no tests
10097 #:make-flags '("CC=gcc")
10098 #:phases
10099 (modify-phases %standard-phases
10100 (replace 'configure
10101 (lambda _
10102 (substitute* "src/BFdriver.c"
10103 (("/bin/bash") (which "bash")))
10104 (chdir "src")
10105 #t))
10106 (replace 'install
10107 (lambda* (#:key outputs #:allow-other-keys)
10108 (let ((tools '("baseml" "basemlg" "codeml"
10109 "pamp" "evolver" "yn00" "chi2"))
10110 (bin (string-append (assoc-ref outputs "out") "/bin"))
10111 (docdir (string-append (assoc-ref outputs "out")
10112 "/share/doc/paml")))
10113 (mkdir-p bin)
10114 (for-each (lambda (file) (install-file file bin)) tools)
10115 (copy-recursively "../doc" docdir)
10116 #t))))))
10117 (home-page "http://abacus.gene.ucl.ac.uk/software/paml.html")
10118 (synopsis "Phylogentic analysis by maximum likelihood")
10119 (description "PAML (for Phylogentic Analysis by Maximum Likelihood)
10120contains a few programs for model fitting and phylogenetic tree reconstruction
10121using nucleotide or amino-acid sequence data.")
10122 ;; GPLv3 only
10123 (license license:gpl3)))
jackhill's copy of GNU Guix: https://guix.gnu.org/