gnu/packages/bioconductor.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2016, 2017, 2018, 2019 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016, 2017, 2018 Roel Janssen <roel@gnu.org>
   4 ;;; Copyright © 2017, 2018 Tobias Geerinckx-Rice <me@tobias.gr>
   5 ;;;
   6 ;;; This file is part of GNU Guix.
   7 ;;;
   8 ;;; GNU Guix is free software; you can redistribute it and/or modify it
   9 ;;; under the terms of the GNU General Public License as published by
  10 ;;; the Free Software Foundation; either version 3 of the License, or (at
  11 ;;; your option) any later version.
  12 ;;;
  13 ;;; GNU Guix is distributed in the hope that it will be useful, but
  14 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  15 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 ;;; GNU General Public License for more details.
  17 ;;;
  18 ;;; You should have received a copy of the GNU General Public License
  19 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  20
  21 (define-module (gnu packages bioconductor)
  22   #:use-module ((guix licenses) #:prefix license:)
  23   #:use-module (guix packages)
  24   #:use-module (guix download)
  25   #:use-module (guix build-system r)
  26   #:use-module (gnu packages)
  27   #:use-module (gnu packages bioinformatics)
  28   #:use-module (gnu packages cran)
  29   #:use-module (gnu packages compression)
  30   #:use-module (gnu packages gcc)
  31   #:use-module (gnu packages graph)
  32   #:use-module (gnu packages maths)
  33   #:use-module (gnu packages statistics)
  34   #:use-module (gnu packages web))
  35
  36 (define-public r-bsgenome-celegans-ucsc-ce6
  37   (package
  38     (name "r-bsgenome-celegans-ucsc-ce6")
  39     (version "1.4.0")
  40     (source (origin
  41               (method url-fetch)
  42               ;; We cannot use bioconductor-uri here because this tarball is
  43               ;; located under "data/annotation/" instead of "bioc/".
  44               (uri (string-append "https://www.bioconductor.org/packages/"
  45                                   "release/data/annotation/src/contrib/"
  46                                   "BSgenome.Celegans.UCSC.ce6_"
  47                                   version ".tar.gz"))
  48               (sha256
  49                (base32
  50                 "0mqzb353xv2c3m3vkb315dkmnxkgczp7ndnknyhpgjlybyf715v9"))))
  51     (properties
  52      `((upstream-name . "BSgenome.Celegans.UCSC.ce6")))
  53     (build-system r-build-system)
  54     ;; As this package provides little more than a very large data file it
  55     ;; doesn't make sense to build substitutes.
  56     (arguments `(#:substitutable? #f))
  57     (propagated-inputs
  58      `(("r-bsgenome" ,r-bsgenome)))
  59     (home-page
  60      "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce6/")
  61     (synopsis "Full genome sequences for Worm")
  62     (description
  63      "This package provides full genome sequences for Caenorhabditis
  64 elegans (Worm) as provided by UCSC (ce6, May 2008) and stored in Biostrings
  65 objects.")
  66     (license license:artistic2.0)))
  67
  68 (define-public r-bsgenome-celegans-ucsc-ce10
  69   (package
  70     (name "r-bsgenome-celegans-ucsc-ce10")
  71     (version "1.4.0")
  72     (source (origin
  73               (method url-fetch)
  74               ;; We cannot use bioconductor-uri here because this tarball is
  75               ;; located under "data/annotation/" instead of "bioc/".
  76               (uri (string-append "https://www.bioconductor.org/packages/"
  77                                   "release/data/annotation/src/contrib/"
  78                                   "BSgenome.Celegans.UCSC.ce10_"
  79                                   version ".tar.gz"))
  80               (sha256
  81                (base32
  82                 "1zaym97jk4npxk14ifvwz2rvhm4zx9xgs33r9vvx9rlynp0gydrk"))))
  83     (properties
  84      `((upstream-name . "BSgenome.Celegans.UCSC.ce10")))
  85     (build-system r-build-system)
  86     ;; As this package provides little more than a very large data file it
  87     ;; doesn't make sense to build substitutes.
  88     (arguments `(#:substitutable? #f))
  89     (propagated-inputs
  90      `(("r-bsgenome" ,r-bsgenome)))
  91     (home-page
  92      "https://www.bioconductor.org/packages/BSgenome.Celegans.UCSC.ce10/")
  93     (synopsis "Full genome sequences for Worm")
  94     (description
  95      "This package provides full genome sequences for Caenorhabditis
  96 elegans (Worm) as provided by UCSC (ce10, Oct 2010) and stored in Biostrings
  97 objects.")
  98     (license license:artistic2.0)))
  99
 100 (define-public r-bsgenome-dmelanogaster-ucsc-dm6
 101   (package
 102     (name "r-bsgenome-dmelanogaster-ucsc-dm6")
 103     (version "1.4.1")
 104     (source (origin
 105               (method url-fetch)
 106               ;; We cannot use bioconductor-uri here because this tarball is
 107               ;; located under "data/annotation/" instead of "bioc/".
 108               (uri (string-append "https://www.bioconductor.org/packages/"
 109                                   "release/data/annotation/src/contrib/"
 110                                   "BSgenome.Dmelanogaster.UCSC.dm6_"
 111                                   version ".tar.gz"))
 112               (sha256
 113                (base32
 114                 "1bhj0rdgf7lspw4xby9y9mf7v7jxxz8001bc8vw8kf04rjsx6060"))))
 115     (properties
 116      `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm6")))
 117     (build-system r-build-system)
 118     ;; As this package provides little more than a very large data file it
 119     ;; doesn't make sense to build substitutes.
 120     (arguments `(#:substitutable? #f))
 121     (propagated-inputs
 122      `(("r-bsgenome" ,r-bsgenome)))
 123     (home-page
 124      "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm6/")
 125     (synopsis "Full genome sequences for Fly")
 126     (description
 127      "This package provides full genome sequences for Drosophila
 128 melanogaster (Fly) as provided by UCSC (dm6) and stored in Biostrings
 129 objects.")
 130     (license license:artistic2.0)))
 131
 132 (define-public r-bsgenome-dmelanogaster-ucsc-dm3
 133   (package
 134     (name "r-bsgenome-dmelanogaster-ucsc-dm3")
 135     (version "1.4.0")
 136     (source (origin
 137               (method url-fetch)
 138               ;; We cannot use bioconductor-uri here because this tarball is
 139               ;; located under "data/annotation/" instead of "bioc/".
 140               (uri (string-append "https://www.bioconductor.org/packages/"
 141                                   "release/data/annotation/src/contrib/"
 142                                   "BSgenome.Dmelanogaster.UCSC.dm3_"
 143                                   version ".tar.gz"))
 144               (sha256
 145                (base32
 146                 "19bm3lkhhkag3gnwp419211fh0cnr0x6fa0r1lr0ycwrikxdxsv8"))))
 147     (properties
 148      `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3")))
 149     (build-system r-build-system)
 150     ;; As this package provides little more than a very large data file it
 151     ;; doesn't make sense to build substitutes.
 152     (arguments `(#:substitutable? #f))
 153     (propagated-inputs
 154      `(("r-bsgenome" ,r-bsgenome)))
 155     (home-page
 156      "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3/")
 157     (synopsis "Full genome sequences for Fly")
 158     (description
 159      "This package provides full genome sequences for Drosophila
 160 melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
 161 Biostrings objects.")
 162     (license license:artistic2.0)))
 163
 164 (define-public r-bsgenome-dmelanogaster-ucsc-dm3-masked
 165   (package
 166     (name "r-bsgenome-dmelanogaster-ucsc-dm3-masked")
 167     (version "1.3.99")
 168     (source (origin
 169               (method url-fetch)
 170               ;; We cannot use bioconductor-uri here because this tarball is
 171               ;; located under "data/annotation/" instead of "bioc/".
 172               (uri (string-append "http://www.bioconductor.org/packages/"
 173                                   "release/data/annotation/src/contrib/"
 174                                   "BSgenome.Dmelanogaster.UCSC.dm3.masked_"
 175                                   version ".tar.gz"))
 176               (sha256
 177                (base32
 178                 "1756csb09f1br9rj1l3f08qyh4hlymdbd0cfn8x3fq39dn45m5ap"))))
 179     (properties
 180      `((upstream-name . "BSgenome.Dmelanogaster.UCSC.dm3.masked")))
 181     (build-system r-build-system)
 182     (propagated-inputs
 183      `(("r-bsgenome" ,r-bsgenome)
 184        ("r-bsgenome-dmelanogaster-ucsc-dm3"
 185         ,r-bsgenome-dmelanogaster-ucsc-dm3)))
 186     (home-page "https://www.bioconductor.org/packages/BSgenome.Dmelanogaster.UCSC.dm3.masked/")
 187     (synopsis "Full masked genome sequences for Fly")
 188     (description
 189      "This package provides full masked genome sequences for Drosophila
 190 melanogaster (Fly) as provided by UCSC (dm3, April 2006) and stored in
 191 Biostrings objects.  The sequences are the same as in
 192 BSgenome.Dmelanogaster.UCSC.dm3, except that each of them has the 4 following
 193 masks on top: (1) the mask of assembly gaps (AGAPS mask), (2) the mask of
 194 intra-contig ambiguities (AMB mask), (3) the mask of repeats from
 195 RepeatMasker (RM mask), and (4) the mask of repeats from Tandem Repeats
 196 Finder (TRF mask).  Only the AGAPS and AMB masks are \"active\" by default.")
 197     (license license:artistic2.0)))
 198
 199 (define-public r-bsgenome-hsapiens-1000genomes-hs37d5
 200   (package
 201     (name "r-bsgenome-hsapiens-1000genomes-hs37d5")
 202     (version "0.99.1")
 203     (source (origin
 204               (method url-fetch)
 205               ;; We cannot use bioconductor-uri here because this tarball is
 206               ;; located under "data/annotation/" instead of "bioc/".
