jandot-bio 1.2.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (335) hide show
  1. data/bin/bioruby +44 -0
  2. data/bin/br_biofetch.rb +47 -0
  3. data/bin/br_bioflat.rb +282 -0
  4. data/bin/br_biogetseq.rb +45 -0
  5. data/bin/br_pmfetch.rb +421 -0
  6. data/doc/Changes-0.7.rd +369 -0
  7. data/doc/KEGG_API.rd +1843 -0
  8. data/doc/KEGG_API.rd.ja +1834 -0
  9. data/doc/Tutorial.rd +1296 -0
  10. data/doc/Tutorial.rd.ja +2640 -0
  11. data/etc/bioinformatics/seqdatabase.ini +210 -0
  12. data/lib/bio.rb +279 -0
  13. data/lib/bio/alignment.rb +2518 -0
  14. data/lib/bio/appl/bl2seq/report.rb +334 -0
  15. data/lib/bio/appl/blast.rb +351 -0
  16. data/lib/bio/appl/blast/format0.rb +1438 -0
  17. data/lib/bio/appl/blast/format8.rb +83 -0
  18. data/lib/bio/appl/blast/report.rb +516 -0
  19. data/lib/bio/appl/blast/rexml.rb +135 -0
  20. data/lib/bio/appl/blast/rpsblast.rb +176 -0
  21. data/lib/bio/appl/blast/wublast.rb +550 -0
  22. data/lib/bio/appl/blast/xmlparser.rb +228 -0
  23. data/lib/bio/appl/blat/report.rb +489 -0
  24. data/lib/bio/appl/clustalw.rb +219 -0
  25. data/lib/bio/appl/clustalw/report.rb +152 -0
  26. data/lib/bio/appl/emboss.rb +203 -0
  27. data/lib/bio/appl/fasta.rb +237 -0
  28. data/lib/bio/appl/fasta/format10.rb +325 -0
  29. data/lib/bio/appl/gcg/msf.rb +212 -0
  30. data/lib/bio/appl/gcg/seq.rb +195 -0
  31. data/lib/bio/appl/genscan/report.rb +552 -0
  32. data/lib/bio/appl/hmmer.rb +126 -0
  33. data/lib/bio/appl/hmmer/report.rb +683 -0
  34. data/lib/bio/appl/iprscan/report.rb +374 -0
  35. data/lib/bio/appl/mafft.rb +259 -0
  36. data/lib/bio/appl/mafft/report.rb +226 -0
  37. data/lib/bio/appl/muscle.rb +52 -0
  38. data/lib/bio/appl/phylip/alignment.rb +129 -0
  39. data/lib/bio/appl/phylip/distance_matrix.rb +96 -0
  40. data/lib/bio/appl/probcons.rb +41 -0
  41. data/lib/bio/appl/psort.rb +548 -0
  42. data/lib/bio/appl/psort/report.rb +457 -0
  43. data/lib/bio/appl/pts1.rb +263 -0
  44. data/lib/bio/appl/sim4.rb +124 -0
  45. data/lib/bio/appl/sim4/report.rb +485 -0
  46. data/lib/bio/appl/sosui/report.rb +151 -0
  47. data/lib/bio/appl/spidey/report.rb +593 -0
  48. data/lib/bio/appl/targetp/report.rb +267 -0
  49. data/lib/bio/appl/tcoffee.rb +55 -0
  50. data/lib/bio/appl/tmhmm/report.rb +222 -0
  51. data/lib/bio/command.rb +337 -0
  52. data/lib/bio/data/aa.rb +349 -0
  53. data/lib/bio/data/codontable.rb +722 -0
  54. data/lib/bio/data/na.rb +223 -0
  55. data/lib/bio/db.rb +329 -0
  56. data/lib/bio/db/aaindex.rb +357 -0
  57. data/lib/bio/db/embl/common.rb +336 -0
  58. data/lib/bio/db/embl/embl.rb +402 -0
  59. data/lib/bio/db/embl/sptr.rb +1283 -0
  60. data/lib/bio/db/embl/swissprot.rb +42 -0
  61. data/lib/bio/db/embl/trembl.rb +41 -0
  62. data/lib/bio/db/embl/uniprot.rb +42 -0
  63. data/lib/bio/db/fantom.rb +599 -0
  64. data/lib/bio/db/fasta.rb +907 -0
  65. data/lib/bio/db/genbank/common.rb +290 -0
  66. data/lib/bio/db/genbank/ddbj.rb +22 -0
  67. data/lib/bio/db/genbank/genbank.rb +215 -0
  68. data/lib/bio/db/genbank/genpept.rb +60 -0
  69. data/lib/bio/db/genbank/refseq.rb +18 -0
  70. data/lib/bio/db/gff.rb +174 -0
  71. data/lib/bio/db/go.rb +481 -0
  72. data/lib/bio/db/kegg/brite.rb +41 -0
  73. data/lib/bio/db/kegg/compound.rb +131 -0
  74. data/lib/bio/db/kegg/drug.rb +98 -0
  75. data/lib/bio/db/kegg/enzyme.rb +148 -0
  76. data/lib/bio/db/kegg/expression.rb +155 -0
  77. data/lib/bio/db/kegg/genes.rb +263 -0
  78. data/lib/bio/db/kegg/genome.rb +241 -0
  79. data/lib/bio/db/kegg/glycan.rb +170 -0
  80. data/lib/bio/db/kegg/keggtab.rb +357 -0
  81. data/lib/bio/db/kegg/kgml.rb +256 -0
