wwood-bioruby 1.2.11

Sign up to get free protection for your applications and to get access to all the features.
Files changed (386) hide show
  1. data/README.rdoc +205 -0
  2. data/README_DEV.rdoc +285 -0
  3. data/VERSION.yml +4 -0
  4. data/bin/bioruby +44 -0
  5. data/bin/br_biofetch.rb +47 -0
  6. data/bin/br_bioflat.rb +293 -0
  7. data/bin/br_biogetseq.rb +45 -0
  8. data/bin/br_pmfetch.rb +421 -0
  9. data/lib/bio.rb +306 -0
  10. data/lib/bio/alignment.rb +2518 -0
  11. data/lib/bio/appl/bl2seq/report.rb +334 -0
  12. data/lib/bio/appl/blast.rb +505 -0
  13. data/lib/bio/appl/blast/ddbj.rb +142 -0
  14. data/lib/bio/appl/blast/format0.rb +1438 -0
  15. data/lib/bio/appl/blast/format8.rb +83 -0
  16. data/lib/bio/appl/blast/genomenet.rb +263 -0
  17. data/lib/bio/appl/blast/ncbioptions.rb +220 -0
  18. data/lib/bio/appl/blast/remote.rb +105 -0
  19. data/lib/bio/appl/blast/report.rb +767 -0
  20. data/lib/bio/appl/blast/rexml.rb +144 -0
  21. data/lib/bio/appl/blast/rpsblast.rb +277 -0
  22. data/lib/bio/appl/blast/wublast.rb +635 -0
  23. data/lib/bio/appl/blast/xmlparser.rb +236 -0
  24. data/lib/bio/appl/blat/report.rb +530 -0
  25. data/lib/bio/appl/clustalw.rb +219 -0
  26. data/lib/bio/appl/clustalw/report.rb +152 -0
  27. data/lib/bio/appl/emboss.rb +203 -0
  28. data/lib/bio/appl/fasta.rb +235 -0
  29. data/lib/bio/appl/fasta/format10.rb +325 -0
  30. data/lib/bio/appl/gcg/msf.rb +212 -0
  31. data/lib/bio/appl/gcg/seq.rb +195 -0
  32. data/lib/bio/appl/genscan/report.rb +552 -0
  33. data/lib/bio/appl/hmmer.rb +126 -0
  34. data/lib/bio/appl/hmmer/report.rb +683 -0
  35. data/lib/bio/appl/iprscan/report.rb +374 -0
  36. data/lib/bio/appl/mafft.rb +259 -0
  37. data/lib/bio/appl/mafft/report.rb +226 -0
  38. data/lib/bio/appl/muscle.rb +52 -0
  39. data/lib/bio/appl/paml/baseml.rb +95 -0
  40. data/lib/bio/appl/paml/baseml/report.rb +32 -0
  41. data/lib/bio/appl/paml/codeml.rb +242 -0
  42. data/lib/bio/appl/paml/codeml/rates.rb +67 -0
  43. data/lib/bio/appl/paml/codeml/report.rb +67 -0
  44. data/lib/bio/appl/paml/common.rb +348 -0
  45. data/lib/bio/appl/paml/common_report.rb +38 -0
  46. data/lib/bio/appl/paml/yn00.rb +103 -0
  47. data/lib/bio/appl/paml/yn00/report.rb +32 -0
  48. data/lib/bio/appl/phylip/alignment.rb +133 -0
  49. data/lib/bio/appl/phylip/distance_matrix.rb +96 -0
  50. data/lib/bio/appl/probcons.rb +41 -0
  51. data/lib/bio/appl/psort.rb +548 -0
  52. data/lib/bio/appl/psort/report.rb +542 -0
  53. data/lib/bio/appl/pts1.rb +263 -0
  54. data/lib/bio/appl/sim4.rb +124 -0
  55. data/lib/bio/appl/sim4/report.rb +485 -0
  56. data/lib/bio/appl/sosui/report.rb +151 -0
  57. data/lib/bio/appl/spidey/report.rb +593 -0
  58. data/lib/bio/appl/targetp/report.rb +267 -0
  59. data/lib/bio/appl/tcoffee.rb +55 -0
  60. data/lib/bio/appl/tmhmm/report.rb +231 -0
  61. data/lib/bio/command.rb +593 -0
  62. data/lib/bio/compat/features.rb +157 -0
  63. data/lib/bio/compat/references.rb +128 -0
  64. data/lib/bio/data/aa.rb +353 -0
  65. data/lib/bio/data/codontable.rb +722 -0
  66. data/lib/bio/data/na.rb +223 -0
  67. data/lib/bio/db.rb +329 -0
  68. data/lib/bio/db/aaindex.rb +357 -0
  69. data/lib/bio/db/biosql/biosql_to_biosequence.rb +67 -0
  70. data/lib/bio/db/biosql/sequence.rb +508 -0
  71. data/lib/bio/db/embl/common.rb +352 -0
  72. data/lib/bio/db/embl/embl.rb +500 -0
  73. data/lib/bio/db/embl/embl_to_biosequence.rb +85 -0
  74. data/lib/bio/db/embl/format_embl.rb +190 -0
  75. data/lib/bio/db/embl/sptr.rb +1283 -0
  76. data/lib/bio/db/embl/swissprot.rb +42 -0
  77. data/lib/bio/db/embl/trembl.rb +41 -0
  78. data/lib/bio/db/embl/uniprot.rb +42 -0
  79. data/lib/bio/db/fantom.rb +597 -0
  80. data/lib/bio/db/fasta.rb +410 -0
  81. data/lib/bio/db/fasta/defline.rb +532 -0
  82. data/lib/bio/db/fasta/fasta_to_biosequence.rb +63 -0
  83. data/lib/bio/db/fasta/format_fasta.rb +97 -0
  84. data/lib/bio/db/genbank/common.rb +307 -0
  85. data/lib/bio/db/genbank/ddbj.rb +22 -0
  86. data/lib/bio/db/genbank/format_genbank.rb +187 -0
  87. data/lib/bio/db/genbank/genbank.rb +250 -0
  88. data/lib/bio/db/genbank/genbank_to_biosequence.rb +86 -0
  89. data/lib/bio/db/genbank/genpept.rb +60 -0
  90. data/lib/bio/db/genbank/refseq.rb +18 -0
  91. data/lib/bio/db/gff.rb +1846 -0
  92. data/lib/bio/db/go.rb +481 -0
  93. data/lib/bio/db/kegg/brite.rb +41 -0
  94. data/lib/bio/db/kegg/compound.rb +131 -0
  95. data/lib/bio/db/kegg/drug.rb +98 -0
  96. data/lib/bio/db/kegg/enzyme.rb +148 -0
  97. data/lib/bio/db/kegg/expression.rb +155 -0
  98. data/lib/bio/db/kegg/genes.rb +263 -0
  99. data/lib/bio/db/kegg/genome.rb +241 -0
  100. data/lib/bio/db/kegg/glycan.rb +166 -0
  101. data/lib/bio/db/kegg/keggtab.rb +357 -0
  102. data/lib/bio/db/kegg/kgml.rb +256 -0
  103. data/lib/bio/db/kegg/orthology.rb +136 -0
  104. data/lib/bio/db/kegg/reaction.rb +82 -0
  105. data/lib/bio/db/kegg/taxonomy.rb +331 -0
  106. data/lib/bio/db/lasergene.rb +209 -0
  107. data/lib/bio/db/litdb.rb +107 -0
  108. data/lib/bio/db/medline.rb +326 -0
  109. data/lib/bio/db/nbrf.rb +191 -0
  110. data/lib/bio/db/newick.rb +658 -0
  111. data/lib/bio/db/nexus.rb +1854 -0
  112. data/lib/bio/db/pdb.rb +29 -0
  113. data/lib/bio/db/pdb/atom.rb +77 -0
  114. data/lib/bio/db/pdb/chain.rb +210 -0
  115. data/lib/bio/db/pdb/chemicalcomponent.rb +224 -0
  116. data/lib/bio/db/pdb/model.rb +148 -0
  117. data/lib/bio/db/pdb/pdb.rb +1911 -0
  118. data/lib/bio/db/pdb/residue.rb +176 -0
  119. data/lib/bio/db/pdb/utils.rb +399 -0
  120. data/lib/bio/db/prosite.rb +597 -0
  121. data/lib/bio/db/rebase.rb +456 -0
  122. data/lib/bio/db/soft.rb +404 -0
  123. data/lib/bio/db/transfac.rb +375 -0
  124. data/lib/bio/db/url.rb +42 -0
  125. data/lib/bio/feature.rb +139 -0
  126. data/lib/bio/io/biosql/biodatabase.rb +64 -0
  127. data/lib/bio/io/biosql/bioentry.rb +29 -0
  128. data/lib/bio/io/biosql/bioentry_dbxref.rb +11 -0
  129. data/lib/bio/io/biosql/bioentry_path.rb +12 -0
  130. data/lib/bio/io/biosql/bioentry_qualifier_value.rb +10 -0
  131. data/lib/bio/io/biosql/bioentry_reference.rb +10 -0
  132. data/lib/bio/io/biosql/bioentry_relationship.rb +10 -0
  133. data/lib/bio/io/biosql/biosequence.rb +11 -0
  134. data/lib/bio/io/biosql/comment.rb +7 -0
  135. data/lib/bio/io/biosql/config/database.yml +20 -0
  136. data/lib/bio/io/biosql/dbxref.rb +13 -0
  137. data/lib/bio/io/biosql/dbxref_qualifier_value.rb +12 -0