 207               (uri (string-append "https://www.bioconductor.org/packages/"
 208                                   "release/data/annotation/src/contrib/"
 209                                   "BSgenome.Hsapiens.1000genomes.hs37d5_"
 210                                   version ".tar.gz"))
 211               (sha256
 212                (base32
 213                 "1cg0g5fqmsvwyw2p9hp2yy4ilk21jkbbrnpgqvb5c36ihjwvc7sr"))))
 214     (properties
 215      `((upstream-name . "BSgenome.Hsapiens.1000genomes.hs37d5")))
 216     (build-system r-build-system)
 217     ;; As this package provides little more than a very large data file it
 218     ;; doesn't make sense to build substitutes.
 219     (arguments `(#:substitutable? #f))
 220     (propagated-inputs
 221      `(("r-bsgenome" ,r-bsgenome)))
 222     (home-page
 223      "https://www.bioconductor.org/packages/BSgenome.Hsapiens.1000genomes.hs37d5/")
 224     (synopsis "Full genome sequences for Homo sapiens")
 225     (description
 226      "This package provides full genome sequences for Homo sapiens from
 227 1000genomes phase2 reference genome sequence (hs37d5), based on NCBI GRCh37.")
 228     (license license:artistic2.0)))
 229
 230 (define-public r-bsgenome-hsapiens-ucsc-hg19-masked
 231   (package
 232     (name "r-bsgenome-hsapiens-ucsc-hg19-masked")
 233     (version "1.3.99")
 234     (source (origin
 235               (method url-fetch)
 236               ;; We cannot use bioconductor-uri here because this tarball is
 237               ;; located under "data/annotation/" instead of "bioc/".
 238               (uri (string-append "http://www.bioconductor.org/packages/"
 239                                   "release/data/annotation/src/contrib/"
 240                                   "BSgenome.Hsapiens.UCSC.hg19.masked_"
 241                                   version ".tar.gz"))
 242               (sha256
 243                (base32
 244                 "0452pyah0kv1vsrsjbrqw4k2rm8lc2vc771dzib45gnnfz86qxrr"))))
 245     (properties
 246      `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19.masked")))
 247     (build-system r-build-system)
 248     (propagated-inputs
 249      `(("r-bsgenome" ,r-bsgenome)
 250        ("r-bsgenome-hsapiens-ucsc-hg19"
 251         ,r-bsgenome-hsapiens-ucsc-hg19)))
 252     (home-page "https://bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19.masked/")
 253     (synopsis "Full masked genome sequences for Homo sapiens")
 254     (description
 255      "This package provides full genome sequences for Homo sapiens (Human) as
 256 provided by UCSC (hg19, Feb. 2009) and stored in Biostrings objects.  The
 257 sequences are the same as in BSgenome.Hsapiens.UCSC.hg19, except that each of
 258 them has the 4 following masks on top: (1) the mask of assembly gaps (AGAPS
 259 mask), (2) the mask of intra-contig ambiguities (AMB mask), (3) the mask of
 260 repeats from RepeatMasker (RM mask), and (4) the mask of repeats from Tandem
 261 Repeats Finder (TRF mask).  Only the AGAPS and AMB masks are \"active\" by
 262 default.")
 263     (license license:artistic2.0)))
 264
 265 (define-public r-bsgenome-mmusculus-ucsc-mm9
 266   (package
 267     (name "r-bsgenome-mmusculus-ucsc-mm9")
 268     (version "1.4.0")
 269     (source (origin
 270               (method url-fetch)
 271               ;; We cannot use bioconductor-uri here because this tarball is
 272               ;; located under "data/annotation/" instead of "bioc/".
 273               (uri (string-append "https://www.bioconductor.org/packages/"
 274                                   "release/data/annotation/src/contrib/"
 275                                   "BSgenome.Mmusculus.UCSC.mm9_"
 276                                   version ".tar.gz"))
 277               (sha256
 278                (base32
 279                 "1birqw30g2azimxpnjfzmkphan7x131yy8b9h85lfz5fjdg7841i"))))
 280     (properties
 281      `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9")))
 282     (build-system r-build-system)
 283     ;; As this package provides little more than a very large data file it
 284     ;; doesn't make sense to build substitutes.
 285     (arguments `(#:substitutable? #f))
 286     (propagated-inputs
 287      `(("r-bsgenome" ,r-bsgenome)))
 288     (home-page
 289      "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9/")
 290     (synopsis "Full genome sequences for Mouse")
 291     (description
 292      "This package provides full genome sequences for Mus musculus (Mouse) as
 293 provided by UCSC (mm9, July 2007) and stored in Biostrings objects.")
 294     (license license:artistic2.0)))
 295
 296 (define-public r-bsgenome-mmusculus-ucsc-mm9-masked
 297   (package
 298     (name "r-bsgenome-mmusculus-ucsc-mm9-masked")
 299     (version "1.3.99")
 300     (source (origin
 301               (method url-fetch)
 302               ;; We cannot use bioconductor-uri here because this tarball is
 303               ;; located under "data/annotation/" instead of "bioc/".
 304               (uri (string-append "http://www.bioconductor.org/packages/"
 305                                   "release/data/annotation/src/contrib/"
 306                                   "BSgenome.Mmusculus.UCSC.mm9.masked_"
 307                                   version ".tar.gz"))
 308               (sha256
 309                (base32
 310                 "00bpbm3havqcxr4g63zhllsbpd9q6svgihks7qp7x73nm4gvq7fn"))))
 311     (properties
 312      `((upstream-name . "BSgenome.Mmusculus.UCSC.mm9.masked")))
 313     (build-system r-build-system)
 314     (propagated-inputs
 315      `(("r-bsgenome" ,r-bsgenome)
 316        ("r-bsgenome-mmusculus-ucsc-mm9"
 317         ,r-bsgenome-mmusculus-ucsc-mm9)))
 318     (home-page "http://bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm9.masked/")
 319     (synopsis "Full masked genome sequences for Mouse")
 320     (description
 321      "This package provides full genome sequences for Mus musculus (Mouse) as
 322 provided by UCSC (mm9, Jul. 2007) and stored in Biostrings objects.  The
 323 sequences are the same as in BSgenome.Mmusculus.UCSC.mm9, except that each of
 324 them has the 4 following masks on top: (1) the mask of assembly gaps (AGAPS
 325 mask), (2) the mask of intra-contig ambiguities (AMB mask), (3) the mask of
 326 repeats from RepeatMasker (RM mask), and (4) the mask of repeats from Tandem
 327 Repeats Finder (TRF mask).  Only the AGAPS and AMB masks are \"active\" by
 328 default."  )
 329     (license license:artistic2.0)))
 330
 331 (define-public r-bsgenome-mmusculus-ucsc-mm10
 332   (package
 333     (name "r-bsgenome-mmusculus-ucsc-mm10")
 334     (version "1.4.0")
 335     (source (origin
 336               (method url-fetch)
 337               ;; We cannot use bioconductor-uri here because this tarball is
 338               ;; located under "data/annotation/" instead of "bioc/".
 339               (uri (string-append "https://www.bioconductor.org/packages/"
 340                                   "release/data/annotation/src/contrib/"
 341                                   "BSgenome.Mmusculus.UCSC.mm10_"
 342                                   version ".tar.gz"))
 343               (sha256
 344                (base32
 345                 "12s0nm2na9brjad4rn9l7d3db2aj8qa1xvz0y1k7gk08wayb6bkf"))))
 346     (properties
 347      `((upstream-name . "BSgenome.Mmusculus.UCSC.mm10")))
 348     (build-system r-build-system)
 349     ;; As this package provides little more than a very large data file it
 350     ;; doesn't make sense to build substitutes.
 351     (arguments `(#:substitutable? #f))
 352     (propagated-inputs
 353      `(("r-bsgenome" ,r-bsgenome)))
 354     (home-page
 355      "https://www.bioconductor.org/packages/BSgenome.Mmusculus.UCSC.mm10/")
 356     (synopsis "Full genome sequences for Mouse")
 357     (description
 358      "This package provides full genome sequences for Mus
 359 musculus (Mouse) as provided by UCSC (mm10, December 2011) and stored
 360 in Biostrings objects.")
 361     (license license:artistic2.0)))
 362
 363 (define-public r-org-ce-eg-db
 364   (package
 365     (name "r-org-ce-eg-db")
 366     (version "3.7.0")
 367     (source (origin
 368               (method url-fetch)
 369               ;; We cannot use bioconductor-uri here because this tarball is
 370               ;; located under "data/annotation/" instead of "bioc/".
 371               (uri (string-append "https://www.bioconductor.org/packages/"
 372                                   "release/data/annotation/src/contrib/"
 373                                   "org.Ce.eg.db_" version ".tar.gz"))
 374               (sha256
 375                (base32
 376                 "1w5br1ss4ha8wv4v2saj7cmbjc2jw0dyj2f2y269l078z31wcnaz"))))
 377     (properties
 378      `((upstream-name . "org.Ce.eg.db")))
 379     (build-system r-build-system)
 380     (propagated-inputs
 381      `(("r-annotationdbi" ,r-annotationdbi)))
 382     (home-page "https://www.bioconductor.org/packages/org.Ce.eg.db/")
 383     (synopsis "Genome wide annotation for Worm")
 384     (description
 385      "This package provides mappings from Entrez gene identifiers to various
 386 annotations for the genome of the model worm Caenorhabditis elegans.")