  82. data/lib/bio/db/kegg/orthology.rb +136 -0
  83. data/lib/bio/db/kegg/reaction.rb +82 -0
  84. data/lib/bio/db/kegg/taxonomy.rb +331 -0
  85. data/lib/bio/db/lasergene.rb +209 -0
  86. data/lib/bio/db/litdb.rb +107 -0
  87. data/lib/bio/db/medline.rb +323 -0
  88. data/lib/bio/db/nbrf.rb +191 -0
  89. data/lib/bio/db/newick.rb +658 -0
  90. data/lib/bio/db/nexus.rb +1854 -0
  91. data/lib/bio/db/pdb.rb +29 -0
  92. data/lib/bio/db/pdb/atom.rb +77 -0
  93. data/lib/bio/db/pdb/chain.rb +210 -0
  94. data/lib/bio/db/pdb/chemicalcomponent.rb +224 -0
  95. data/lib/bio/db/pdb/model.rb +148 -0
  96. data/lib/bio/db/pdb/pdb.rb +1911 -0
  97. data/lib/bio/db/pdb/residue.rb +176 -0
  98. data/lib/bio/db/pdb/utils.rb +399 -0
  99. data/lib/bio/db/prosite.rb +597 -0
  100. data/lib/bio/db/rebase.rb +457 -0
  101. data/lib/bio/db/soft.rb +404 -0
  102. data/lib/bio/db/transfac.rb +375 -0
  103. data/lib/bio/feature.rb +226 -0
  104. data/lib/bio/io/das.rb +461 -0
  105. data/lib/bio/io/dbget.rb +194 -0
  106. data/lib/bio/io/ddbjxml.rb +581 -0
  107. data/lib/bio/io/ebisoap.rb +158 -0
  108. data/lib/bio/io/ensembl.rb +229 -0
  109. data/lib/bio/io/fastacmd.rb +163 -0
  110. data/lib/bio/io/fetch.rb +181 -0
  111. data/lib/bio/io/flatfile.rb +1309 -0
  112. data/lib/bio/io/flatfile/bdb.rb +253 -0
  113. data/lib/bio/io/flatfile/index.rb +1371 -0
  114. data/lib/bio/io/flatfile/indexer.rb +787 -0
  115. data/lib/bio/io/higet.rb +73 -0
  116. data/lib/bio/io/hinv.rb +442 -0
  117. data/lib/bio/io/keggapi.rb +805 -0
  118. data/lib/bio/io/ncbirest.rb +256 -0
  119. data/lib/bio/io/ncbisoap.rb +155 -0
  120. data/lib/bio/io/pubmed.rb +307 -0
  121. data/lib/bio/io/registry.rb +292 -0
  122. data/lib/bio/io/soapwsdl.rb +119 -0
  123. data/lib/bio/io/sql.rb +365 -0
  124. data/lib/bio/location.rb +772 -0
  125. data/lib/bio/map.rb +410 -0
  126. data/lib/bio/pathway.rb +854 -0
  127. data/lib/bio/reference.rb +623 -0
  128. data/lib/bio/sequence.rb +475 -0
  129. data/lib/bio/sequence/aa.rb +125 -0
  130. data/lib/bio/sequence/common.rb +333 -0
  131. data/lib/bio/sequence/compat.rb +123 -0
  132. data/lib/bio/sequence/format.rb +181 -0
  133. data/lib/bio/sequence/generic.rb +24 -0
  134. data/lib/bio/sequence/na.rb +491 -0
  135. data/lib/bio/shell.rb +44 -0
  136. data/lib/bio/shell/core.rb +578 -0
  137. data/lib/bio/shell/demo.rb +146 -0
  138. data/lib/bio/shell/interface.rb +218 -0
  139. data/lib/bio/shell/irb.rb +95 -0
  140. data/lib/bio/shell/object.rb +71 -0
  141. data/lib/bio/shell/plugin/blast.rb +42 -0
  142. data/lib/bio/shell/plugin/codon.rb +218 -0
  143. data/lib/bio/shell/plugin/das.rb +58 -0
  144. data/lib/bio/shell/plugin/emboss.rb +23 -0
  145. data/lib/bio/shell/plugin/entry.rb +105 -0
  146. data/lib/bio/shell/plugin/flatfile.rb +101 -0
  147. data/lib/bio/shell/plugin/keggapi.rb +181 -0
  148. data/lib/bio/shell/plugin/midi.rb +430 -0
  149. data/lib/bio/shell/plugin/obda.rb +45 -0
  150. data/lib/bio/shell/plugin/psort.rb +56 -0
  151. data/lib/bio/shell/plugin/seq.rb +247 -0
  152. data/lib/bio/shell/plugin/soap.rb +87 -0
  153. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/bioruby_generator.rb +29 -0
  154. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_classes.rhtml +4 -0
  155. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_log.rhtml +27 -0
  156. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_methods.rhtml +11 -0
  157. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_modules.rhtml +4 -0
  158. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/_variables.rhtml +7 -0