  138. data/lib/bio/io/biosql/location.rb +32 -0
  139. data/lib/bio/io/biosql/location_qualifier_value.rb +11 -0
  140. data/lib/bio/io/biosql/ontology.rb +10 -0
  141. data/lib/bio/io/biosql/reference.rb +9 -0
  142. data/lib/bio/io/biosql/seqfeature.rb +32 -0
  143. data/lib/bio/io/biosql/seqfeature_dbxref.rb +11 -0
  144. data/lib/bio/io/biosql/seqfeature_path.rb +11 -0
  145. data/lib/bio/io/biosql/seqfeature_qualifier_value.rb +20 -0
  146. data/lib/bio/io/biosql/seqfeature_relationship.rb +11 -0
  147. data/lib/bio/io/biosql/taxon.rb +12 -0
  148. data/lib/bio/io/biosql/taxon_name.rb +9 -0
  149. data/lib/bio/io/biosql/term.rb +27 -0
  150. data/lib/bio/io/biosql/term_dbxref.rb +11 -0
  151. data/lib/bio/io/biosql/term_path.rb +12 -0
  152. data/lib/bio/io/biosql/term_relationship.rb +13 -0
  153. data/lib/bio/io/biosql/term_relationship_term.rb +11 -0
  154. data/lib/bio/io/biosql/term_synonym.rb +10 -0
  155. data/lib/bio/io/das.rb +461 -0
  156. data/lib/bio/io/dbget.rb +194 -0
  157. data/lib/bio/io/ddbjxml.rb +638 -0
  158. data/lib/bio/io/ebisoap.rb +158 -0
  159. data/lib/bio/io/ensembl.rb +229 -0
  160. data/lib/bio/io/fastacmd.rb +163 -0
  161. data/lib/bio/io/fetch.rb +195 -0
  162. data/lib/bio/io/flatfile.rb +482 -0
  163. data/lib/bio/io/flatfile/autodetection.rb +545 -0
  164. data/lib/bio/io/flatfile/bdb.rb +253 -0
  165. data/lib/bio/io/flatfile/buffer.rb +237 -0
  166. data/lib/bio/io/flatfile/index.rb +1381 -0
  167. data/lib/bio/io/flatfile/indexer.rb +805 -0
  168. data/lib/bio/io/flatfile/splitter.rb +297 -0
  169. data/lib/bio/io/higet.rb +73 -0
  170. data/lib/bio/io/hinv.rb +442 -0
  171. data/lib/bio/io/keggapi.rb +805 -0
  172. data/lib/bio/io/ncbirest.rb +733 -0
  173. data/lib/bio/io/ncbisoap.rb +155 -0
  174. data/lib/bio/io/pubmed.rb +307 -0
  175. data/lib/bio/io/registry.rb +292 -0
  176. data/lib/bio/io/soapwsdl.rb +119 -0
  177. data/lib/bio/io/sql.rb +186 -0
  178. data/lib/bio/location.rb +867 -0
  179. data/lib/bio/map.rb +410 -0
  180. data/lib/bio/pathway.rb +960 -0
  181. data/lib/bio/reference.rb +602 -0
  182. data/lib/bio/sequence.rb +456 -0
  183. data/lib/bio/sequence/aa.rb +152 -0
  184. data/lib/bio/sequence/adapter.rb +108 -0
  185. data/lib/bio/sequence/common.rb +310 -0
  186. data/lib/bio/sequence/compat.rb +123 -0
  187. data/lib/bio/sequence/dblink.rb +54 -0
  188. data/lib/bio/sequence/format.rb +358 -0
  189. data/lib/bio/sequence/format_raw.rb +23 -0
  190. data/lib/bio/sequence/generic.rb +24 -0
  191. data/lib/bio/sequence/na.rb +491 -0
  192. data/lib/bio/shell.rb +44 -0
  193. data/lib/bio/shell/core.rb +578 -0
  194. data/lib/bio/shell/demo.rb +146 -0
  195. data/lib/bio/shell/interface.rb +218 -0
  196. data/lib/bio/shell/irb.rb +95 -0
  197. data/lib/bio/shell/object.rb +71 -0
  198. data/lib/bio/shell/plugin/blast.rb +42 -0
  199. data/lib/bio/shell/plugin/codon.rb +218 -0
  200. data/lib/bio/shell/plugin/das.rb +58 -0
  201. data/lib/bio/shell/plugin/emboss.rb +23 -0
  202. data/lib/bio/shell/plugin/entry.rb +105 -0
  203. data/lib/bio/shell/plugin/flatfile.rb +101 -0
  204. data/lib/bio/shell/plugin/keggapi.rb +181 -0
  205. data/lib/bio/shell/plugin/midi.rb +430 -0
  206. data/lib/bio/shell/plugin/obda.rb +45 -0
  207. data/lib/bio/shell/plugin/psort.rb +56 -0
  208. data/lib/bio/shell/plugin/seq.rb +247 -0
  209. data/lib/bio/shell/plugin/soap.rb +87 -0
  210. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/bioruby_generator.rb +29 -0
  211. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_classes.rhtml +4 -0
  212. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_log.rhtml +27 -0
  213. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_methods.rhtml +11 -0
  214. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_modules.rhtml +4 -0
  215. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_variables.rhtml +7 -0
  216. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-bg.gif +0 -0
  217. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-gem.png +0 -0
  218. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-link.gif +0 -0
  219. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.css +368 -0
  220. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.rhtml +47 -0
  221. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_controller.rb +144 -0
  222. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_helper.rb +47 -0
  223. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/commands.rhtml +8 -0
  224. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/history.rhtml +10 -0
  225. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/index.rhtml +26 -0
  226. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/spinner.gif +0 -0
  227. data/lib/bio/shell/script.rb +25 -0
  228. data/lib/bio/shell/setup.rb +109 -0
  229. data/lib/bio/shell/web.rb +102 -0
  230. data/lib/bio/tree.rb +852 -0
  231. data/lib/bio/util/color_scheme.rb +191 -0
  232. data/lib/bio/util/color_scheme/buried.rb +59 -0
  233. data/lib/bio/util/color_scheme/helix.rb +59 -0
  234. data/lib/bio/util/color_scheme/hydropathy.rb +64 -0
  235. data/lib/bio/util/color_scheme/nucleotide.rb +31 -0
  236. data/lib/bio/util/color_scheme/strand.rb +59 -0
  237. data/lib/bio/util/color_scheme/taylor.rb +50 -0
  238. data/lib/bio/util/color_scheme/turn.rb +59 -0
  239. data/lib/bio/util/color_scheme/zappo.rb +50 -0
  240. data/lib/bio/util/contingency_table.rb +370 -0
  241. data/lib/bio/util/restriction_enzyme.rb +228 -0
  242. data/lib/bio/util/restriction_enzyme/analysis.rb +249 -0
  243. data/lib/bio/util/restriction_enzyme/analysis_basic.rb +217 -0
  244. data/lib/bio/util/restriction_enzyme/cut_symbol.rb +107 -0
  245. data/lib/bio/util/restriction_enzyme/double_stranded.rb +321 -0
  246. data/lib/bio/util/restriction_enzyme/double_stranded/aligned_strands.rb +130 -0
  247. data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair.rb +103 -0
  248. data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair_in_enzyme_notation.rb +38 -0
  249. data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations.rb +76 -0
  250. data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations_in_enzyme_notation.rb +107 -0
  251. data/lib/bio/util/restriction_enzyme/enzymes.yaml +7061 -0
  252. data/lib/bio/util/restriction_enzyme/range/cut_range.rb +24 -0
  253. data/lib/bio/util/restriction_enzyme/range/cut_ranges.rb +47 -0
  254. data/lib/bio/util/restriction_enzyme/range/horizontal_cut_range.rb +67 -0