 387     (license license:artistic2.0)))
 388
 389 (define-public r-org-dr-eg-db
 390   (package
 391     (name "r-org-dr-eg-db")
 392     (version "3.7.0")
 393     (source (origin
 394               (method url-fetch)
 395               ;; We cannot use bioconductor-uri here because this tarball is
 396               ;; located under "data/annotation/" instead of "bioc/".
 397               (uri (string-append "https://www.bioconductor.org/packages/"
 398                                   "release/data/annotation/src/contrib/"
 399                                   "org.Dr.eg.db_" version ".tar.gz"))
 400               (sha256
 401                (base32
 402                 "1xs5wsbcpy0iwbjyiv7fax57djqc529ai5fk1qfsdcvlja3cpglx"))))
 403     (properties
 404      `((upstream-name . "org.Dr.eg.db")))
 405     (build-system r-build-system)
 406     (propagated-inputs
 407      `(("r-annotationdbi" ,r-annotationdbi)))
 408     (home-page "https://www.bioconductor.org/packages/org.Dr.eg.db/")
 409     (synopsis "Annotation for Zebrafish")
 410     (description
 411      "This package provides genome wide annotations for Zebrafish, primarily
 412 based on mapping using Entrez Gene identifiers.")
 413     (license license:artistic2.0)))
 414
 415 (define-public r-bsgenome-hsapiens-ucsc-hg19
 416   (package
 417     (name "r-bsgenome-hsapiens-ucsc-hg19")
 418     (version "1.4.0")
 419     (source (origin
 420               (method url-fetch)
 421               ;; We cannot use bioconductor-uri here because this tarball is
 422               ;; located under "data/annotation/" instead of "bioc/".
 423               (uri (string-append "https://www.bioconductor.org/packages/"
 424                                   "release/data/annotation/src/contrib/"
 425                                   "BSgenome.Hsapiens.UCSC.hg19_"
 426                                   version ".tar.gz"))
 427               (sha256
 428                (base32
 429                 "1y0nqpk8cw5a34sd9hmin3z4v7iqm6hf6l22cl81vlbxqbjibxc8"))))
 430     (properties
 431      `((upstream-name . "BSgenome.Hsapiens.UCSC.hg19")))
 432     (build-system r-build-system)
 433     ;; As this package provides little more than a very large data file it
 434     ;; doesn't make sense to build substitutes.
 435     (arguments `(#:substitutable? #f))
 436     (propagated-inputs
 437      `(("r-bsgenome" ,r-bsgenome)))
 438     (home-page
 439      "https://www.bioconductor.org/packages/BSgenome.Hsapiens.UCSC.hg19/")
 440     (synopsis "Full genome sequences for Homo sapiens")
 441     (description
 442      "This package provides full genome sequences for Homo sapiens as provided
 443 by UCSC (hg19, February 2009) and stored in Biostrings objects.")
 444     (license license:artistic2.0)))
 445
 446 (define-public r-genelendatabase
 447   (package
 448     (name "r-genelendatabase")
 449     (version "1.16.0")
 450     (source
 451      (origin
 452        (method url-fetch)
 453        ;; We cannot use bioconductor-uri here because this tarball is
 454        ;; located under "data/experiment/" instead of "bioc/".
 455        (uri (string-append "https://bioconductor.org/packages/"
 456                            "release/data/experiment/src/contrib"
 457                            "/geneLenDataBase_" version ".tar.gz"))
 458        (sha256
 459         (base32
 460          "07mmmn53kb7v14msc13dsbm8ghssbvwcrhifrk15hn37bw2p4ja5"))))
 461     (properties
 462      `((upstream-name . "geneLenDataBase")))
 463     (build-system r-build-system)
 464     (propagated-inputs
 465      `(("r-rtracklayer" ,r-rtracklayer)
 466        ("r-genomicfeatures" ,r-genomicfeatures)))
 467     (home-page "https://bioconductor.org/packages/geneLenDataBase/")
 468     (synopsis "Lengths of mRNA transcripts for a number of genomes")
 469     (description
 470      "This package provides the lengths of mRNA transcripts for a number of
 471 genomes and gene ID formats, largely based on the UCSC table browser.")
 472     (license license:lgpl2.0+)))
 473
 474 (define-public r-txdb-hsapiens-ucsc-hg19-knowngene
 475   (package
 476     (name "r-txdb-hsapiens-ucsc-hg19-knowngene")
 477     (version "3.2.2")
 478     (source (origin
 479               (method url-fetch)
 480               ;; We cannot use bioconductor-uri here because this tarball is
 481               ;; located under "data/annotation/" instead of "bioc/".
 482               (uri (string-append "https://bioconductor.org/packages/"
 483                                   "release/data/annotation/src/contrib"
 484                                   "/TxDb.Hsapiens.UCSC.hg19.knownGene_"
 485                                   version ".tar.gz"))
 486               (sha256
 487                (base32
 488                 "1sajhcqqwazgz2lqbik7rd935i7kpnh08zxbp2ra10j72yqy4g86"))))
 489     (properties
 490      `((upstream-name . "TxDb.Hsapiens.UCSC.hg19.knownGene")))
 491     (build-system r-build-system)
 492     ;; As this package provides little more than a very large data file it
 493     ;; doesn't make sense to build substitutes.
 494     (arguments `(#:substitutable? #f))
 495     (propagated-inputs
 496      `(("r-genomicfeatures" ,r-genomicfeatures)))
 497     (home-page
 498      "https://bioconductor.org/packages/TxDb.Hsapiens.UCSC.hg19.knownGene/")
 499     (synopsis "Annotation package for human genome in TxDb format")
 500     (description
 501      "This package provides an annotation database of Homo sapiens genome
 502 data.  It is derived from the UCSC hg19 genome and based on the \"knownGene\"
 503 track.  The database is exposed as a @code{TxDb} object.")
 504     (license license:artistic2.0)))
 505
 506 (define-public r-txdb-mmusculus-ucsc-mm9-knowngene
 507   (package
 508     (name "r-txdb-mmusculus-ucsc-mm9-knowngene")
 509     (version "3.2.2")
 510     (source (origin
 511               (method url-fetch)
 512               ;; We cannot use bioconductor-uri here because this tarball is
 513               ;; located under "data/annotation/" instead of "bioc/".
 514               (uri (string-append "https://bioconductor.org/packages/"
 515                                   "release/data/annotation/src/contrib"
 516                                   "/TxDb.Mmusculus.UCSC.mm9.knownGene_"
 517                                   version ".tar.gz"))
 518               (sha256
 519                (base32
 520                 "16bjxy00363hf91ik2mqlqls86i07gia72qh92xc3l1ncch61mx2"))))
 521     (properties
 522      `((upstream-name . "TxDb.Mmusculus.UCSC.mm9.knownGene")))
 523     (build-system r-build-system)
 524     (propagated-inputs
 525      `(("r-genomicfeatures" ,r-genomicfeatures)
 526        ("r-annotationdbi" ,r-annotationdbi)))
 527     (home-page
 528      "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm9.knownGene/")
 529     (synopsis "Annotation package for mouse genome in TxDb format")
 530     (description
 531      "This package provides an annotation database of Mouse genome data.  It
 532 is derived from the UCSC mm9 genome and based on the \"knownGene\" track.  The
 533 database is exposed as a @code{TxDb} object.")
 534     (license license:artistic2.0)))
 535
 536 (define-public r-txdb-mmusculus-ucsc-mm10-knowngene
 537   (package
 538     (name "r-txdb-mmusculus-ucsc-mm10-knowngene")
 539     (version "3.4.4")
 540     (source (origin
 541               (method url-fetch)
 542               ;; We cannot use bioconductor-uri here because this tarball is
 543               ;; located under "data/annotation/" instead of "bioc/".
 544               (uri (string-append "https://www.bioconductor.org/packages/"
 545                                   "release/data/annotation/src/contrib/"
 546                                   "TxDb.Mmusculus.UCSC.mm10.knownGene_"
 547                                   version ".tar.gz"))
 548               (sha256
 549                (base32
 550                 "01lgxc1fx5nhlpbwjd5zqghkkbmh6axd98ikx4b0spv0jdg6gf39"))))
 551     (properties
 552      `((upstream-name . "TxDb.Mmusculus.UCSC.mm10.knownGene")))
 553     (build-system r-build-system)
 554     ;; As this package provides little more than a very large data file it
 555     ;; doesn't make sense to build substitutes.
 556     (arguments `(#:substitutable? #f))
 557     (propagated-inputs
 558      `(("r-bsgenome" ,r-bsgenome)
 559        ("r-genomicfeatures" ,r-genomicfeatures)
 560        ("r-annotationdbi" ,r-annotationdbi)))
 561     (home-page
 562      "https://bioconductor.org/packages/TxDb.Mmusculus.UCSC.mm10.knownGene/")
 563     (synopsis "Annotation package for TxDb knownGene object(s) for Mouse")
 564     (description
 565      "This package loads a TxDb object, which is an R interface to
 566 prefabricated databases contained in this package.  This package provides
 567 the TxDb object of Mouse data as provided by UCSC (mm10, December 2011)
 568 based on the knownGene track.")