  159. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-bg.gif +0 -0
  160. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-gem.png +0 -0
  161. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby-link.gif +0 -0
  162. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.css +368 -0
  163. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby.rhtml +47 -0
  164. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_controller.rb +144 -0
  165. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/bioruby_helper.rb +47 -0
  166. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/commands.rhtml +8 -0
  167. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/history.rhtml +10 -0
  168. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/index.rhtml +26 -0
  169. data/lib/bio/shell/rails/vendor/plugins/bioruby/generators/bioruby/templates/spinner.gif +0 -0
  170. data/lib/bio/shell/script.rb +25 -0
  171. data/lib/bio/shell/setup.rb +109 -0
  172. data/lib/bio/shell/web.rb +102 -0
  173. data/lib/bio/tree.rb +850 -0
  174. data/lib/bio/util/color_scheme.rb +191 -0
  175. data/lib/bio/util/color_scheme/buried.rb +59 -0
  176. data/lib/bio/util/color_scheme/helix.rb +59 -0
  177. data/lib/bio/util/color_scheme/hydropathy.rb +64 -0
  178. data/lib/bio/util/color_scheme/nucleotide.rb +31 -0
  179. data/lib/bio/util/color_scheme/strand.rb +59 -0
  180. data/lib/bio/util/color_scheme/taylor.rb +50 -0
  181. data/lib/bio/util/color_scheme/turn.rb +59 -0
  182. data/lib/bio/util/color_scheme/zappo.rb +50 -0
  183. data/lib/bio/util/contingency_table.rb +370 -0
  184. data/lib/bio/util/restriction_enzyme.rb +228 -0
  185. data/lib/bio/util/restriction_enzyme/analysis.rb +249 -0
  186. data/lib/bio/util/restriction_enzyme/analysis_basic.rb +217 -0
  187. data/lib/bio/util/restriction_enzyme/cut_symbol.rb +107 -0
  188. data/lib/bio/util/restriction_enzyme/double_stranded.rb +321 -0
  189. data/lib/bio/util/restriction_enzyme/double_stranded/aligned_strands.rb +130 -0
  190. data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair.rb +103 -0
  191. data/lib/bio/util/restriction_enzyme/double_stranded/cut_location_pair_in_enzyme_notation.rb +38 -0
  192. data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations.rb +76 -0
  193. data/lib/bio/util/restriction_enzyme/double_stranded/cut_locations_in_enzyme_notation.rb +107 -0
  194. data/lib/bio/util/restriction_enzyme/enzymes.yaml +7061 -0
  195. data/lib/bio/util/restriction_enzyme/range/cut_range.rb +24 -0
  196. data/lib/bio/util/restriction_enzyme/range/cut_ranges.rb +47 -0
  197. data/lib/bio/util/restriction_enzyme/range/horizontal_cut_range.rb +67 -0
  198. data/lib/bio/util/restriction_enzyme/range/sequence_range.rb +257 -0
  199. data/lib/bio/util/restriction_enzyme/range/sequence_range/calculated_cuts.rb +242 -0
  200. data/lib/bio/util/restriction_enzyme/range/sequence_range/fragment.rb +51 -0
  201. data/lib/bio/util/restriction_enzyme/range/sequence_range/fragments.rb +41 -0
  202. data/lib/bio/util/restriction_enzyme/range/vertical_cut_range.rb +77 -0
  203. data/lib/bio/util/restriction_enzyme/single_strand.rb +199 -0
  204. data/lib/bio/util/restriction_enzyme/single_strand/cut_locations_in_enzyme_notation.rb +135 -0
  205. data/lib/bio/util/restriction_enzyme/single_strand_complement.rb +23 -0
  206. data/lib/bio/util/restriction_enzyme/string_formatting.rb +111 -0
  207. data/lib/bio/util/sirna.rb +288 -0
  208. data/sample/any2fasta.rb +59 -0
  209. data/sample/biofetch.rb +475 -0
  210. data/sample/color_scheme_na.rb +91 -0
  211. data/sample/dbget +37 -0