  255. data/lib/bio/util/restriction_enzyme/range/sequence_range.rb +257 -0
  256. data/lib/bio/util/restriction_enzyme/range/sequence_range/calculated_cuts.rb +242 -0
  257. data/lib/bio/util/restriction_enzyme/range/sequence_range/fragment.rb +51 -0
  258. data/lib/bio/util/restriction_enzyme/range/sequence_range/fragments.rb +41 -0
  259. data/lib/bio/util/restriction_enzyme/range/vertical_cut_range.rb +77 -0
  260. data/lib/bio/util/restriction_enzyme/single_strand.rb +200 -0
  261. data/lib/bio/util/restriction_enzyme/single_strand/cut_locations_in_enzyme_notation.rb +135 -0
  262. data/lib/bio/util/restriction_enzyme/single_strand_complement.rb +23 -0
  263. data/lib/bio/util/restriction_enzyme/string_formatting.rb +111 -0
  264. data/lib/bio/util/sirna.rb +288 -0
  265. data/test/data/HMMER/hmmpfam.out +64 -0
  266. data/test/data/HMMER/hmmsearch.out +88 -0
  267. data/test/data/SOSUI/sample.report +11 -0
  268. data/test/data/TMHMM/sample.report +21 -0
  269. data/test/data/aaindex/DAYM780301 +30 -0
  270. data/test/data/aaindex/PRAM900102 +20 -0
  271. data/test/data/bl2seq/cd8a_cd8b_blastp.bl2seq +53 -0
  272. data/test/data/bl2seq/cd8a_p53_e-5blastp.bl2seq +37 -0
  273. data/test/data/blast/2.2.15.blastp.m7 +876 -0
  274. data/test/data/blast/b0002.faa +15 -0
  275. data/test/data/blast/b0002.faa.m0 +128 -0
  276. data/test/data/blast/b0002.faa.m7 +65 -0
  277. data/test/data/blast/b0002.faa.m8 +1 -0
  278. data/test/data/blast/blastp-multi.m7 +188 -0
  279. data/test/data/command/echoarg2.bat +1 -0
  280. data/test/data/embl/AB090716.embl +65 -0
  281. data/test/data/embl/AB090716.embl.rel89 +63 -0
  282. data/test/data/fasta/example1.txt +75 -0
  283. data/test/data/fasta/example2.txt +21 -0
  284. data/test/data/genscan/sample.report +63 -0
  285. data/test/data/iprscan/merged.raw +32 -0
  286. data/test/data/iprscan/merged.txt +74 -0
  287. data/test/data/paml/codeml/control_file.txt +30 -0
  288. data/test/data/paml/codeml/output.txt +78 -0
  289. data/test/data/paml/codeml/rates +217 -0
  290. data/test/data/prosite/prosite.dat +2233 -0
  291. data/test/data/refseq/nm_126355.entret +64 -0
  292. data/test/data/rpsblast/misc.rpsblast +193 -0
  293. data/test/data/soft/GDS100_partial.soft +92 -0
  294. data/test/data/soft/GSE3457_family_partial.soft +874 -0
  295. data/test/data/uniprot/p53_human.uniprot +1456 -0
  296. data/test/functional/bio/appl/test_pts1.rb +115 -0
  297. data/test/functional/bio/io/test_ensembl.rb +229 -0
  298. data/test/functional/bio/io/test_soapwsdl.rb +52 -0
  299. data/test/functional/bio/sequence/test_output_embl.rb +51 -0
  300. data/test/functional/bio/test_command.rb +301 -0
  301. data/test/runner.rb +14 -0
  302. data/test/unit/bio/appl/bl2seq/test_report.rb +134 -0
  303. data/test/unit/bio/appl/blast/test_ncbioptions.rb +112 -0
  304. data/test/unit/bio/appl/blast/test_report.rb +1135 -0
  305. data/test/unit/bio/appl/blast/test_rpsblast.rb +398 -0
  306. data/test/unit/bio/appl/genscan/test_report.rb +182 -0
  307. data/test/unit/bio/appl/hmmer/test_report.rb +342 -0
  308. data/test/unit/bio/appl/iprscan/test_report.rb +338 -0
  309. data/test/unit/bio/appl/mafft/test_report.rb +63 -0
  310. data/test/unit/bio/appl/paml/codeml/test_rates.rb +45 -0
  311. data/test/unit/bio/appl/paml/codeml/test_report.rb +45 -0
  312. data/test/unit/bio/appl/paml/test_codeml.rb +174 -0
  313. data/test/unit/bio/appl/sosui/test_report.rb +81 -0
  314. data/test/unit/bio/appl/targetp/test_report.rb +146 -0
  315. data/test/unit/bio/appl/test_blast.rb +277 -0
  316. data/test/unit/bio/appl/test_fasta.rb +130 -0
  317. data/test/unit/bio/appl/test_psort.rb +57 -0
  318. data/test/unit/bio/appl/test_pts1.rb +77 -0
  319. data/test/unit/bio/appl/tmhmm/test_report.rb +126 -0
  320. data/test/unit/bio/data/test_aa.rb +90 -0
  321. data/test/unit/bio/data/test_codontable.rb +107 -0
  322. data/test/unit/bio/data/test_na.rb +80 -0
  323. data/test/unit/bio/db/embl/test_common.rb +117 -0
  324. data/test/unit/bio/db/embl/test_embl.rb +214 -0
  325. data/test/unit/bio/db/embl/test_embl_rel89.rb +219 -0
  326. data/test/unit/bio/db/embl/test_embl_to_bioseq.rb +203 -0
  327. data/test/unit/bio/db/embl/test_sptr.rb +1812 -0
  328. data/test/unit/bio/db/embl/test_uniprot.rb +31 -0
  329. data/test/unit/bio/db/kegg/test_genes.rb +45 -0
  330. data/test/unit/bio/db/pdb/test_pdb.rb +152 -0
  331. data/test/unit/bio/db/test_aaindex.rb +197 -0
  332. data/test/unit/bio/db/test_fasta.rb +250 -0
  333. data/test/unit/bio/db/test_gff.rb +1190 -0
  334. data/test/unit/bio/db/test_lasergene.rb +95 -0
  335. data/test/unit/bio/db/test_medline.rb +127 -0
  336. data/test/unit/bio/db/test_newick.rb +293 -0
  337. data/test/unit/bio/db/test_nexus.rb +364 -0
  338. data/test/unit/bio/db/test_prosite.rb +1437 -0
  339. data/test/unit/bio/db/test_rebase.rb +101 -0
  340. data/test/unit/bio/db/test_soft.rb +138 -0
  341. data/test/unit/bio/db/test_url.rb +36 -0
  342. data/test/unit/bio/io/flatfile/test_autodetection.rb +375 -0
  343. data/test/unit/bio/io/flatfile/test_buffer.rb +251 -0
  344. data/test/unit/bio/io/flatfile/test_splitter.rb +369 -0
  345. data/test/unit/bio/io/test_ddbjxml.rb +80 -0
  346. data/test/unit/bio/io/test_ensembl.rb +109 -0
  347. data/test/unit/bio/io/test_fastacmd.rb +42 -0
  348. data/test/unit/bio/io/test_flatfile.rb +505 -0
  349. data/test/unit/bio/io/test_soapwsdl.rb +32 -0
  350. data/test/unit/bio/sequence/test_aa.rb +115 -0
  351. data/test/unit/bio/sequence/test_common.rb +373 -0
  352. data/test/unit/bio/sequence/test_compat.rb +69 -0
  353. data/test/unit/bio/sequence/test_dblink.rb +58 -0
  354. data/test/unit/bio/sequence/test_na.rb +330 -0
  355. data/test/unit/bio/shell/plugin/test_seq.rb +185 -0
  356. data/test/unit/bio/test_alignment.rb +1025 -0
  357. data/test/unit/bio/test_command.rb +349 -0
  358. data/test/unit/bio/test_db.rb +96 -0
  359. data/test/unit/bio/test_feature.rb +144 -0
  360. data/test/unit/bio/test_location.rb +599 -0
  361. data/test/unit/bio/test_map.rb +230 -0
  362. data/test/unit/bio/test_pathway.rb +499 -0
  363. data/test/unit/bio/test_reference.rb +252 -0
  364. data/test/unit/bio/test_sequence.rb +329 -0
  365. data/test/unit/bio/test_shell.rb +18 -0
  366. data/test/unit/bio/test_tree.rb +593 -0
  367. data/test/unit/bio/util/restriction_enzyme/analysis/test_calculated_cuts.rb +299 -0
  368. data/test/unit/bio/util/restriction_enzyme/analysis/test_cut_ranges.rb +103 -0