 569     (license license:artistic2.0)))
 570
 571 \f
 572 (define-public r-biocgenerics
 573   (package
 574     (name "r-biocgenerics")
 575     (version "0.28.0")
 576     (source (origin
 577               (method url-fetch)
 578               (uri (bioconductor-uri "BiocGenerics" version))
 579               (sha256
 580                (base32
 581                 "0cvpsrhg7sn7lpqgxvqrsagv6j7xj5rafq5xdjfd8zc4gxrs5rb8"))))
 582     (properties
 583      `((upstream-name . "BiocGenerics")))
 584     (build-system r-build-system)
 585     (home-page "https://bioconductor.org/packages/BiocGenerics")
 586     (synopsis "S4 generic functions for Bioconductor")
 587     (description
 588      "This package provides S4 generic functions needed by many Bioconductor
 589 packages.")
 590     (license license:artistic2.0)))
 591
 592 (define-public r-annotate
 593   (package
 594     (name "r-annotate")
 595     (version "1.60.0")
 596     (source
 597      (origin
 598        (method url-fetch)
 599        (uri (bioconductor-uri "annotate" version))
 600        (sha256
 601         (base32
 602          "0p6c96lay23a67dyirgnwzm2yw22m592z780vy6p4nqwla8ha18n"))))
 603     (build-system r-build-system)
 604     (propagated-inputs
 605      `(("r-annotationdbi" ,r-annotationdbi)
 606        ("r-biobase" ,r-biobase)
 607        ("r-biocgenerics" ,r-biocgenerics)
 608        ("r-dbi" ,r-dbi)
 609        ("r-rcurl" ,r-rcurl)
 610        ("r-xml" ,r-xml)
 611        ("r-xtable" ,r-xtable)))
 612     (home-page
 613      "https://bioconductor.org/packages/annotate")
 614     (synopsis "Annotation for microarrays")
 615     (description "This package provides R environments for the annotation of
 616 microarrays.")
 617     (license license:artistic2.0)))
 618
 619 (define-public r-hpar
 620   (package
 621     (name "r-hpar")
 622     (version "1.24.0")
 623     (source
 624      (origin
 625        (method url-fetch)
 626        (uri (bioconductor-uri "hpar" version))
 627        (sha256
 628         (base32
 629          "1pm3k8apgynmdzv2d0chca3b636kcai3b1x861fyv1m3xs6msgxn"))))
 630     (build-system r-build-system)
 631     (home-page "https://bioconductor.org/packages/hpar/")
 632     (synopsis "Human Protein Atlas in R")
 633     (description "This package provides a simple interface to and data from
 634 the Human Protein Atlas project.")
 635     (license license:artistic2.0)))
 636
 637 (define-public r-regioner
 638   (package
 639     (name "r-regioner")
 640     (version "1.14.0")
 641     (source
 642      (origin
 643        (method url-fetch)
 644        (uri (bioconductor-uri "regioneR" version))
 645        (sha256
 646         (base32
 647          "19la74swgzxp90z2nr3pzsgkxd7wp70zl6i2ipv3plg841f6k5zd"))))
 648     (properties `((upstream-name . "regioneR")))
 649     (build-system r-build-system)
 650     (propagated-inputs
 651      `(("r-biostrings" ,r-biostrings)
 652        ("r-bsgenome" ,r-bsgenome)
 653        ("r-genomeinfodb" ,r-genomeinfodb)
 654        ("r-genomicranges" ,r-genomicranges)
 655        ("r-iranges" ,r-iranges)
 656        ("r-memoise" ,r-memoise)
 657        ("r-rtracklayer" ,r-rtracklayer)
 658        ("r-s4vectors" ,r-s4vectors)))
 659     (home-page "https://bioconductor.org/packages/regioneR/")
 660     (synopsis "Association analysis of genomic regions")
 661     (description "This package offers a statistical framework based on
 662 customizable permutation tests to assess the association between genomic
 663 region sets and other genomic features.")
 664     (license license:artistic2.0)))
 665
 666 (define-public r-geneplotter
 667   (package
 668     (name "r-geneplotter")
 669     (version "1.60.0")
 670     (source
 671      (origin
 672        (method url-fetch)
 673        (uri (bioconductor-uri "geneplotter" version))
 674        (sha256
 675         (base32
 676          "10khr0pznxf3m0f5gzck9ymljrwcv3vamfmpskd51yjh36lhllqz"))))
 677     (build-system r-build-system)
 678     (propagated-inputs
 679      `(("r-annotate" ,r-annotate)
 680        ("r-annotationdbi" ,r-annotationdbi)
 681        ("r-biobase" ,r-biobase)
 682        ("r-biocgenerics" ,r-biocgenerics)
 683        ("r-lattice" ,r-lattice)
 684        ("r-rcolorbrewer" ,r-rcolorbrewer)))
 685     (home-page "https://bioconductor.org/packages/geneplotter")
 686     (synopsis "Graphics functions for genomic data")
 687     (description
 688      "This package provides functions for plotting genomic data.")
 689     (license license:artistic2.0)))
 690
 691 (define-public r-diffbind
 692   (package
 693     (name "r-diffbind")
 694     (version "2.10.0")
 695     (source
 696      (origin
 697        (method url-fetch)
 698        (uri (bioconductor-uri "DiffBind" version))
 699        (sha256
 700         (base32
 701          "0j8pal40lr1gv8sss96hhkj7l1qn9sy4q4l2kqd4rfwl7qrcnfw5"))))
 702     (properties `((upstream-name . "DiffBind")))
 703     (build-system r-build-system)
 704     (inputs
 705      `(("zlib" ,zlib)))
 706     (propagated-inputs
 707      `(("r-amap" ,r-amap)
 708        ("r-biocparallel" ,r-biocparallel)
 709        ("r-deseq2" ,r-deseq2)
 710        ("r-dplyr" ,r-dplyr)
 711        ("r-edger" ,r-edger)
 712        ("r-genomicalignments" ,r-genomicalignments)
 713        ("r-genomicranges" ,r-genomicranges)
 714        ("r-ggplot2" ,r-ggplot2)
 715        ("r-ggrepel" ,r-ggrepel)
 716        ("r-gplots" ,r-gplots)
 717        ("r-iranges" ,r-iranges)
 718        ("r-lattice" ,r-lattice)
 719        ("r-limma" ,r-limma)
 720        ("r-locfit" ,r-locfit)
 721        ("r-rcolorbrewer" , r-rcolorbrewer)
 722        ("r-rcpp" ,r-rcpp)
 723        ("r-rsamtools" ,r-rsamtools)
 724        ("r-s4vectors" ,r-s4vectors)
 725        ("r-summarizedexperiment" ,r-summarizedexperiment)
 726        ("r-systempiper" ,r-systempiper)
 727        ("r-zlibbioc" ,r-zlibbioc)))
 728     (home-page "http://bioconductor.org/packages/DiffBind")
 729     (synopsis "Differential binding analysis of ChIP-Seq peak data")
 730     (description
 731      "This package computes differentially bound sites from multiple
 732 ChIP-seq experiments using affinity (quantitative) data.  Also enables
 733 occupancy (overlap) analysis and plotting functions.")
 734     (license license:artistic2.0)))
 735
 736 (define-public r-ripseeker
 737   (package
 738     (name "r-ripseeker")
 739     (version "1.22.0")
 740     (source
 741      (origin
 742        (method url-fetch)
 743        (uri (bioconductor-uri "RIPSeeker" version))
 744        (sha256
 745         (base32
 746          "1x2n1iyik4s67bxq0fl6fpf602k51g4pxjpjpxkgx1a5fsk61f2i"))))
 747     (properties `((upstream-name . "RIPSeeker")))
 748     (build-system r-build-system)
 749     (propagated-inputs
 750      `(("r-s4vectors" ,r-s4vectors)
 751        ("r-iranges" ,r-iranges)
 752        ("r-genomicranges" ,r-genomicranges)
 753        ("r-summarizedexperiment" ,r-summarizedexperiment)
 754        ("r-rsamtools" ,r-rsamtools)
 755        ("r-genomicalignments" ,r-genomicalignments)
 756        ("r-rtracklayer" ,r-rtracklayer)))
 757     (home-page "http://bioconductor.org/packages/RIPSeeker")
 758     (synopsis
 759      "Identifying protein-associated transcripts from RIP-seq experiments")
 760     (description
 761      "This package infers and discriminates RIP peaks from RIP-seq alignments
 762 using two-state HMM with negative binomial emission probability.  While
 763 RIPSeeker is specifically tailored for RIP-seq data analysis, it also provides
 764 a suite of bioinformatics tools integrated within this self-contained software
 765 package comprehensively addressing issues ranging from post-alignments
 766 processing to visualization and annotation.")
 767     (license license:gpl2)))
 768
 769 (define-public r-multtest
 770   (package
 771     (name "r-multtest")
 772     (version "2.38.0")
 773     (source
 774      (origin
 775        (method url-fetch)
 776        (uri (bioconductor-uri "multtest" version))
 777        (sha256
 778         (base32
 779          "0lq62jw81hz9k840969n5byj57pwd0jqga3hqvhb6abb3g10yz7k"))))
 780     (build-system r-build-system)
 781     (propagated-inputs
 782      `(("r-survival" ,r-survival)
 783        ("r-biocgenerics" ,r-biocgenerics)
 784        ("r-biobase" ,r-biobase)
 785        ("r-mass" ,r-mass)))
 786     (home-page "http://bioconductor.org/packages/multtest")
 787     (synopsis "Resampling-based multiple hypothesis testing")
 788     (description
 789      "This package can do non-parametric bootstrap and permutation
 790 resampling-based multiple testing procedures (including empirical Bayes
 791 methods) for controlling the family-wise error rate (FWER), generalized
 792 family-wise error rate (gFWER), tail probability of the proportion of
 793 false positives (TPPFP), and false discovery rate (FDR).  Several choices
 794 of bootstrap-based null distribution are implemented (centered, centered
 795 and scaled, quantile-transformed).  Single-step and step-wise methods are
 796 available.  Tests based on a variety of T- and F-statistics (including
 797 T-statistics based on regression parameters from linear and survival models
 798 as well as those based on correlation parameters) are included.  When probing
 799 hypotheses with T-statistics, users may also select a potentially faster null
 800 distribution which is multivariate normal with mean zero and variance
 801 covariance matrix derived from the vector influence function.  Results are
 802 reported in terms of adjusted P-values, confidence regions and test statistic
 803 cutoffs.  The procedures are directly applicable to identifying differentially
 804 expressed genes in DNA microarray experiments.")