  212. data/sample/enzymes.rb +78 -0
  213. data/sample/fasta2tab.rb +99 -0
  214. data/sample/fastagrep.rb +72 -0
  215. data/sample/fastasort.rb +54 -0
  216. data/sample/fsplit.rb +51 -0
  217. data/sample/gb2fasta.rb +30 -0
  218. data/sample/gb2tab.rb +325 -0
  219. data/sample/gbtab2mysql.rb +161 -0
  220. data/sample/genes2nuc.rb +33 -0
  221. data/sample/genes2pep.rb +33 -0
  222. data/sample/genes2tab.rb +81 -0
  223. data/sample/genome2rb.rb +29 -0
  224. data/sample/genome2tab.rb +76 -0
  225. data/sample/goslim.rb +303 -0
  226. data/sample/gt2fasta.rb +47 -0
  227. data/sample/na2aa.rb +34 -0
  228. data/sample/pmfetch.rb +42 -0
  229. data/sample/pmsearch.rb +42 -0
  230. data/sample/psortplot_html.rb +214 -0
  231. data/sample/ssearch2tab.rb +96 -0
  232. data/sample/tdiary.rb +158 -0
  233. data/sample/tfastx2tab.rb +100 -0
  234. data/sample/vs-genes.rb +212 -0
  235. data/test/data/HMMER/hmmpfam.out +64 -0
  236. data/test/data/HMMER/hmmsearch.out +88 -0
  237. data/test/data/SOSUI/sample.report +11 -0
  238. data/test/data/TMHMM/sample.report +21 -0
  239. data/test/data/aaindex/DAYM780301 +30 -0
  240. data/test/data/aaindex/PRAM900102 +20 -0
  241. data/test/data/bl2seq/cd8a_cd8b_blastp.bl2seq +53 -0
  242. data/test/data/bl2seq/cd8a_p53_e-5blastp.bl2seq +37 -0
  243. data/test/data/blast/2.2.15.blastp.m7 +876 -0
  244. data/test/data/blast/b0002.faa +15 -0
  245. data/test/data/blast/b0002.faa.m0 +128 -0
  246. data/test/data/blast/b0002.faa.m7 +65 -0
  247. data/test/data/blast/b0002.faa.m8 +1 -0
  248. data/test/data/embl/AB090716.embl +65 -0
  249. data/test/data/embl/AB090716.embl.rel89 +63 -0
  250. data/test/data/fasta/example1.txt +75 -0
  251. data/test/data/fasta/example2.txt +21 -0
  252. data/test/data/genscan/sample.report +63 -0
  253. data/test/data/iprscan/merged.raw +32 -0
  254. data/test/data/iprscan/merged.txt +74 -0
  255. data/test/data/prosite/prosite.dat +2233 -0
  256. data/test/data/refseq/nm_126355.entret +64 -0
  257. data/test/data/soft/GDS100_partial.soft +92 -0
  258. data/test/data/soft/GSE3457_family_partial.soft +874 -0
  259. data/test/data/uniprot/p53_human.uniprot +1456 -0
  260. data/test/functional/bio/io/test_ensembl.rb +186 -0
  261. data/test/functional/bio/io/test_soapwsdl.rb +52 -0
  262. data/test/runner.rb +14 -0
  263. data/test/unit/bio/appl/bl2seq/test_report.rb +134 -0
  264. data/test/unit/bio/appl/blast/test_report.rb +417 -0
  265. data/test/unit/bio/appl/blast/test_xmlparser.rb +388 -0
  266. data/test/unit/bio/appl/genscan/test_report.rb +182 -0
  267. data/test/unit/bio/appl/hmmer/test_report.rb +342 -0
  268. data/test/unit/bio/appl/iprscan/test_report.rb +338 -0
  269. data/test/unit/bio/appl/mafft/test_report.rb +63 -0
  270. data/test/unit/bio/appl/sosui/test_report.rb +81 -0
  271. data/test/unit/bio/appl/targetp/test_report.rb +146 -0
  272. data/test/unit/bio/appl/test_blast.rb +163 -0
  273. data/test/unit/bio/appl/test_fasta.rb +130 -0
  274. data/test/unit/bio/appl/test_pts1.rb +140 -0
  275. data/test/unit/bio/appl/tmhmm/test_report.rb +126 -0
  276. data/test/unit/bio/data/test_aa.rb +90 -0
  277. data/test/unit/bio/data/test_codontable.rb +107 -0
  278. data/test/unit/bio/data/test_na.rb +80 -0
  279. data/test/unit/bio/db/embl/test_common.rb +117 -0
  280. data/test/unit/bio/db/embl/test_embl.rb +214 -0
  281. data/test/unit/bio/db/embl/test_embl_rel89.rb +219 -0
  282. data/test/unit/bio/db/embl/test_sptr.rb +1775 -0
  283. data/test/unit/bio/db/embl/test_uniprot.rb +31 -0
  284. data/test/unit/bio/db/kegg/test_genes.rb +45 -0
  285. data/test/unit/bio/db/pdb/test_pdb.rb +152 -0