  369. data/test/unit/bio/util/restriction_enzyme/analysis/test_sequence_range.rb +240 -0
  370. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_aligned_strands.rb +101 -0
  371. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair.rb +75 -0
  372. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair_in_enzyme_notation.rb +73 -0
  373. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations.rb +53 -0
  374. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations_in_enzyme_notation.rb +104 -0
  375. data/test/unit/bio/util/restriction_enzyme/single_strand/test_cut_locations_in_enzyme_notation.rb +83 -0
  376. data/test/unit/bio/util/restriction_enzyme/test_analysis.rb +246 -0
  377. data/test/unit/bio/util/restriction_enzyme/test_cut_symbol.rb +44 -0
  378. data/test/unit/bio/util/restriction_enzyme/test_double_stranded.rb +115 -0
  379. data/test/unit/bio/util/restriction_enzyme/test_single_strand.rb +147 -0
  380. data/test/unit/bio/util/restriction_enzyme/test_single_strand_complement.rb +147 -0
  381. data/test/unit/bio/util/restriction_enzyme/test_string_formatting.rb +60 -0
  382. data/test/unit/bio/util/test_color_scheme.rb +33 -0
  383. data/test/unit/bio/util/test_contingency_table.rb +94 -0
  384. data/test/unit/bio/util/test_restriction_enzyme.rb +42 -0
  385. data/test/unit/bio/util/test_sirna.rb +245 -0
  386. metadata +543 -0
@@ -0,0 +1,144 @@
1
+ #
2
+ # = bio/appl/blast/rexml.rb - BLAST XML output (-m 7) parser by REXML
3
+ #
4
+ # Copyright:: Copyright (C) 2002, 2003 Toshiaki Katayama <k@bioruby.org>
5
+ # License:: The Ruby License
6
+ #
7
+ # $Id:$
8
+ #
9
+ # == Note
10
+ #
11
+ # This file is automatically loaded by bio/appl/blast/report.rb
12
+ #
13
+
14
+ begin
15
+ require 'rexml/document'
16
+ rescue LoadError
17
+ end
18
+
19
+ module Bio
20
+ class Blast
21
+ class Report
22
+
23
+ private
24
+
25
+ def rexml_parse(xml)
26
+ dom = REXML::Document.new(xml)
27
+ rexml_parse_program(dom)
28
+ dom.elements.each("*//Iteration") do |e|
29
+ @iterations.push(rexml_parse_iteration(e))
30
+ end
31
+ end
32
+
33
+ def rexml_parse_program(dom)
34
+ hash = {}
35
+ dom.root.each_element_with_text do |e|
36
+ name, text = e.name, e.text
37
+ case name
38
+ when 'BlastOutput_param'
39
+ e.elements["Parameters"].each_element_with_text do |p|
40
+ xml_set_parameter(p.name, p.text)
41
+ end
42
+ else
43
+ hash[name] = text if text.strip.size > 0
44
+ end
45
+ end
46
+ @program = hash['BlastOutput_program']
47
+ @version = hash['BlastOutput_version']
48
+ @reference = hash['BlastOutput_reference']
49
+ @db = hash['BlastOutput_db']
50
+ @query_id = hash['BlastOutput_query-ID']
51
+ @query_def = hash['BlastOutput_query-def']
52
+ @query_len = hash['BlastOutput_query-len'].to_i
53
+ end
54
+
55
+ def rexml_parse_iteration(e)
56
+ iteration = Iteration.new
57
+ e.elements.each do |i|
58
+ case i.name
59
+ when 'Iteration_iter-num'
60
+ iteration.num = i.text.to_i
61
+ when 'Iteration_hits'
62
+ i.elements.each("Hit") do |h|
63
+ iteration.hits.push(rexml_parse_hit(h))
64
+ end
65
+ when 'Iteration_message'
66
+ iteration.message = i.text
67
+ when 'Iteration_stat'
68
+ i.elements["Statistics"].each_element_with_text do |s|
69
+ k = s.name.sub(/Statistics_/, '')
70
+ v = s.text =~ /\D/ ? s.text.to_f : s.text.to_i
71
+ iteration.statistics[k] = v
72
+ end
73
+
74
+ # for new BLAST XML format
75
+ when 'Iteration_query-ID'
76
+ iteration.query_id = i.text
77
+ when 'Iteration_query-def'
78
+ iteration.query_def = i.text
79
+ when 'Iteration_query-len'
80
+ iteration.query_len = i.text.to_i
81
+ end
82
+ end #case i.name
83
+
84
+ return iteration
85
+ end
86
+
87
+ def rexml_parse_hit(e)
88
+ hit = Hit.new
89
+ hash = {}
90
+ hit.query_id = @query_id
91
+ hit.query_def = @query_def
92
+ hit.query_len = @query_len
93
+ e.elements.each do |h|
94
+ case h.name
95
+ when 'Hit_hsps'
96
+ h.elements.each("Hsp") do |s|
97
+ hit.hsps.push(rexml_parse_hsp(s))
98
+ end
99
+ else
100
+ hash[h.name] = h.text
101
+ end
102
+ end
103
+ hit.num = hash['Hit_num'].to_i
104
+ hit.hit_id = hash['Hit_id']
105
+ hit.len = hash['Hit_len'].to_i
106
+ hit.definition = hash['Hit_def']
107
+ hit.accession = hash['Hit_accession']
108
+ return hit
109
+ end
110
+
111
+ def rexml_parse_hsp(e)
112
+ hsp = Hsp.new
113
+ hash = {}
114
+ e.each_element_with_text do |h|
115
+ hash[h.name] = h.text
116
+ end
117
+ hsp.num = hash['Hsp_num'].to_i
118
+ hsp.bit_score = hash['Hsp_bit-score'].to_f
119
+ hsp.score = hash['Hsp_score'].to_i
120
+ hsp.evalue = hash['Hsp_evalue'].to_f
121
+ hsp.query_from = hash['Hsp_query-from'].to_i
122
+ hsp.query_to = hash['Hsp_query-to'].to_i
123
+ hsp.hit_from = hash['Hsp_hit-from'].to_i
124
+ hsp.hit_to = hash['Hsp_hit-to'].to_i
125
+ hsp.pattern_from = hash['Hsp_pattern-from'].to_i
126
+ hsp.pattern_to = hash['Hsp_pattern-to'].to_i
127
+ hsp.query_frame = hash['Hsp_query-frame'].to_i
128
+ hsp.hit_frame = hash['Hsp_hit-frame'].to_i
129
+ hsp.identity = hash['Hsp_identity'].to_i
130
+ hsp.positive = hash['Hsp_positive'].to_i
131
+ hsp.gaps = hash['Hsp_gaps'].to_i
132
+ hsp.align_len = hash['Hsp_align-len'].to_i
133
+ hsp.density = hash['Hsp_density'].to_i
134
+ hsp.qseq = hash['Hsp_qseq']
135
+ hsp.hseq = hash['Hsp_hseq']
136
+ hsp.midline = hash['Hsp_midline']
137
+ return hsp
138
+ end
139
+
140
+ end
141
+ end
142
+ end
143
+
144
+
@@ -0,0 +1,277 @@
1
+ #
2
+ # = bio/appl/blast/rpsblast.rb - NCBI RPS Blast default output parser
3
+ #
4
+ # Copyright:: Copyright (C) 2008 Naohisa Goto <ng@bioruby.org>
5
+ # License:: The Ruby License
6
+ #
7
+ # $Id:$
8
+ #
9
+ # == Description
10
+ #
11
+ # NCBI RPS Blast (Reversed Position Specific Blast) default
12
+ # (-m 0 option) output parser class, Bio::Blast::RPSBlast::Report
13
+ # and related classes/modules.
14
+ #
15
+ # == References
16
+ #
17
+ # * Altschul, Stephen F., Thomas L. Madden, Alejandro A. Schaffer,
18
+ # Jinghui Zhang, Zheng Zhang, Webb Miller, and David J. Lipman (1997),
19
+ # "Gapped BLAST and PSI-BLAST: a new generation of protein database search
20
+ # programs", Nucleic Acids Res. 25:3389-3402.
21
+ # * ftp://ftp.ncbi.nih.gov/blast/documents/rpsblast.html
22
+ # * http://www.ncbi.nlm.nih.gov/Structure/cdd/cdd_help.shtml
23
+ #
24
+
25
+ require 'bio/appl/blast/format0'
26
+ require 'bio/io/flatfile'
27
+
28
+ module Bio
29
+ class Blast
30
+
31
+ # NCBI RPS Blast (Reversed Position Specific Blast) namespace.