 805     (license license:lgpl3)))
 806
 807 (define-public r-graph
 808   (package
 809     (name "r-graph")
 810     (version "1.60.0")
 811     (source (origin
 812               (method url-fetch)
 813               (uri (bioconductor-uri "graph" version))
 814               (sha256
 815                (base32
 816                 "1kgnsm6f0vmb9qbkmmrnvxbwqc0gar17dq5gv1v10hrksw6mh64i"))))
 817     (build-system r-build-system)
 818     (propagated-inputs
 819      `(("r-biocgenerics" ,r-biocgenerics)))
 820     (home-page "https://bioconductor.org/packages/graph")
 821     (synopsis "Handle graph data structures in R")
 822     (description
 823      "This package implements some simple graph handling capabilities for R.")
 824     (license license:artistic2.0)))
 825
 826 (define-public r-codedepends
 827   (package
 828     (name "r-codedepends")
 829     (version "0.6.5")
 830     (source
 831      (origin
 832        (method url-fetch)
 833        (uri (cran-uri "CodeDepends" version))
 834        (sha256
 835         (base32
 836          "0l7kiv3awx50glf5cs841b4zzsff1ml90f0zr868ygvwsr4ps1hq"))))
 837     (properties `((upstream-name . "CodeDepends")))
 838     (build-system r-build-system)
 839     (propagated-inputs
 840      `(("r-codetools" ,r-codetools)
 841        ("r-graph" ,r-graph)
 842        ("r-xml" ,r-xml)))
 843     (home-page "http://cran.r-project.org/web/packages/CodeDepends")
 844     (synopsis "Analysis of R code for reproducible research and code comprehension")
 845     (description
 846      "This package provides tools for analyzing R expressions or blocks of
 847 code and determining the dependencies between them.  It focuses on R scripts,
 848 but can be used on the bodies of functions.  There are many facilities
 849 including the ability to summarize or get a high-level view of code,
 850 determining dependencies between variables, code improvement suggestions.")
 851     ;; Any version of the GPL
 852     (license (list license:gpl2+ license:gpl3+))))
 853
 854 (define-public r-chippeakanno
 855   (package
 856     (name "r-chippeakanno")
 857     (version "3.16.1")
 858     (source
 859      (origin
 860        (method url-fetch)
 861        (uri (bioconductor-uri "ChIPpeakAnno" version))
 862        (sha256
 863         (base32
 864          "1x98d8iwrxjwdz1s5cnvi6flynw9gdkmara9gwf205qxgmy7j3a3"))))
 865     (properties `((upstream-name . "ChIPpeakAnno")))
 866     (build-system r-build-system)
 867     (propagated-inputs
 868      `(("r-biocgenerics" ,r-biocgenerics)
 869        ("r-biocmanager" ,r-biocmanager)
 870        ("r-biostrings" ,r-biostrings)
 871        ("r-delayedarray" ,r-delayedarray)
 872        ("r-go-db" ,r-go-db)
 873        ("r-biomart" ,r-biomart)
 874        ("r-bsgenome" ,r-bsgenome)
 875        ("r-genomicfeatures" ,r-genomicfeatures)
 876        ("r-genomicranges" ,r-genomicranges)
 877        ("r-genomeinfodb" ,r-genomeinfodb)
 878        ("r-iranges" ,r-iranges)
 879        ("r-matrixstats" ,r-matrixstats)
 880        ("r-annotationdbi" ,r-annotationdbi)
 881        ("r-limma" ,r-limma)
 882        ("r-multtest" ,r-multtest)
 883        ("r-rbgl" ,r-rbgl)
 884        ("r-graph" ,r-graph)
 885        ("r-regioner" ,r-regioner)
 886        ("r-dbi" ,r-dbi)
 887        ("r-ensembldb" ,r-ensembldb)
 888        ("r-biobase" ,r-biobase)
 889        ("r-s4vectors" ,r-s4vectors)
 890        ("r-seqinr" ,r-seqinr)
 891        ("r-idr" ,r-idr)
 892        ("r-genomicalignments" ,r-genomicalignments)
 893        ("r-summarizedexperiment" ,r-summarizedexperiment)
 894        ("r-rsamtools" ,r-rsamtools)
 895        ("r-venndiagram" ,r-venndiagram)))
 896     (home-page "http://bioconductor.org/packages/ChIPpeakAnno")
 897     (synopsis "Peaks annotation from ChIP-seq and ChIP-chip experiments")
 898     (description
 899      "The package includes functions to retrieve the sequences around the peak,
 900 obtain enriched Gene Ontology (GO) terms, find the nearest gene, exon, miRNA or
 901 custom features such as most conserved elements and other transcription factor
 902 binding sites supplied by users.  Starting 2.0.5, new functions have been added
 903 for finding the peaks with bi-directional promoters with summary statistics
 904 (peaksNearBDP), for summarizing the occurrence of motifs in peaks
 905 (summarizePatternInPeaks) and for adding other IDs to annotated peaks or
 906 enrichedGO (addGeneIDs).")
 907     (license license:gpl2+)))
 908
 909 (define-public r-marray
 910   (package
 911     (name "r-marray")
 912     (version "1.60.0")
 913     (source (origin
 914               (method url-fetch)
 915               (uri (bioconductor-uri "marray" version))
 916               (sha256
 917                (base32 "1sh7l3c28x6zhdv99c9x05ii2yxmh9alkazp98kdi4fdb23rlzky"))))
 918     (build-system r-build-system)
 919     (propagated-inputs
 920      `(("r-limma" ,r-limma)))
 921     (home-page "http://bioconductor.org/packages/marray")
 922     (synopsis "Exploratory analysis for two-color spotted microarray data")
 923     (description "This package contains class definitions for two-color spotted
 924 microarray data.  It also includes fuctions for data input, diagnostic plots,
 925 normalization and quality checking.")
 926     (license license:lgpl2.0+)))
 927
 928 (define-public r-cghbase
 929   (package
 930    (name "r-cghbase")
 931    (version "1.42.0")
 932    (source (origin
 933             (method url-fetch)
 934             (uri (bioconductor-uri "CGHbase" version))
 935             (sha256
 936              (base32 "0ghxp49xdi09p3f2qwrdrq2p4qjafj4z1rr08ycgbf11gb22h1sc"))))
 937    (properties `((upstream-name . "CGHbase")))
 938    (build-system r-build-system)
 939    (propagated-inputs
 940     `(("r-biobase" ,r-biobase)
 941       ("r-marray" ,r-marray)))
 942    (home-page "http://bioconductor.org/packages/CGHbase")
 943    (synopsis "Base functions and classes for arrayCGH data analysis")
 944    (description "This package contains functions and classes that are needed by
 945 the @code{arrayCGH} packages.")
 946    (license license:gpl2+)))
 947
 948 (define-public r-cghcall
 949   (package
 950    (name "r-cghcall")
 951    (version "2.44.0")
 952    (source (origin
 953             (method url-fetch)
 954             (uri (bioconductor-uri "CGHcall" version))
 955             (sha256
 956              (base32 "1k65kaiqvjyllzbpa2367n6f6kkmsy463kpflzs66hqhx2fshsmi"))))
 957    (properties `((upstream-name . "CGHcall")))
 958    (build-system r-build-system)
 959    (propagated-inputs
 960     `(("r-biobase" ,r-biobase)
 961       ("r-cghbase" ,r-cghbase)
 962       ("r-impute" ,r-impute)
 963       ("r-dnacopy" ,r-dnacopy)
 964       ("r-snowfall" ,r-snowfall)))
 965    (home-page "http://bioconductor.org/packages/CGHcall")
 966    (synopsis "Base functions and classes for arrayCGH data analysis")
 967    (description "This package contains functions and classes that are needed by
 968 @code{arrayCGH} packages.")
 969    (license license:gpl2+)))
 970
 971 (define-public r-qdnaseq
 972   (package
 973     (name "r-qdnaseq")
 974     (version "1.18.0")
 975     (source (origin
 976               (method url-fetch)
 977               (uri (bioconductor-uri "QDNAseq" version))
 978               (sha256
 979                (base32 "04ff9nbckzrlb45mr2j0c3mlh1wcggq5bbl81zklhq203c5x1wc2"))))
 980     (properties `((upstream-name . "QDNAseq")))
 981     (build-system r-build-system)
 982     (propagated-inputs
 983      `(("r-biobase" ,r-biobase)
 984        ("r-biocparallel" ,r-biocparallel)
 985        ("r-cghbase" ,r-cghbase)
 986        ("r-cghcall" ,r-cghcall)
 987        ("r-dnacopy" ,r-dnacopy)
 988        ("r-genomicranges" ,r-genomicranges)
 989        ("r-iranges" ,r-iranges)
 990        ("r-matrixstats" ,r-matrixstats)
 991        ("r-r-utils" ,r-r-utils)
 992        ("r-rsamtools" ,r-rsamtools)))
 993     (home-page "http://bioconductor.org/packages/QDNAseq")
 994     (synopsis "Quantitative DNA sequencing for chromosomal aberrations")
 995     (description "The genome is divided into non-overlapping fixed-sized bins,
 996 number of sequence reads in each counted, adjusted with a simultaneous
 997 two-dimensional loess correction for sequence mappability and GC content, and
 998 filtered to remove spurious regions in the genome.  Downstream steps of
 999 segmentation and calling are also implemented via packages DNAcopy and CGHcall,
1000 respectively.")