  286. data/test/unit/bio/db/test_aaindex.rb +197 -0
  287. data/test/unit/bio/db/test_fasta.rb +250 -0
  288. data/test/unit/bio/db/test_gff.rb +127 -0
  289. data/test/unit/bio/db/test_lasergene.rb +95 -0
  290. data/test/unit/bio/db/test_newick.rb +293 -0
  291. data/test/unit/bio/db/test_nexus.rb +360 -0
  292. data/test/unit/bio/db/test_prosite.rb +1437 -0
  293. data/test/unit/bio/db/test_rebase.rb +101 -0
  294. data/test/unit/bio/db/test_soft.rb +138 -0
  295. data/test/unit/bio/io/test_ddbjxml.rb +75 -0
  296. data/test/unit/bio/io/test_ensembl.rb +109 -0
  297. data/test/unit/bio/io/test_fastacmd.rb +42 -0
  298. data/test/unit/bio/io/test_flatfile.rb +237 -0
  299. data/test/unit/bio/io/test_soapwsdl.rb +32 -0
  300. data/test/unit/bio/sequence/test_aa.rb +103 -0
  301. data/test/unit/bio/sequence/test_common.rb +174 -0
  302. data/test/unit/bio/sequence/test_compat.rb +69 -0
  303. data/test/unit/bio/sequence/test_na.rb +330 -0
  304. data/test/unit/bio/shell/plugin/test_seq.rb +185 -0
  305. data/test/unit/bio/test_alignment.rb +1025 -0
  306. data/test/unit/bio/test_command.rb +288 -0
  307. data/test/unit/bio/test_db.rb +96 -0
  308. data/test/unit/bio/test_feature.rb +116 -0
  309. data/test/unit/bio/test_location.rb +39 -0
  310. data/test/unit/bio/test_map.rb +230 -0
  311. data/test/unit/bio/test_pathway.rb +473 -0
  312. data/test/unit/bio/test_reference.rb +224 -0
  313. data/test/unit/bio/test_sequence.rb +329 -0
  314. data/test/unit/bio/test_shell.rb +18 -0
  315. data/test/unit/bio/test_tree.rb +593 -0
  316. data/test/unit/bio/util/restriction_enzyme/analysis/test_calculated_cuts.rb +299 -0
  317. data/test/unit/bio/util/restriction_enzyme/analysis/test_cut_ranges.rb +103 -0
  318. data/test/unit/bio/util/restriction_enzyme/analysis/test_sequence_range.rb +240 -0
  319. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_aligned_strands.rb +101 -0
  320. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair.rb +75 -0
  321. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_location_pair_in_enzyme_notation.rb +73 -0
  322. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations.rb +53 -0
  323. data/test/unit/bio/util/restriction_enzyme/double_stranded/test_cut_locations_in_enzyme_notation.rb +104 -0
  324. data/test/unit/bio/util/restriction_enzyme/single_strand/test_cut_locations_in_enzyme_notation.rb +83 -0
  325. data/test/unit/bio/util/restriction_enzyme/test_analysis.rb +246 -0
  326. data/test/unit/bio/util/restriction_enzyme/test_cut_symbol.rb +44 -0
  327. data/test/unit/bio/util/restriction_enzyme/test_double_stranded.rb +115 -0
  328. data/test/unit/bio/util/restriction_enzyme/test_single_strand.rb +147 -0
  329. data/test/unit/bio/util/restriction_enzyme/test_single_strand_complement.rb +147 -0
  330. data/test/unit/bio/util/restriction_enzyme/test_string_formatting.rb +60 -0
  331. data/test/unit/bio/util/test_color_scheme.rb +33 -0
  332. data/test/unit/bio/util/test_contingency_table.rb +94 -0
  333. data/test/unit/bio/util/test_restriction_enzyme.rb +42 -0
  334. data/test/unit/bio/util/test_sirna.rb +245 -0
  335. metadata +479 -0
@@ -0,0 +1,1025 @@
1
+ #
2
+ # test/unit/bio/test_alignment.rb - Unit test for Bio::Alignment
3
+ #
4
+ # Copyright:: Copyright (C) 2004
5
+ # Moses Hohman <mmhohman@northwestern.edu>
6
+ # 2005 Naohisa Goto <ng@bioruby.org>
7
+ # License:: The Ruby License
8
+ #
9
+ # $Id: test_alignment.rb,v 1.12 2007/12/26 13:55:40 ngoto Exp $
10
+ #
11
+
12
+ require 'pathname'
13
+ libpath = Pathname.new(File.join(File.dirname(__FILE__), ['..'] * 3, 'lib')).cleanpath.to_s