32
+ # Currently, this module is existing only for separating namespace.
33
+ # To parse RPSBlast results, see Bio::Blast::RPSBlast::Report documents.
34
+ module RPSBlast
35
+
36
+ # Flatfile splitter for RPS-BLAST reports.
37
+ # It is internally used when reading RPS-BLAST report.
38
+ # Normally, users do not need to use it directly.
39
+ #
40
+ # Note for Windows: RPS-BLAST results generated in Microsoft Windows
41
+ # may not be parsed correctly due to the line feed code problem.
42
+ # For a workaroud, convert line feed codes from Windows(DOS) to UNIX.
43
+ #
44
+ class RPSBlastSplitter < Bio::FlatFile::Splitter::Template
45
+
46
+ # Separator used to distinguish start of each report
47
+ ReportHead = /\A\n*(RPS\-BLAST|Query\=)/
48
+
49
+ # Delimiter used for IO#gets
50
+ Delimiter = "\n\n"
51
+
52
+ # creates a new splitter object
53
+ def initialize(klass, bstream)
54
+ super(klass, bstream)
55
+ @entry_head = nil
56
+ end
57
+
58
+ # Skips leader of the entry.
59
+ # In this class, only skips space characters.
60
+ def skip_leader
61
+ stream.skip_spaces
62
+ return nil
63
+ end
64
+
65
+ # Rewinds the stream
66
+ def rewind
67
+ @entry_head = nil
68
+ super
69
+ end
70
+
71
+ # gets an entry
72
+ def get_entry
73
+ p0 = stream_pos()
74
+ pieces = []
75
+ flag_head = false # reached to start of header
76
+ flag_body = false # reached to start of body (Query=...)
77
+ while x = stream.gets(Delimiter)
78
+ if ReportHead =~ x then
79
+ case $1
80
+ when 'RPS-BLAST'
81
+ if pieces.empty? then
82
+ @entry_head = nil
83
+ flag_head = true
84
+ else
85
+ stream.ungets(x)
86
+ break
87
+ end
88
+ when 'Query='
89
+ if flag_body then
90
+ stream.ungets(x)
91
+ break
92
+ else
93
+ @entry_head = pieces.join('') if flag_head
94
+ flag_body = true
95
+ end
96
+ else
97
+ raise 'Bug: should not reach here'
98
+ end
99
+ end #if ReportHead...
100
+ pieces.push x
101
+ end #while
102
+ p1 = stream_pos()
103
+
104
+ self.entry_start_pos = p0
105
+ self.entry =
106
+ if pieces.empty? then
107
+ nil
108
+ elsif !flag_head and @entry_head then
109
+ @entry_head + pieces.join('')
110
+ else
111
+ pieces.join('')
112
+ end
113
+ self.entry_ended_pos = p1
114
+ return self.entry
115
+ end
116
+ end #class RPSBlastSplitter
117
+
118
+ # NCBI RPS Blast (Reversed Position Specific Blast)
119
+ # default output parser.
120
+ #
121
+ # It supports defalut (-m 0 option) output of the "rpsblast" command.
122
+ #
123
+ # Because this class inherits Bio::Blast::Default::Report,
124
+ # almost all methods are eqaul to Bio::Blast::Default::Report.
125
+ # Only DELIMITER (and RS) and few methods are different.
126
+ #
127
+ # By using Bio::FlatFile, (for example, Bio::FlatFile.open),
128
+ # rpsblast result generated from multiple query sequences is
129
+ # automatically splitted into multiple
130
+ # Bio::BLast::RPSBlast::Report objects corresponding to
131
+ # query sequences.
132
+ #
133
+ # Note for multi-fasta results WITH using Bio::FlatFile:
134
+ # Each splitted result is concatenated with header of the
135
+ # result which describes RPS-BLAST version and database
136
+ # information, if possible.
137
+ #
138
+ # Note for multi-fasta results WITHOUT using Bio::FlatFile:
139
+ # When parsing an output of rpsblast command running with
140
+ # multi-fasta sequences WITHOUT using Bio::FlatFile,
141
+ # each query's result is stored as an "iteration" of PSI-Blast.
142
+ # This behavior may be changed in the future.
143
+ #
144
+ # Note for nucleotide results: This class is not tested with
145
+ # nucleotide query and/or nucleotide databases.
146
+ #
147
+ class Report < Bio::Blast::Default::Report
148
+ # Delimter of each entry for RPS-BLAST.
149
+ DELIMITER = RS = "\nRPS-BLAST"
150
+
151
+ # (Integer) excess read size included in DELIMITER.
152
+ DELIMITER_OVERRUN = 9 # "RPS-BLAST"
153
+
154
+ # splitter for Bio::FlatFile support
155
+ FLATFILE_SPLITTER = RPSBlastSplitter
156
+
157
+ # Creates a new Report object from a string.
158
+ #
159
+ # Using Bio::FlatFile.open (or some other methods)
160
+ # is recommended instead of using this method directly.
161
+ # Refer Bio::Blast::RPSBlast::Report document for more information.
162
+ #
163
+ # Note for multi-fasta results WITHOUT using Bio::FlatFile:
164
+ # When parsing an output of rpsblast command running with
165
+ # multi-fasta sequences WITHOUT using Bio::FlatFile,
166
+ # each query's result is stored as an "iteration" of PSI-Blast.
167
+ # This behavior may be changed in the future.
168
+ #
169
+ # Note for nucleotide results: This class is not tested with
170
+ # nucleotide query and/or nucleotide databases.
171
+ #
172
+ def initialize(str)
173
+ str = str.sub(/\A\s+/, '')
174
+ # remove trailing entries for sure
175
+ str.sub!(/\n(RPS\-BLAST.*)/m, "\n")
176
+ @entry_overrun = $1
177
+ @entry = str
178
+ data = str.split(/(?:^[ \t]*\n)+/)
179
+
180
+ if data[0] and /\AQuery\=/ !~ data[0] then
181
+ format0_split_headers(data)
182
+ end
183
+ @iterations = format0_split_search(data)
184
+ format0_split_stat_params(data)
185
+ end
186
+
187
+ # Returns definition of the query.
188
+ # For a result of multi-fasta input, the first query's definition
189
+ # is returned (The same as <tt>iterations.first.query_def</tt>).
190
+ def query_def
191
+ iterations.first.query_def
192
+ end
193
+
194
+ # Returns length of the query.
195
+ # For a result of multi-fasta input, the first query's length
196
+ # is returned (The same as <tt>iterations.first.query_len</tt>).
197
+ def query_len
198
+ iterations.first.query_len
199
+ end
200
+
201
+ private
202
+
203
+ # Splits headers into the first line, reference, query line and
204
+ # database line.
205
+ def format0_split_headers(data)
206
+ @f0header = data.shift
207
+ @f0references = []
208
+ while data[0] and /\ADatabase\:/ !~ data[0]
209
+ @f0references.push data.shift
210
+ end
211
+ @f0database = data.shift
212
+ # In special case, a void line is inserted after database name.
213
+ if /\A +[\d\,]+ +sequences\; +[\d\,]+ total +letters\s*\z/ =~ data[0] then
214
+ @f0database.concat "\n"
215
+ @f0database.concat data.shift
216
+ end
217
+ end
218
+
219
+ # Splits the search results.
220
+ def format0_split_search(data)
221
+ iterations = []
222
+ dummystr = 'Searching..................................................done'
223
+ if r = data[0] and /^Searching/ =~ r then
224
+ dummystr = data.shift
225
+ end
226
+ while r = data[0] and /^Query\=/ =~ r
227
+ iterations << Iteration.new(data, dummystr)
228
+ end
229
+ iterations
230
+ end
231
+
232
+ # Iteration class for RPS-Blast.
233
+ # Though RPS-Blast does not iterate like PSI-BLAST,
234
+ # it aims to store a result of single query sequence.
235
+ #
236
+ # Normally, the instance of the class is generated
237
+ # by Bio::Blast::RPSBlast::Report object.
238
+ #
239
+ class Iteration < Bio::Blast::Default::Report::Iteration
240
+ # Creates a new Iteration object.
241
+ # It is designed to be called only internally from
242
+ # the Bio::Blast::RPSBlast::Report class.
243
+ # Users shall not use the method directly.