1001     (license license:gpl2+)))
1002
1003 (define-public r-bayseq
1004   (package
1005     (name "r-bayseq")
1006     (version "2.16.0")
1007     (source
1008      (origin
1009        (method url-fetch)
1010        (uri (bioconductor-uri "baySeq" version))
1011        (sha256
1012         (base32
1013          "0f6yckihm5cwh3dycv2g54hf7nddhcqya4yrqwbir96y5k1d1km5"))))
1014     (properties `((upstream-name . "baySeq")))
1015     (build-system r-build-system)
1016     (propagated-inputs
1017      `(("r-abind" ,r-abind)
1018        ("r-edger" ,r-edger)
1019        ("r-genomicranges" ,r-genomicranges)))
1020     (home-page "https://bioconductor.org/packages/baySeq/")
1021     (synopsis "Bayesian analysis of differential expression patterns in count data")
1022     (description
1023      "This package identifies differential expression in high-throughput count
1024 data, such as that derived from next-generation sequencing machines,
1025 calculating estimated posterior likelihoods of differential expression (or
1026 more complex hypotheses) via empirical Bayesian methods.")
1027     (license license:gpl3)))
1028
1029 (define-public r-chipcomp
1030   (package
1031     (name "r-chipcomp")
1032     (version "1.12.0")
1033     (source
1034      (origin
1035        (method url-fetch)
1036        (uri (bioconductor-uri "ChIPComp" version))
1037        (sha256
1038         (base32
1039          "1sypdsvwzssraanlhddhzpf9p0xs3qlflc0hp7yfbp0aplsifx85"))))
1040     (properties `((upstream-name . "ChIPComp")))
1041     (build-system r-build-system)
1042     (propagated-inputs
1043      `(("r-biocgenerics" ,r-biocgenerics)
1044        ("r-bsgenome-hsapiens-ucsc-hg19" ,r-bsgenome-hsapiens-ucsc-hg19)
1045        ("r-bsgenome-mmusculus-ucsc-mm9" ,r-bsgenome-mmusculus-ucsc-mm9)
1046        ("r-genomeinfodb" ,r-genomeinfodb)
1047        ("r-genomicranges" ,r-genomicranges)
1048        ("r-iranges" ,r-iranges)
1049        ("r-limma" ,r-limma)
1050        ("r-rsamtools" ,r-rsamtools)
1051        ("r-rtracklayer" ,r-rtracklayer)
1052        ("r-s4vectors" ,r-s4vectors)))
1053     (home-page "https://bioconductor.org/packages/ChIPComp")
1054     (synopsis "Quantitative comparison of multiple ChIP-seq datasets")
1055     (description
1056      "ChIPComp implements a statistical method for quantitative comparison of
1057 multiple ChIP-seq datasets.  It detects differentially bound sharp binding
1058 sites across multiple conditions considering matching control in ChIP-seq
1059 datasets.")
1060     ;; Any version of the GPL.
1061     (license license:gpl3+)))
1062
1063 (define-public r-riboprofiling
1064   (package
1065     (name "r-riboprofiling")
1066     (version "1.12.0")
1067     (source
1068      (origin
1069        (method url-fetch)
1070        (uri (bioconductor-uri "RiboProfiling" version))
1071        (sha256
1072         (base32
1073          "1njvkd1khmf3rbp3dkz5z63wp79z4wmk4kzd3p3amky3w5by070z"))))
1074     (properties `((upstream-name . "RiboProfiling")))
1075     (build-system r-build-system)
1076     (propagated-inputs
1077      `(("r-biocgenerics" ,r-biocgenerics)
1078        ("r-biostrings" ,r-biostrings)
1079        ("r-data-table" ,r-data-table)
1080        ("r-genomeinfodb" ,r-genomeinfodb)
1081        ("r-genomicalignments" ,r-genomicalignments)
1082        ("r-genomicfeatures" ,r-genomicfeatures)
1083        ("r-genomicranges" ,r-genomicranges)
1084        ("r-ggbio" ,r-ggbio)
1085        ("r-ggplot2" ,r-ggplot2)
1086        ("r-iranges" ,r-iranges)
1087        ("r-plyr" ,r-plyr)
1088        ("r-reshape2" ,r-reshape2)
1089        ("r-rsamtools" ,r-rsamtools)
1090        ("r-rtracklayer" ,r-rtracklayer)
1091        ("r-s4vectors" ,r-s4vectors)
1092        ("r-sqldf" ,r-sqldf)))
1093     (home-page "https://bioconductor.org/packages/RiboProfiling/")
1094     (synopsis "Ribosome profiling data analysis")
1095     (description "Starting with a BAM file, this package provides the
1096 necessary functions for quality assessment, read start position recalibration,
1097 the counting of genomic sequence reads on CDS, 3'UTR, and 5'UTR, and plotting
1098 of count data: pairs, log fold-change, codon frequency and coverage
1099 assessment, principal component analysis on codon coverage.")
1100     (license license:gpl3)))
1101
1102 (define-public r-riboseqr
1103   (package
1104     (name "r-riboseqr")
1105     (version "1.16.0")
1106     (source
1107      (origin
1108        (method url-fetch)
1109        (uri (bioconductor-uri "riboSeqR" version))
1110        (sha256
1111         (base32
1112          "1nacsbsz77fw4a10nqj2ncsf25q3aaa0gd5w1q0ray2prx7qmlqb"))))
1113     (properties `((upstream-name . "riboSeqR")))
1114     (build-system r-build-system)
1115     (propagated-inputs
1116      `(("r-abind" ,r-abind)
1117        ("r-bayseq" ,r-bayseq)
1118        ("r-genomeinfodb" ,r-genomeinfodb)
1119        ("r-genomicranges" ,r-genomicranges)
1120        ("r-iranges" ,r-iranges)
1121        ("r-rsamtools" ,r-rsamtools)
1122        ("r-seqlogo" ,r-seqlogo)))
1123     (home-page "https://bioconductor.org/packages/riboSeqR/")
1124     (synopsis "Analysis of sequencing data from ribosome profiling experiments")
1125     (description
1126      "This package provides plotting functions, frameshift detection and
1127 parsing of genetic sequencing data from ribosome profiling experiments.")
1128     (license license:gpl3)))
1129
1130 (define-public r-interactionset
1131   (package
1132     (name "r-interactionset")
1133     (version "1.10.0")
1134     (source
1135      (origin
1136        (method url-fetch)
1137        (uri (bioconductor-uri "InteractionSet" version))
1138        (sha256
1139         (base32
1140          "0wirfhmpmkmnmhbqslw4bzvi2msqyqpjy1rrwr5qbd9k5rhx3bzb"))))
1141     (properties
1142      `((upstream-name . "InteractionSet")))
1143     (build-system r-build-system)
1144     (propagated-inputs
1145      `(("r-biocgenerics" ,r-biocgenerics)
1146        ("r-genomeinfodb" ,r-genomeinfodb)
1147        ("r-genomicranges" ,r-genomicranges)
1148        ("r-iranges" ,r-iranges)
1149        ("r-matrix" ,r-matrix)
1150        ("r-rcpp" ,r-rcpp)
1151        ("r-s4vectors" ,r-s4vectors)
1152        ("r-summarizedexperiment" ,r-summarizedexperiment)))
1153     (home-page "https://bioconductor.org/packages/InteractionSet")
1154     (synopsis "Base classes for storing genomic interaction data")
1155     (description
1156      "This packages provides the @code{GInteractions},
1157 @code{InteractionSet} and @code{ContactMatrix} objects and associated methods
1158 for storing and manipulating genomic interaction data from Hi-C and ChIA-PET
1159 experiments.")
1160     (license license:gpl3)))
1161
1162 (define-public r-genomicinteractions
1163   (package
1164     (name "r-genomicinteractions")
1165     (version "1.16.0")
1166     (source
1167      (origin
1168        (method url-fetch)
1169        (uri (bioconductor-uri "GenomicInteractions" version))
1170        (sha256
1171         (base32
1172          "0zy5isp2lqpjm0n0n1gly5bs4izn22yciibyqrnlrr60rmn5s67q"))))
1173     (properties
1174      `((upstream-name . "GenomicInteractions")))
1175     (build-system r-build-system)
1176     (propagated-inputs
1177      `(("r-biobase" ,r-biobase)
1178        ("r-biocgenerics" ,r-biocgenerics)
1179        ("r-data-table" ,r-data-table)
1180        ("r-dplyr" ,r-dplyr)
1181        ("r-genomeinfodb" ,r-genomeinfodb)
1182        ("r-genomicranges" ,r-genomicranges)
1183        ("r-ggplot2" ,r-ggplot2)
1184        ("r-gridextra" ,r-gridextra)
1185        ("r-gviz" ,r-gviz)
1186        ("r-igraph" ,r-igraph)
1187        ("r-interactionset" ,r-interactionset)
1188        ("r-iranges" ,r-iranges)
1189        ("r-rsamtools" ,r-rsamtools)
1190        ("r-rtracklayer" ,r-rtracklayer)
1191        ("r-s4vectors" ,r-s4vectors)
1192        ("r-stringr" ,r-stringr)))
1193     (home-page "https://github.com/ComputationalRegulatoryGenomicsICL/GenomicInteractions/")
1194     (synopsis "R package for handling genomic interaction data")
1195     (description
1196      "This R package provides tools for handling genomic interaction data,
1197 such as ChIA-PET/Hi-C, annotating genomic features with interaction
1198 information and producing various plots and statistics.")