14
+ $:.unshift(libpath) unless $:.include?(libpath)
15
+
16
+ require 'test/unit'
17
+ require 'bio/alignment'
18
+
19
+ module Bio
20
+
21
+ class TestAlignmentPropertyMethods < Test::Unit::TestCase
22
+
23
+ def setup
24
+ @obj = Object.new
25
+ @obj.extend(Alignment::PropertyMethods)
26
+ end
27
+
28
+ def test_is_gap_default_false
29
+ assert_equal(false, @obj.is_gap?('a'), "\"a\" isn't a gap")
30
+ end
31
+
32
+ def test_is_gap_default_true
33
+ assert_equal(true, @obj.is_gap?('-'), '"-" is a gap')
34
+ end
35
+
36
+ def test_gap_regexp
37
+ assert_not_nil(@obj.gap_regexp)
38
+ end
39
+
40
+ def test_gap_regexp_never_nil
41
+ @obj.gap_regexp = nil
42
+ assert_not_nil(@obj.gap_regexp)
43
+ end
44
+
45
+ def test_gap_regexp=()
46
+ @obj.gap_regexp = /[^a-zA-Z0-9]/
47
+ assert_equal(/[^a-zA-Z0-9]/, @obj.gap_regexp)
48
+ end
49
+
50
+ def test_is_gap_nodefault_false
51
+ @obj.gap_regexp = /[^a-zA-Z0-9]/
52
+ assert_equal(false, @obj.is_gap?('3'))
53
+ end
54
+
55
+ def test_is_gap_nodefault_true
56
+ @obj.gap_regexp = /[^atgc]/
57
+ assert_equal(true, @obj.is_gap?('z'))
58
+ end
59
+
60
+ def test_gap_char_default
61
+ assert_not_nil(@obj.gap_char)
62
+ end
63
+
64
+ def test_gap_char_never_nil
65
+ @obj.gap_char = nil
66
+ assert_not_nil(@obj.gap_char)
67
+ end
68
+
69
+ def test_gap_char=()
70
+ @obj.gap_char = '#'
71
+ assert_equal('#', @obj.gap_char)
72
+ end
73
+
74
+ def test_missing_char_default
75
+ assert_not_nil(@obj.missing_char)
76
+ end
77
+
78
+ def test_missing_char_never_nil
79
+ @obj.missing_char = nil
80
+ assert_not_nil(@obj.missing_char)
81
+ end
82
+
83
+ def test_missing_char=()
84
+ @obj.missing_char = '_'
85
+ assert_equal('_', @obj.missing_char)
86
+ end
87
+
88
+ def test_seqclass_default
89
+ assert_not_nil(@obj.seqclass)
90
+ end
91
+
92
+ def test_seqclass_never_nil
93
+ @obj.seqclass = nil
94
+ assert_not_nil(@obj.seqclass)
95
+ end
96
+
97
+ def test_seqclass=()
98
+ @obj.seqclass = Sequence::NA
99
+ assert_equal(Sequence::NA, @obj.seqclass)
100
+ end
101
+
102
+ def test_get_all_property_default
103
+ assert_equal({}, @obj.get_all_property)
104
+ end
105
+
106
+ def test_get_all_property_nodefault
107
+ @obj.gap_regexp = /[^acgt]/
108
+ @obj.gap_char = '#'
109
+ @obj.missing_char = '_'
110
+ @obj.seqclass = Sequence::NA
111
+ assert_equal({ :gap_regexp => /[^acgt]/,
112
+ :gap_char => '#',
113
+ :missing_char => '_',
114
+ :seqclass => Sequence::NA },
115
+ @obj.get_all_property)
116
+ end
117
+
118
+ def test_set_all_property
119
+ h = { :gap_regexp => /[^acgt]/,
120
+ :gap_char => '#',
121
+ :missing_char => '_',
122
+ :seqclass => Sequence::NA }
123
+ @obj.set_all_property(h)
124
+ assert_equal(h, @obj.get_all_property)
125
+ end
126
+ end #class TestAlignmentPropertyMethods
127
+
128
+ # This is a unit test of Bio::Alignment::Site class and
129
+ # Bio::Alignment::SiteMethods module.
130
+ # Since Bio::Alignment::Site includes Bio::Alignment::SiteMethods,
131
+ # we can test both at a time.
132
+ class TestAlignmentSite < Test::Unit::TestCase
133
+
134
+ def test_has_gap_true
135
+ site = Alignment::Site[ 'a', '-', 'c', 'g', 't' ]
136
+ assert_equal(true, site.has_gap?)
137
+ end
138
+
139
+ def test_has_gap_false
140
+ site = Alignment::Site[ 'a', 'c', 'g', 't' ]
141
+ assert_equal(false, site.has_gap?)
142
+ end
143
+
144
+ def test_remove_gaps!
145
+ site = Alignment::Site[ 'a', '-', 'c', '-' ]
146
+ assert_equal(Alignment::Site['a', 'c'], site.remove_gaps!)
147
+ end
148
+
149
+ def test_remove_gaps_bang_not_removed
150
+ site = Alignment::Site[ 'a', 'c']
151
+ assert_equal(nil, site.remove_gaps!)