244
+ def initialize(data, dummystr)
245
+ if /\AQuery\=/ =~ data[0] then
246
+ sc = StringScanner.new(data.shift)
247
+ sc.skip(/\s*/)
248
+ if sc.skip_until(/Query\= */) then
249
+ q = []
250
+ begin
251
+ q << sc.scan(/.*/)
252
+ sc.skip(/\s*^ ?/)
253
+ end until !sc.rest or r = sc.skip(/ *\( *([\,\d]+) *letters *\)\s*\z/)
254
+ @query_len = sc[1].delete(',').to_i if r
255
+ @query_def = q.join(' ')
256
+ end
257
+ end
258
+ data.unshift(dummystr)
259
+
260
+ super(data)
261
+ end
262
+
263
+ # definition of the query
264
+ attr_reader :query_def
265
+
266
+ # length of the query sequence
267
+ attr_reader :query_len
268
+
269
+ end #class Iteration
270
+
271
+ end #class Report
272
+
273
+ end #module RPSBlast
274
+
275
+ end #module Blast
276
+ end #module Bio
277
+
@@ -0,0 +1,635 @@
1
+ #
2
+ # = bio/appl/blast/wublast.rb - WU-BLAST default output parser
3
+ #
4
+ # Copyright:: Copyright (C) 2003, 2008 Naohisa GOTO <ng@bioruby.org>
5
+ # License:: The Ruby License
6
+ #
7
+ # $Id:$
8
+ #
9
+ # == Description
10
+ #
11
+ # WU-BLAST default output parser.
12
+ #
13
+ # The parser is still incomplete and may contain many bugs,
14
+ # because I didn't have WU-BLAST license.
15
+ # It was tested under web-based WU-BLAST results and
16
+ # obsolete version downloaded from http://blast.wustl.edu/ .
17
+ #
18
+ # == References
19
+ #
20
+ # * http://blast.wustl.edu/
21
+ # * http://www.ebi.ac.uk/blast2/
22
+ #
23
+
24
+ require 'bio/appl/blast/format0'
25
+
26
+ module Bio
27
+ class Blast
28
+ module WU #:nodoc:
29
+
30
+ # Bio::Blast::WU::Report parses WU-BLAST default output
31
+ # and stores information in the data.
32
+ # It may contain a Bio::Blast::WU::Report::Iteration object.
33
+ # Because it inherits Bio::Blast::Default::Report,
34
+ # please also refer Bio::Blast::Default::Report.
35
+ class Report < Default::Report
36
+
37
+ # Returns parameters (???)
38
+ def parameters
39
+ parse_parameters
40
+ @parameters
41
+ end
42
+
43
+ # Returns parameter matrix (???)
44
+ def parameter_matrix
45
+ parse_parameters
46
+ @parameter_matrix
47
+ end
48
+
49
+ # Returns e-value threshold specified when BLAST was executed.
50
+ def expect; parse_parameters; @parameters['E']; end
51
+
52
+ # Returns warning messages.
53
+ def warnings
54
+ unless defined?(@warnings)
55
+ @warnings = @f0warnings
56
+ iterations.each { |x| @warnings.concat(x.warnings) }
57
+ end
58
+ @warnings
59
+ end
60
+
61
+ # Returns notice messages.
62
+ def notice
63
+ unless defined?(@notice)
64
+ @notice = @f0notice.to_s.gsub(/\s+/, ' ').strip
65
+ end #unless
66
+ @notice
67
+ end
68
+
69
+ # (WU-BLAST) Returns record number of the query.
70
+ # It may only be available for reports with multiple queries.
71
+ # Returns an Integer or nil.
72
+ def query_record_number
73
+ format0_parse_query
74
+ @query_record_number
75
+ end
76
+
77
+ # (WU-BLAST) Returns exit code for the execution.
78
+ # Returns an Integer or nil.
79
+ def exit_code
80
+ if defined? @exit_code then
81
+ @exit_code
82
+ else
83
+ nil
84
+ end
85
+ end
86
+
87
+ # (WU-BLAST) Returns fatal error information.
88
+ # Returns nil or an array containing String.
89
+ def fatal_errors
90
+ if defined? @fatal_errors then
91
+ @fatal_errors
92
+ else
93
+ nil
94
+ end
95
+ end
96
+
97
+ # Returns the name (filename or title) of the database.
98
+ def db
99
+ unless defined?(@db)
100
+ if /Database *\: *(.*)/m =~ @f0database then
101
+ a = $1.split(/^/)
102
+ if a.size > 1 and /\ASearching\..+ done\s*\z/ =~ a[-1] then
103
+ a.pop
104
+ end
105
+ if a.size > 1 and /\A +[\d\,]+ +sequences\; +[\d\,]+ total +letters\.?\s*\z/ =~ a[-1] then
106
+ a.pop
107
+ end
108
+ @db = a.collect { |x| x.sub(/\s+\z/, '') }.join(' ')
109
+ end
110
+ end #unless
111
+ @db
112
+ end
113
+
114
+ private
115
+ # Parses the query lines (begins with "Query = ").
116
+ def format0_parse_query
117
+ unless defined?(@query_def)
118
+ sc = StringScanner.new(@f0query)
119
+ sc.skip(/\s*/)
120
+ if sc.skip_until(/Query\= */) then
121
+ q = []
122
+ begin
123
+ q << sc.scan(/.*/)
124
+ sc.skip(/\s*^ ?/)
125
+ end until !sc.rest or r = sc.skip(/ *\( *([\,\d]+) *letters *(\; *record *([\,\d]+) *)?\)\s*\z/)
126
+ @query_len = sc[1].delete(',').to_i if r
127
+ @query_record_number = sc[3].delete(',').to_i if r and sc[2]
128
+ @query_def = q.join(' ')
129
+ end
130
+ end
131
+ end
132
+
133
+ # Splits headers.
134
+ def format0_split_headers(data)
135
+ @f0header = data.shift
136
+ @f0references = []
137
+ while r = data.first
138
+ case r
139
+ when /^Reference\: /
140
+ @f0references.push data.shift
141
+ when /^Copyright /
142
+ @f0copyright = data.shift
143
+ when /^Notice\: /
144
+ @f0notice = data.shift
145
+ when /^Query\= /
146
+ break
147
+ else
148
+ break
149
+ end
150
+ end
151
+ @f0query = data.shift
152
+ @f0warnings ||= []
153
+ while r = data.first and r =~ /^WARNING\: /
154
+ @f0warnings << data.shift
155
+ end
156
+ return if r = data.first and /\AParameters\:/ =~ r
157
+ if r = data.first and !(/^Database\: / =~ r)
158
+ @f0translate_info = data.shift
159
+ end
160
+ @f0database = data.shift
161
+ end
162
+
163
+ # Splits search data.
164
+ def format0_split_search(data)
165
+ @f0warnings ||= []
166
+ while r = data.first and r =~ /^WARNING\: /
167
+ @f0warnings << data.shift
168
+ end
169
+ [ Iteration.new(data) ]
170
+ end
171
+
172
+ # Splits statistics parameters.
173
+ def format0_split_stat_params(data)
174
+ @f0warnings ||= []
175
+ while r = data.first and r =~ /^WARNING\: /
176
+ @f0warnings << data.shift
177
+ end
178
+ @f0wu_params = []
179
+ @f0wu_stats = []
180
+ ary = @f0wu_params
181
+ while r = data.shift
182
+ case r
183
+ when /\AStatistics\:/
184
+ ary = @f0wu_stats
185
+ when /\AEXIT CODE *(\d+)\s*$/
186
+ @exit_code = $1.to_i
187
+ r = nil
188
+ when /\AFATAL\: /
189
+ @fatal_errors ||= []
190
+ @fatal_errors.push r
191
+ r = nil
192
+ when /\AWARNING\: /
193
+ @f0warnings ||= []
194
+ @f0warnings << r
195
+ r = nil
196
+ end
197
+ ary << r if r
198
+ end
199
+ @f0dbstat = F0dbstat.new(@f0wu_stats)
200
+ itr = @iterations[0]
201
+ x = @f0dbstat
202
+ itr.instance_eval { @f0dbstat = x } if itr
203
+ end
204
+
205
+ # Splits parameters.
206
+ def parse_parameters
207
+ unless defined?(@parse_parameters)
208
+ @parameters = {}
209
+ @parameter_matrix = []
210
+ @f0wu_params.each do |x|
211
+ if /^ Query/ =~ x then
212
+ @parameter_matrix << x
213
+ else
214
+ x.split(/^/).each do |y|
215
+ if /\A\s*(.+)\s*\=\s*(.*)\s*/ =~ y then
216
+ @parameters[$1] = $2
217
+ elsif /\AParameters\:/ =~ y then
218
+ ; #ignore this
219
+ elsif /\A\s*(.+)\s*$/ =~ y then
220
+ @parameters[$1] = true
221
+ end
222
+ end
223
+ end
224
+ end
225
+ if ev = @parameters['E'] then
226
+ ev = '1' + ev if ev[0] == ?e
227
+ @parameters['E'] = ev.to_f
228
+ end
229
+ @parse_parameters = true
230
+ end
231
+ end
232
+
233
+ # Stores database statistics.