1199     (license license:gpl3)))
1200
1201 (define-public r-ctc
1202   (package
1203     (name "r-ctc")
1204     (version "1.56.0")
1205     (source
1206      (origin
1207        (method url-fetch)
1208        (uri (bioconductor-uri "ctc" version))
1209        (sha256
1210         (base32
1211          "0yp7c0abk48jx1wf8n1gawh7dm15idahqc8va24v8cm0bzxgnmh2"))))
1212     (build-system r-build-system)
1213     (propagated-inputs `(("r-amap" ,r-amap)))
1214     (home-page "https://bioconductor.org/packages/ctc/")
1215     (synopsis "Cluster and tree conversion")
1216     (description
1217      "This package provides tools for exporting and importing classification
1218 trees and clusters to other programs.")
1219     (license license:gpl2)))
1220
1221 (define-public r-goseq
1222   (package
1223     (name "r-goseq")
1224     (version "1.34.1")
1225     (source
1226      (origin
1227        (method url-fetch)
1228        (uri (bioconductor-uri "goseq" version))
1229        (sha256
1230         (base32
1231          "1j87j98cajcjqabv6rb6zmcqxsqxxhbb3w60w1iink4rhsh8m3mn"))))
1232     (build-system r-build-system)
1233     (propagated-inputs
1234      `(("r-annotationdbi" ,r-annotationdbi)
1235        ("r-biasedurn" ,r-biasedurn)
1236        ("r-biocgenerics" ,r-biocgenerics)
1237        ("r-genelendatabase" ,r-genelendatabase)
1238        ("r-go-db" ,r-go-db)
1239        ("r-mgcv" ,r-mgcv)))
1240     (home-page "https://bioconductor.org/packages/goseq/")
1241     (synopsis "Gene Ontology analyser for RNA-seq and other length biased data")
1242     (description
1243      "This package provides tools to detect Gene Ontology and/or other user
1244 defined categories which are over/under represented in RNA-seq data.")
1245     (license license:lgpl2.0+)))
1246
1247 (define-public r-glimma
1248   (package
1249     (name "r-glimma")
1250     (version "1.10.1")
1251     (source
1252      (origin
1253        (method url-fetch)
1254        (uri (bioconductor-uri "Glimma" version))
1255        (sha256
1256         (base32
1257          "1ihrww55sa7ipi1rpp0rmn081sbqdwdmm5mz30zfrjr1xxqcdbcv"))))
1258     (properties `((upstream-name . "Glimma")))
1259     (build-system r-build-system)
1260     (propagated-inputs
1261      `(("r-edger" ,r-edger)
1262        ("r-jsonlite" ,r-jsonlite)
1263        ("r-s4vectors" ,r-s4vectors)))
1264     (home-page "https://github.com/Shians/Glimma")
1265     (synopsis "Interactive HTML graphics")
1266     (description
1267      "This package generates interactive visualisations for analysis of
1268 RNA-sequencing data using output from limma, edgeR or DESeq2 packages in an
1269 HTML page.  The interactions are built on top of the popular static
1270 representations of analysis results in order to provide additional
1271 information.")
1272     (license license:lgpl3)))
1273
1274 (define-public r-rots
1275   (package
1276     (name "r-rots")
1277     (version "1.10.1")
1278     (source
1279      (origin
1280        (method url-fetch)
1281        (uri (bioconductor-uri "ROTS" version))
1282        (sha256
1283         (base32
1284          "1d5ggkk47xybcaizfy756qimbf2falg9cld46mhqjp3xfbfvzsg6"))))
1285     (properties `((upstream-name . "ROTS")))
1286     (build-system r-build-system)
1287     (propagated-inputs
1288      `(("r-biobase" ,r-biobase)
1289        ("r-rcpp" ,r-rcpp)))
1290     (home-page "https://bioconductor.org/packages/ROTS/")
1291     (synopsis "Reproducibility-Optimized Test Statistic")
1292     (description
1293      "This package provides tools for calculating the
1294 @dfn{Reproducibility-Optimized Test Statistic} (ROTS) for differential testing
1295 in omics data.")
1296     (license license:gpl2+)))
1297
1298 (define-public r-plgem
1299   (package
1300     (name "r-plgem")
1301     (version "1.54.1")
1302     (source
1303      (origin
1304        (method url-fetch)
1305        (uri (bioconductor-uri "plgem" version))
1306        (sha256
1307         (base32
1308          "1330635db3p8xm5y8fwrk1l37r6bgypsq70s3rx954i775zp6szg"))))
1309     (build-system r-build-system)
1310     (propagated-inputs
1311      `(("r-biobase" ,r-biobase)
1312        ("r-mass" ,r-mass)))
1313     (home-page "http://www.genopolis.it")
1314     (synopsis "Detect differential expression in microarray and proteomics datasets")
1315     (description
1316      "The Power Law Global Error Model (PLGEM) has been shown to faithfully
1317 model the variance-versus-mean dependence that exists in a variety of
1318 genome-wide datasets, including microarray and proteomics data.  The use of
1319 PLGEM has been shown to improve the detection of differentially expressed
1320 genes or proteins in these datasets.")
1321     (license license:gpl2)))
1322
1323 (define-public r-inspect
1324   (package
1325     (name "r-inspect")
1326     (version "1.12.1")
1327     (source
1328      (origin
1329        (method url-fetch)
1330        (uri (bioconductor-uri "INSPEcT" version))
1331        (sha256
1332         (base32
1333          "07q5msw9rnamx957mbiawnv3p9kr5ahwawzvv9xzla7d3lkk62xp"))))
1334     (properties `((upstream-name . "INSPEcT")))
1335     (build-system r-build-system)
1336     (propagated-inputs
1337      `(("r-biobase" ,r-biobase)
1338        ("r-biocgenerics" ,r-biocgenerics)
1339        ("r-biocparallel" ,r-biocparallel)
1340        ("r-deseq2" ,r-deseq2)
1341        ("r-desolve" ,r-desolve)
1342        ("r-genomicalignments" ,r-genomicalignments)
1343        ("r-genomicfeatures" ,r-genomicfeatures)
1344        ("r-genomicranges" ,r-genomicranges)
1345        ("r-iranges" ,r-iranges)
1346        ("r-plgem" ,r-plgem)
1347        ("r-preprocesscore" ,r-preprocesscore)
1348        ("r-proc" ,r-proc)
1349        ("r-rootsolve" ,r-rootsolve)
1350        ("r-rsamtools" ,r-rsamtools)
1351        ("r-s4vectors" ,r-s4vectors)
1352        ("r-shiny" ,r-shiny)
1353        ("r-summarizedexperiment" ,r-summarizedexperiment)
1354        ("r-txdb-mmusculus-ucsc-mm9-knowngene"
1355         ,r-txdb-mmusculus-ucsc-mm9-knowngene)))
1356     (home-page "https://bioconductor.org/packages/INSPEcT")
1357     (synopsis "Analysis of 4sU-seq and RNA-seq time-course data")
1358     (description
1359      "INSPEcT (INference of Synthesis, Processing and dEgradation rates in
1360 Time-Course experiments) analyses 4sU-seq and RNA-seq time-course data in
1361 order to evaluate synthesis, processing and degradation rates and assess via
1362 modeling the rates that determines changes in mature mRNA levels.")
1363     (license license:gpl2)))
1364
1365 (define-public r-dnabarcodes
1366   (package
1367     (name "r-dnabarcodes")
1368     (version "1.12.0")
1369     (source
1370      (origin
1371        (method url-fetch)
1372        (uri (bioconductor-uri "DNABarcodes" version))
1373        (sha256
1374         (base32
1375          "0g6j7ish0fk9jcib94wssjgp1m8ldcp42hyyg1ypr945fa3xghx0"))))
1376     (properties `((upstream-name . "DNABarcodes")))
1377     (build-system r-build-system)
1378     (propagated-inputs
1379      `(("r-bh" ,r-bh)
1380        ("r-matrix" ,r-matrix)
1381        ("r-rcpp" ,r-rcpp)))
1382     (home-page "https://bioconductor.org/packages/DNABarcodes")
1383     (synopsis "Create and analyze DNA barcodes")
1384     (description
1385      "This package offers tools to create DNA barcode sets capable of
1386 correcting insertion, deletion, and substitution errors.  Existing barcodes
1387 can be analyzed regarding their minimal, maximal and average distances between
1388 barcodes.  Finally, reads that start with a (possibly mutated) barcode can be
1389 demultiplexed, i.e. assigned to their original reference barcode.")
1390     (license license:gpl2)))
1391
1392 (define-public r-ruvseq
1393   (package
1394     (name "r-ruvseq")
1395     (version "1.16.1")
1396     (source
1397      (origin
1398        (method url-fetch)
1399        (uri (bioconductor-uri "RUVSeq" version))
1400        (sha256
1401         (base32
1402          "0qk7q3ab7k133divfkp54zsmvsmb9p8r09pkh2caswrzrn8achzv"))))
1403     (properties `((upstream-name . "RUVSeq")))
1404     (build-system r-build-system)
1405     (propagated-inputs
1406      `(("r-biobase" ,r-biobase)
1407        ("r-edaseq" ,r-edaseq)
1408        ("r-edger" ,r-edger)
1409        ("r-mass" ,r-mass)))
1410     (home-page "https://github.com/drisso/RUVSeq")
1411     (synopsis "Remove unwanted variation from RNA-Seq data")
1412     (description
1413      "This package implements methods to @dfn{remove unwanted variation} (RUV)
1414 of Risso et al. (2014) for the normalization of RNA-Seq read counts between
1415 samples.")