152
+ end
153
+
154
+ def test_consensus_string_default
155
+ site = Alignment::Site[ 'a', 'a', 'a', 'a']
156
+ assert_equal('a', site.consensus_string)
157
+ end
158
+
159
+ def test_consensus_string_default_nil
160
+ site = Alignment::Site[ 'a', 'a', 'a', 'c']
161
+ assert_nil(site.consensus_string)
162
+ end
163
+
164
+ def test_consensus_string_50percent
165
+ site = Alignment::Site[ 'a', 'a', 'c', 'g']
166
+ assert_equal('a', site.consensus_string(0.5))
167
+ end
168
+
169
+ def test_consensus_string_50percent_nil
170
+ site = Alignment::Site[ 'a', 'c', 'g', 't']
171
+ assert_nil(site.consensus_string(0.5))
172
+ end
173
+
174
+ def test_consensus_iupac
175
+ data = {
176
+ 'a' => [ 'a' ],
177
+ 'c' => [ 'c' ],
178
+ 'g' => [ 'g' ],
179
+ 't' => [ 't' ],
180
+ 't' => [ 't', 'u' ],
181
+ 'm' => [ 'a', 'c' ],
182
+ 'r' => [ 'a', 'g' ],
183
+ 'w' => [ 'a', 't' ],
184
+ 's' => [ 'c', 'g' ],
185
+ 'y' => [ 'c', 't' ],
186
+ 'k' => [ 'g', 't' ],
187
+ 'v' => [ 'a', 'c', 'g' ],
188
+ 'h' => [ 'a', 'c', 't' ],
189
+ 'd' => [ 'a', 'g', 't' ],
190
+ 'b' => [ 'c', 'g', 't' ],
191
+ 'n' => [ 'a', 'c', 'g', 't' ],
192
+ nil => [ 'z', 'a' ]
193
+ }
194
+ data.each do |cons, testdata|
195
+ site = Alignment::Site[ *testdata ]
196
+ assert_equal(cons, site.consensus_iupac,
197
+ "IUPAC consensus of #{testdata.join(',')} is #{cons}")
198
+ end
199
+ end
200
+
201
+ def test_match_line_amino_missing
202
+ site = Alignment::Site[ 'P', 'Q', 'R', 'S' ]
203
+ assert_equal(' ', site.match_line_amino)
204
+ end
205
+
206
+ def test_match_line_amino_100percent
207
+ site = Alignment::Site[ 'M', 'M', 'M', 'M' ]
208
+ assert_equal('*', site.match_line_amino)
209
+ end
210
+
211
+ def test_match_line_amino_strong
212
+ site = Alignment::Site[ 'N', 'E', 'Q', 'K' ]
213
+ assert_equal(':', site.match_line_amino)
214
+ end
215
+
216
+ def test_match_line_amino_weak
217
+ site = Alignment::Site[ 'S', 'G', 'N', 'D' ]
218
+ assert_equal('.', site.match_line_amino)
219
+ end
220
+
221
+ def test_match_line_nuc_missing
222
+ site = Alignment::Site[ 'A', 'C', 'G', 'T' ]
223
+ assert_equal(' ', site.match_line_nuc)
224
+ end
225
+
226
+ def test_match_line_nuc_100percent
227
+ site = Alignment::Site[ 'G', 'G', 'G', 'G' ]
228
+ assert_equal('*', site.match_line_nuc)
229
+ end
230
+ end #class TestAlignmentSite
231
+
232
+ # This is sample class for testing Bio::Alignment::EnumerableExtension.
233
+ class A < Array
234
+ include Alignment::EnumerableExtension
235
+ end
236
+
237
+ class TestAlignmentEnumerableExtension < Test::Unit::TestCase
238
+ def test_each_seq
239
+ expected_results = [ 'atg', 'aag', 'acg' ]
240
+ a = A[ *expected_results ]
241
+ a.each_seq do |x|
242
+ assert_equal(expected_results.shift, x)
243
+ end
244
+ assert(expected_results.empty?)
245
+ end
246
+
247
+ def test_seqclass_default
248
+ a = A.new
249
+ assert_equal(String, a.seqclass)
250
+ end
251
+
252
+ def test_seqclass
253
+ a = A[ Bio::Sequence::NA.new('atg') ]
254
+ assert_equal(Bio::Sequence::NA, a.seqclass)
255
+ end
256
+
257
+ def test_seqclass=()
258
+ a = A.new
259
+ assert_equal(String, a.seqclass)
260
+ a << Bio::Sequence::NA.new('a')
261
+ assert_equal(Bio::Sequence::NA, a.seqclass)
262
+ a.seqclass = Bio::Sequence::AA
263
+ assert_equal(Bio::Sequence::AA, a.seqclass)
264
+ end
265
+
266
+ def test_alignment_length
267
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
268
+ assert_equal(4, a.alignment_length)
269
+ end
270
+
271
+ def test_private_alignment_site
272
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
273
+ assert_equal(Alignment::Site[ '-', 't', 't', 't', '-' ],
274
+ a.instance_eval { _alignment_site(1) })
275
+ end
276
+
277
+ def test_alignment_site
278
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
279
+ assert_equal(Alignment::Site[ '-', 't', 't', 't', '-' ],
280
+ a.__send__(:_alignment_site, 1))
281
+ end
282
+
283
+ def test_each_site
284
+ expected_results = [
285
+ Alignment::Site[ 'a', 'a', 'a', 'a', '-' ],
286
+ Alignment::Site[ '-', 't', 't', 't', '-' ],
287
+ Alignment::Site[ '-', '-', 'g', 'g', '-' ],
288
+ Alignment::Site[ '-', '-', 'c', '-', '-' ]
289
+ ]
290
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
291
+ a.each_site do |site|
292
+ assert_equal(expected_results.shift, site)
293
+ end
294
+ assert(expected_results.empty?)