234
+ # Internal use only. Users must not use the class.
235
+ class F0dbstat < Default::Report::F0dbstat #:nodoc:
236
+ def initialize(ary)
237
+ @f0stat = ary
238
+ @hash = {}
239
+ end
240
+
241
+ #--
242
+ #undef :f0params
243
+ #undef :matrix, :gap_open, :gap_extend,
244
+ # :eff_space, :expect, :sc_match, :sc_mismatch,
245
+ # :num_hits
246
+ #++
247
+
248
+ # Parses database statistics.
249
+ def parse_dbstat
250
+ unless defined?(@parse_dbstat)
251
+ parse_colon_separated_params(@hash, @f0stat)
252
+ @database = @hash['Database']
253
+ @posted_date = @hash['Posted']
254
+ if val = @hash['# of letters in database'] then
255
+ @db_len = val.tr(',', '').to_i
256
+ end
257
+ if val = @hash['# of sequences in database'] then
258
+ @db_num = val.tr(',', '').to_i
259
+ end
260
+ @parse_dbstat = true
261
+ end #unless
262
+ end #def
263
+ private :parse_dbstat
264
+
265
+ end #class F0dbstat
266
+
267
+ #--
268
+ #class Frame
269
+ #end #class FrameParams
270
+ #++
271
+
272
+ # Iteration class for WU-BLAST report.
273
+ # Though WU-BLAST does not iterate like PSI-BLAST,
274
+ # Bio::Blast::WU::Report::Iteration aims to keep compatibility
275
+ # with Bio::Blast::Default::Report::* classes.
276
+ # It may contain some Bio::Blast::WU::Report::Hit objects.
277
+ # Because it inherits Bio::Blast::Default::Report::Iteration,
278
+ # please also refer Bio::Blast::Default::Report::Iteration.
279
+ class Iteration < Default::Report::Iteration
280
+ # Creates a new Iteration object.
281
+ # It is designed to be called only internally from
282
+ # the Bio::Blast::WU::Report class.
283
+ # Users shall not use the method directly.
284
+ def initialize(data)
285
+ @f0stat = []
286
+ @f0dbstat = Default::Report::AlwaysNil.instance
287
+ @f0hitlist = []
288
+ @hits = []
289
+ @num = 1
290
+ @f0message = []
291
+ @f0warnings = []
292
+ return unless r = data.first
293
+ return if /\AParameters\:$/ =~ r
294
+ @f0hitlist << data.shift
295
+ return unless r = data.shift
296
+ unless /\*{3} +NONE +\*{3}/ =~ r then
297
+ @f0hitlist << r
298
+ while r = data.first and /^WARNING\: / =~ r
299
+ @f0warnings << data.shift
300
+ end
301
+ while r = data.first and /^\>/ =~ r
302
+ @hits << Hit.new(data)
303
+ end
304
+ end #unless
305
+ end
306
+
307
+ # Returns warning messages.
308
+ def warnings
309
+ @f0warnings
310
+ end
311
+
312
+ private
313
+ # Parses hit list.
314
+ def parse_hitlist
315
+ unless defined?(@parse_hitlist)
316
+ r = @f0hitlist.shift.to_s
317
+ if /Reading/ =~ r and /Frame/ =~ r then
318
+ flag_tblast = true
319
+ spnum = 5
320
+ else
321
+ flag_tblast = nil
322
+ spnum = 4
323
+ end
324
+ i = 0
325
+ @f0hitlist.each do |x|
326
+ b = x.split(/^/)
327
+ b.collect! { |y| y.empty? ? nil : y }
328
+ b.compact!
329
+ b.each do |y|
330
+ y.strip!
331
+ y.reverse!
332
+ z = y.split(/\s+/, spnum)
333
+ z.each { |y| y.reverse! }
334
+ dfl = z.pop
335
+ h = @hits[i]
336
+ unless h then
337
+ h = Hit.new([ dfl.to_s.sub(/\.+\z/, '') ])
338
+ @hits[i] = h
339
+ end
340
+ z.pop if flag_tblast #ignore Reading Frame
341
+ scr = z.pop
342
+ scr = (scr ? scr.to_i : nil)
343
+ pval = z.pop.to_s
344
+ pval = '1' + pval if pval[0] == ?e
345
+ pval = (pval.empty? ? (1.0/0.0) : pval.to_f)
346
+ nnum = z.pop.to_i
347
+ h.instance_eval {
348
+ @score = scr
349
+ @pvalue = pval
350
+ @n_number = nnum
351
+ }
352
+ i += 1
353
+ end
354
+ end #each
355
+ @parse_hitlist = true
356
+ end #unless
357
+ end
358
+ end #class Iteration
359
+
360
+ # Bio::Blast::WU::Report::Hit contains information about a hit.
361
+ # It may contain some Bio::Blast::WU::Report::HSP objects.
362
+ #
363
+ # Because it inherits Bio::Blast::Default::Report::Hit,
364
+ # please also refer Bio::Blast::Default::Report::Hit.
365
+ class Hit < Default::Report::Hit
366
+ # Creates a new Hit object.
367
+ # It is designed to be called only internally from the
368
+ # Bio::Blast::WU::Report::Iteration class.
369
+ # Users should not call the method directly.
370
+ def initialize(data)
371
+ @f0hitname = data.shift
372
+ @hsps = []
373
+ while r = data.first
374
+ if r =~ /^\s*(?:Plus|Minus) +Strand +HSPs\:/ then
375
+ data.shift
376
+ r = data.first
377
+ end
378
+ if /\A\s+Score/ =~ r then
379
+ @hsps << HSP.new(data)
380
+ else
381
+ break
382
+ end
383
+ end
384
+ @again = false
385
+ end
386
+
387
+ # Returns score.
388
+ def score
389
+ @score
390
+ end
391
+ # p-value
392
+ attr_reader :pvalue
393
+ # n-number (???)
394
+ attr_reader :n_number
395
+ end #class Hit
396
+
397
+ # Bio::Blast::WU::Report::HSP holds information about the hsp
398
+ # (high-scoring segment pair).
399
+ #
400
+ # Because it inherits Bio::Blast::Default::Report::HSP,
401
+ # please also refer Bio::Blast::Default::Report::HSP.
402
+ class HSP < Default::Report::HSP
403
+ # p-value
404
+ attr_reader :pvalue if false #dummy
405
+ method_after_parse_score :pvalue
406
+ # p_sum_n (???)
407
+ attr_reader :p_sum_n if false #dummy
408
+ method_after_parse_score :p_sum_n
409
+ end #class HSP
410
+
411
+ end #class Report
412
+
413
+ # WU-BLAST default output parser for TBLAST.
414
+ # All methods are equal to Bio::Blast::WU::Report.
415
+ # Only DELIMITER (and RS) is different.
416
+ class Report_TBlast < Report
417
+ # Delimter of each entry for TBLAST. Bio::FlatFile uses it.
418
+ DELIMITER = RS = "\nTBLAST"
419
+
420
+ # (Integer) excess read size included in DELIMITER.