1416     (license license:artistic2.0)))
1417
1418 (define-public r-biocneighbors
1419   (package
1420     (name "r-biocneighbors")
1421     (version "1.0.0")
1422     (source
1423      (origin
1424        (method url-fetch)
1425        (uri (bioconductor-uri "BiocNeighbors" version))
1426        (sha256
1427         (base32
1428          "1fsb96acidlxwf0h65xv7fnwdi58ckmq00gmwykrlawh88wgn1ll"))))
1429     (properties `((upstream-name . "BiocNeighbors")))
1430     (build-system r-build-system)
1431     (propagated-inputs
1432      `(("r-biocparallel" ,r-biocparallel)
1433        ("r-rcpp" ,r-rcpp)
1434        ("r-rcppannoy" ,r-rcppannoy)
1435        ("r-s4vectors" ,r-s4vectors)))
1436     (home-page "https://bioconductor.org/packages/BiocNeighbors")
1437     (synopsis "Nearest Neighbor Detection for Bioconductor packages")
1438     (description
1439      "This package implements exact and approximate methods for nearest
1440 neighbor detection, in a framework that allows them to be easily switched
1441 within Bioconductor packages or workflows.  The exact algorithm is implemented
1442 using pre-clustering with the k-means algorithm.  Functions are also provided
1443 to search for all neighbors within a given distance.  Parallelization is
1444 achieved for all methods using the BiocParallel framework.")
1445     (license license:gpl3)))
1446
1447 (define-public r-destiny
1448   (package
1449     (name "r-destiny")
1450     (version "2.12.0")
1451     (source
1452      (origin
1453        (method url-fetch)
1454        (uri (bioconductor-uri "destiny" version))
1455        (sha256
1456         (base32
1457          "1iay17mrhsfmpwl920rh1nip5b6ybva4h6bna0yld04paq5yva67"))))
1458     (build-system r-build-system)
1459     (propagated-inputs
1460      `(("r-biobase" ,r-biobase)
1461        ("r-biocgenerics" ,r-biocgenerics)
1462        ("r-fnn" ,r-fnn)
1463        ("r-ggthemes" ,r-ggthemes)
1464        ("r-hmisc" ,r-hmisc)
1465        ("r-igraph" ,r-igraph)
1466        ("r-matrix" ,r-matrix)
1467        ("r-proxy" ,r-proxy)
1468        ("r-rcpp" ,r-rcpp)
1469        ("r-rcppeigen" ,r-rcppeigen)
1470        ("r-scales" ,r-scales)
1471        ("r-scatterplot3d" ,r-scatterplot3d)
1472        ("r-smoother" ,r-smoother)
1473        ("r-summarizedexperiment" ,r-summarizedexperiment)
1474        ("r-vim" ,r-vim)))
1475     (home-page "https://bioconductor.org/packages/destiny/")
1476     (synopsis "Create and plot diffusion maps")
1477     (description "This package provides tools to create and plot diffusion
1478 maps.")
1479     ;; Any version of the GPL
1480     (license license:gpl3+)))
1481
1482 (define-public r-savr
1483   (package
1484     (name "r-savr")
1485     (version "1.20.0")
1486     (source
1487      (origin
1488        (method url-fetch)
1489        (uri (bioconductor-uri "savR" version))
1490        (sha256
1491         (base32
1492          "13bwq2a2pygdkmhrcmvz525wsi5i01j911711zgs6x93wj20b2w7"))))
1493     (properties `((upstream-name . "savR")))
1494     (build-system r-build-system)
1495     (propagated-inputs
1496      `(("r-ggplot2" ,r-ggplot2)
1497        ("r-gridextra" ,r-gridextra)
1498        ("r-reshape2" ,r-reshape2)
1499        ("r-scales" ,r-scales)
1500        ("r-xml" ,r-xml)))
1501     (home-page "https://github.com/bcalder/savR")
1502     (synopsis "Parse and analyze Illumina SAV files")
1503     (description
1504      "This package provides tools to parse Illumina Sequence Analysis
1505 Viewer (SAV) files, access data, and generate QC plots.")
1506     (license license:agpl3+)))
1507
1508 (define-public r-chipexoqual
1509   (package
1510     (name "r-chipexoqual")
1511     (version "1.6.0")
1512     (source
1513      (origin
1514        (method url-fetch)
1515        (uri (bioconductor-uri "ChIPexoQual" version))
1516        (sha256
1517         (base32
1518          "1773bpiybn4g9jlv46z29x19q4dpcvn7lairr3lq5pdqbqmz5hnp"))))
1519     (properties `((upstream-name . "ChIPexoQual")))
1520     (build-system r-build-system)
1521     (propagated-inputs
1522      `(("r-biocparallel" ,r-biocparallel)
1523        ("r-biovizbase" ,r-biovizbase)
1524        ("r-broom" ,r-broom)
1525        ("r-data-table" ,r-data-table)
1526        ("r-dplyr" ,r-dplyr)
1527        ("r-genomeinfodb" ,r-genomeinfodb)
1528        ("r-genomicalignments" ,r-genomicalignments)
1529        ("r-genomicranges" ,r-genomicranges)
1530        ("r-ggplot2" ,r-ggplot2)
1531        ("r-hexbin" ,r-hexbin)
1532        ("r-iranges" ,r-iranges)
1533        ("r-rcolorbrewer" ,r-rcolorbrewer)
1534        ("r-rmarkdown" ,r-rmarkdown)
1535        ("r-rsamtools" ,r-rsamtools)
1536        ("r-s4vectors" ,r-s4vectors)
1537        ("r-scales" ,r-scales)
1538        ("r-viridis" ,r-viridis)))
1539     (home-page "https://github.com/keleslab/ChIPexoQual")
1540     (synopsis "Quality control pipeline for ChIP-exo/nexus data")
1541     (description
1542      "This package provides a quality control pipeline for ChIP-exo/nexus
1543 sequencing data.")
1544     (license license:gpl2+)))
1545
1546 (define-public r-copynumber
1547   (package
1548     (name "r-copynumber")
1549     (version "1.22.0")
1550     (source (origin
1551               (method url-fetch)
1552               (uri (bioconductor-uri "copynumber" version))
1553               (sha256
1554                (base32
1555                 "0ipwj9i5p1bwhg5d80jdjagm02krpj2v0j47qdgw41h8wncdyal3"))))
1556     (build-system r-build-system)
1557     (propagated-inputs
1558      `(("r-s4vectors" ,r-s4vectors)
1559        ("r-iranges" ,r-iranges)
1560        ("r-genomicranges" ,r-genomicranges)
1561        ("r-biocgenerics" ,r-biocgenerics)))
1562     (home-page "https://bioconductor.org/packages/copynumber")
1563     (synopsis "Segmentation of single- and multi-track copy number data")
1564     (description
1565      "This package segments single- and multi-track copy number data by a
1566 penalized least squares regression method.")
1567     (license license:artistic2.0)))
1568
1569 (define-public r-dnacopy
1570   (package
1571     (name "r-dnacopy")
1572     (version "1.56.0")
1573     (source
1574      (origin
1575        (method url-fetch)
1576        (uri (bioconductor-uri "DNAcopy" version))
1577        (sha256
1578         (base32
1579          "04cqdqxhva66xwh1s2vffi56b9fcrqd4slcrvqasj5lp2rkjli82"))))
1580     (properties `((upstream-name . "DNAcopy")))
1581     (build-system r-build-system)
1582     (native-inputs `(("gfortran" ,gfortran)))
1583     (home-page "https://bioconductor.org/packages/DNAcopy")
1584     (synopsis "DNA copy number data analysis")
1585     (description
1586      "This package implements the @dfn{circular binary segmentation} (CBS)
1587 algorithm to segment DNA copy number data and identify genomic regions with
1588 abnormal copy number.")
1589     (license license:gpl2+)))
1590
1591 ;; This is a CRAN package, but it uncharacteristically depends on a
1592 ;; Bioconductor package.
1593 (define-public r-htscluster
1594   (package
1595     (name "r-htscluster")
1596     (version "2.0.8")
1597     (source
1598      (origin
1599        (method url-fetch)
1600        (uri (cran-uri "HTSCluster" version))
1601        (sha256
1602         (base32
1603          "0wnbfh6hdx8692jilgmv8sys1zm6fqc6mim7vvjhyqlmpm8gm0kg"))))
1604     (properties `((upstream-name . "HTSCluster")))
1605     (build-system r-build-system)
1606     (propagated-inputs
1607      `(("r-capushe" ,r-capushe)
1608        ("r-edger" ,r-edger)
1609        ("r-plotrix" ,r-plotrix)))
1610     (home-page "https://cran.r-project.org/web/packages/HTSCluster")
1611     (synopsis "Clustering high-throughput transcriptome sequencing (HTS) data")
1612     (description
1613      "This package provides a Poisson mixture model is implemented to cluster
1614 genes from high-throughput transcriptome sequencing (RNA-seq) data.  Parameter
1615 estimation is performed using either the EM or CEM algorithm, and the slope
1616 heuristics are used for model selection (i.e., to choose the number of
1617 clusters).")
1618     (license license:gpl3+)))