421
+ DELIMITER_OVERRUN = 6 # "TBLAST"
422
+ end #class Report_TBlast
423
+
424
+ end #module WU
425
+ end #class Blast
426
+ end #module Bio
427
+
428
+ ######################################################################
429
+
430
+ if __FILE__ == $0
431
+
432
+ Bio::FlatFile.open(Bio::Blast::WU::Report, ARGF) do |ff|
433
+ ff.each do |rep|
434
+
435
+ print "# === Bio::Blast::WU::Report\n"
436
+ puts
437
+ print " rep.program #=> "; p rep.program
438
+ print " rep.version #=> "; p rep.version
439
+ print " rep.reference #=> "; p rep.reference
440
+ print " rep.notice #=> "; p rep.notice
441
+ print " rep.db #=> "; p rep.db
442
+ #print " rep.query_id #=> "; p rep.query_id
443
+ print " rep.query_def #=> "; p rep.query_def
444
+ print " rep.query_len #=> "; p rep.query_len
445
+ #puts
446
+ print " rep.version_number #=> "; p rep.version_number
447
+ print " rep.version_date #=> "; p rep.version_date
448
+ puts
449
+
450
+ print "# === Parameters\n"
451
+ #puts
452
+ print " rep.parameters #=> "; p rep.parameters
453
+ puts
454
+ #@#print " rep.matrix #=> "; p rep.matrix
455
+ print " rep.expect #=> "; p rep.expect
456
+ #print " rep.inclusion #=> "; p rep.inclusion
457
+ #@#print " rep.sc_match #=> "; p rep.sc_match
458
+ #@#print " rep.sc_mismatch #=> "; p rep.sc_mismatch
459
+ #@#print " rep.gap_open #=> "; p rep.gap_open
460
+ #@#print " rep.gap_extend #=> "; p rep.gap_extend
461
+ #print " rep.filter #=> "; p rep.filter
462
+ #@#print " rep.pattern #=> "; p rep.pattern
463
+ #print " rep.entrez_query #=> "; p rep.entrez_query
464
+ #puts
465
+ #@#print " rep.pattern_positions #=> "; p rep.pattern_positions
466
+ puts
467
+
468
+ print "# === Statistics (last iteration's)\n"
469
+ #puts
470
+ #print " rep.statistics #=> "; p rep.statistics
471
+ puts
472
+ print " rep.db_num #=> "; p rep.db_num
473
+ print " rep.db_len #=> "; p rep.db_len
474
+ #print " rep.hsp_len #=> "; p rep.hsp_len
475
+ #@#print " rep.eff_space #=> "; p rep.eff_space
476
+ #@#print " rep.kappa #=> "; p rep.kappa
477
+ #@#print " rep.lambda #=> "; p rep.lambda
478
+ #@#print " rep.entropy #=> "; p rep.entropy
479
+ puts
480
+ #@#print " rep.num_hits #=> "; p rep.num_hits
481
+ #@#print " rep.gapped_kappa #=> "; p rep.gapped_kappa
482
+ #@#print " rep.gapped_lambda #=> "; p rep.gapped_lambda
483
+ #@#print " rep.gapped_entropy #=> "; p rep.gapped_entropy
484
+ #@#print " rep.posted_date #=> "; p rep.posted_date
485
+ puts
486
+
487
+ #@#print "# === Message (last iteration's)\n"
488
+ #@#puts
489
+ #@#print " rep.message #=> "; p rep.message
490
+ #puts
491
+ #@#print " rep.converged? #=> "; p rep.converged?
492
+ #puts
493
+
494
+ print "# === Warning messages\n"
495
+ print " rep.warnings #=> "; p rep.warnings
496
+
497
+ print "# === Iterations\n"
498
+ puts
499
+ print " rep.itrerations.each do |itr|\n"
500
+ puts
501
+
502
+ rep.iterations.each do |itr|
503
+
504
+ print "# --- Bio::Blast::WU::Report::Iteration\n"
505
+ puts
506
+
507
+ print " itr.num #=> "; p itr.num
508
+ #print " itr.statistics #=> "; p itr.statistics
509
+ puts
510
+ print " itr.warnings #=> "; p itr.warnings
511
+ print " itr.message #=> "; p itr.message
512
+ print " itr.hits.size #=> "; p itr.hits.size
513
+ #puts
514
+ #@#print " itr.hits_newly_found.size #=> "; p itr.hits_newly_found.size;
515
+ #@#print " itr.hits_found_again.size #=> "; p itr.hits_found_again.size;
516
+ if itr.hits_for_pattern then
517
+ itr.hits_for_pattern.each_with_index do |hp, hpi|
518
+ print " itr.hits_for_pattern[#{hpi}].size #=> "; p hp.size;
519
+ end
520
+ end
521
+ print " itr.converged? #=> "; p itr.converged?
522
+ puts
523
+
524
+ print " itr.hits.each do |hit|\n"
525
+ puts
526
+
527
+ itr.hits.each_with_index do |hit, i|
528
+
529
+ print "# --- Bio::Blast::WU::Report::Hit"
530
+ print " ([#{i}])\n"
531
+ puts
532
+
533
+ #print " hit.num #=> "; p hit.num
534
+ #print " hit.hit_id #=> "; p hit.hit_id
535
+ print " hit.len #=> "; p hit.len
536
+ print " hit.definition #=> "; p hit.definition
537
+ #print " hit.accession #=> "; p hit.accession
538
+ #puts
539
+ print " hit.found_again? #=> "; p hit.found_again?
540
+ #puts
541
+ print " hit.score #=> "; p hit.score
542
+ print " hit.pvalue #=> "; p hit.pvalue
543
+ print " hit.n_number #=> "; p hit.n_number
544
+
545
+ print " --- compatible/shortcut ---\n"
546
+ #print " hit.query_id #=> "; p hit.query_id
547
+ #print " hit.query_def #=> "; p hit.query_def
548
+ #print " hit.query_len #=> "; p hit.query_len
549
+ #print " hit.target_id #=> "; p hit.target_id
550
+ print " hit.target_def #=> "; p hit.target_def
551
+ print " hit.target_len #=> "; p hit.target_len
552
+
553
+ print " --- first HSP's values (shortcut) ---\n"
554
+ print " hit.evalue #=> "; p hit.evalue
555
+ print " hit.bit_score #=> "; p hit.bit_score
556
+ print " hit.identity #=> "; p hit.identity
557
+ #print " hit.overlap #=> "; p hit.overlap
558
+
559
+ print " hit.query_seq #=> "; p hit.query_seq
560
+ print " hit.midline #=> "; p hit.midline
561
+ print " hit.target_seq #=> "; p hit.target_seq
562
+
563
+ print " hit.query_start #=> "; p hit.query_start
564
+ print " hit.query_end #=> "; p hit.query_end
565
+ print " hit.target_start #=> "; p hit.target_start
566
+ print " hit.target_end #=> "; p hit.target_end
567
+ print " hit.lap_at #=> "; p hit.lap_at
568
+ print " --- first HSP's vaules (shortcut) ---\n"
569
+ print " --- compatible/shortcut ---\n"
570
+
571
+ puts
572
+ print " hit.hsps.size #=> "; p hit.hsps.size
573
+ if hit.hsps.size == 0 then
574
+ puts " (HSP not found: please see blastall's -b and -v options)"
575
+ puts
576
+ else
577
+
578
+ puts
579
+ print " hit.hsps.each do |hsp|\n"
580
+ puts
581
+
582
+ hit.hsps.each_with_index do |hsp, j|
583
+
584
+ print "# --- Bio::Blast::WU::Report::Hsp"
585
+ print " ([#{j}])\n"
586
+ puts
587
+ #print " hsp.num #=> "; p hsp.num
588
+ print " hsp.bit_score #=> "; p hsp.bit_score
589
+ print " hsp.score #=> "; p hsp.score
590
+ print " hsp.evalue #=> "; p hsp.evalue
591
+ print " hsp.identity #=> "; p hsp.identity
592
+ print " hsp.gaps #=> "; p hsp.gaps
593
+ print " hsp.positive #=> "; p hsp.positive
594
+ print " hsp.align_len #=> "; p hsp.align_len
595
+ #print " hsp.density #=> "; p hsp.density
596
+ puts
597
+ print " hsp.pvalue #=> "; p hsp.pvalue
598
+ print " hsp.p_sum_n #=> "; p hsp.p_sum_n
599
+ puts
600
+
601
+ print " hsp.query_frame #=> "; p hsp.query_frame
602
+ print " hsp.query_from #=> "; p hsp.query_from
603
+ print " hsp.query_to #=> "; p hsp.query_to
604
+
605
+ print " hsp.hit_frame #=> "; p hsp.hit_frame
606
+ print " hsp.hit_from #=> "; p hsp.hit_from
607
+ print " hsp.hit_to #=> "; p hsp.hit_to
608
+
609
+ #print " hsp.pattern_from#=> "; p hsp.pattern_from
610
+ #print " hsp.pattern_to #=> "; p hsp.pattern_to
611
+
612
+ print " hsp.qseq #=> "; p hsp.qseq
613
+ print " hsp.midline #=> "; p hsp.midline
614
+ print " hsp.hseq #=> "; p hsp.hseq
615
+ puts
616
+ print " hsp.percent_identity #=> "; p hsp.percent_identity
617
+ #print " hsp.mismatch_count #=> "; p hsp.mismatch_count
618
+ #
619
+ print " hsp.query_strand #=> "; p hsp.query_strand
620
+ print " hsp.hit_strand #=> "; p hsp.hit_strand
621
+ print " hsp.percent_positive #=> "; p hsp.percent_positive
622
+ print " hsp.percent_gaps #=> "; p hsp.percent_gaps
623
+ puts
624
+
625
+ end #each
626
+ end #if hit.hsps.size == 0
627
+ end
628
+ end
629
+ end #ff.each
630
+ end #FlatFile.open
631
+
632
+ end #if __FILE__ == $0
633
+
634
+ ######################################################################
635
+