295
+ end
296
+
297
+ def test_each_site_step
298
+ expected_results = [
299
+ Alignment::Site[ '-', 't', 't', 't', '-' ], # site 1
300
+ Alignment::Site[ '-', 'a', 'g', 't', '-' ], # site 3
301
+ ]
302
+ a = A[ 'a', 'atgatc', 'atggcc', 'atgtga', '' ]
303
+ a.each_site_step(1, 4, 2) do |site|
304
+ assert_equal(expected_results.shift, site)
305
+ end
306
+ assert(expected_results.empty?)
307
+ end
308
+
309
+ def test_alignment_collect
310
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
311
+ assert_equal(Alignment::SequenceArray[ 'a', 'au', 'augc', 'aug', '' ],
312
+ a.alignment_collect { |x| x.gsub(/t/, 'u') })
313
+ end
314
+
315
+ def test_alignment_window
316
+ a = A[ 'a', 'at', 'atgca', 'atg', '' ]
317
+ assert_equal(Alignment::SequenceArray[ '', 't', 'tgc', 'tg', '' ],
318
+ a.alignment_window(1, 3))
319
+ end
320
+
321
+ def test_each_window
322
+ expected_results = [
323
+ Alignment::SequenceArray[ 'atg', 'tcg', '' ], # 0..2
324
+ Alignment::SequenceArray[ 'gca', 'gat', '' ], # 2..4
325
+ Alignment::SequenceArray[ 'atg', 'tgc', '' ], # 4..6
326
+ Alignment::SequenceArray[ 'c', 'a', '' ] # 7..7
327
+ ]
328
+ a = A[ 'atgcatgc', 'tcgatgca', '' ]
329
+ r = a.each_window(3, 2) do |x|
330
+ assert_equal(expected_results.shift, x)
331
+ end
332
+ assert_equal(expected_results.shift, r)
333
+ assert(expected_results.empty?)
334
+ end
335
+
336
+ def test_collect_each_site
337
+ a = A[ 'a', 'at', 'atgc', 'atg', '' ]
338
+ assert_equal(["aaaa-", "-ttt-", "--gg-", "--c--" ],
339
+ a.collect_each_site { |x| x.join('') })
340
+ end
341
+
342
+ def test_consensus_each_site_default
343
+ expected_results = [
344
+ Alignment::Site[ 'a', 'a', 'a', 'a', 'a' ],
345
+ Alignment::Site[ 'a', 'c', 'g', 't', '-' ]
346
+ ]
347
+
348
+ a = A[ 'aa', 'ac', 'ag', 'at', 'a-' ]
349
+ result = a.consensus_each_site do |site|
350
+ assert_equal(expected_results.shift, site)
351
+ 'x'
352
+ end
353
+ assert_equal('xx', result)
354
+ assert(expected_results.empty?)
355
+ end
356
+
357
+ def test_consensus_each_site_gap_mode_1
358
+ expected_results = [
359
+ Alignment::Site[ 'a', 'a', 'a', 'a', 'a' ]
360
+ ]
361
+
362
+ a = A[ 'aa', 'ac', 'ag', 'at', 'a-' ]
363
+ result = a.consensus_each_site(:gap_mode => 1) do |site|
364
+ assert_equal(expected_results.shift, site)
365
+ 'x'
366
+ end
367
+ assert_equal('x-', result)
368
+ assert(expected_results.empty?)
369
+ end
370
+
371
+ def test_consensus_each_site_gap_mode_minus1
372
+ expected_results = [
373
+ Alignment::Site[ 'a', 'a', 'a', 'a', 'a' ],
374
+ Alignment::Site[ 'a', 'c', 'g', 't' ]
375
+ ]
376
+
377
+ a = A[ 'aa', 'ac', 'ag', 'at', 'a-' ]
378
+ result = a.consensus_each_site(:gap_mode => -1) do |site|
379
+ assert_equal(expected_results.shift, site)
380
+ 'x'
381
+ end
382
+ assert_equal('xx', result)
383
+ assert(expected_results.empty?)
384
+ end
385
+
386
+ def test_consensus_string_default
387
+ a = A[ 'ata', 'aac', 'aag', 'aat' ]
388
+ assert_equal('a??', a.consensus_string)
389
+ end
390
+
391
+ def test_consensus_string_half
392
+ a = A[ 'ata', 'aac', 'aag', 'aat' ]
393
+ assert_equal('aa?', a.consensus_string(0.5))
394
+ end
395
+
396
+ def test_consensus_iupac
397
+ a = A[
398
+ 'acgtaaaccgaaacaz',
399
+ 'acgtaaaccgccggcz',
400
+ 'acgtcgtgttgtttgz',
401
+ 'acgtcgtgttaaactz'
402
+ ]
403
+ assert_equal('acgtmrwsykvhdbn?', a.consensus_iupac)
404
+ end
405
+
406
+ def test_match_line_amino
407
+ a = A[
408
+ 'M-SNNNQMMHF-CASSSSSNNFH-AW',
409
+ 'M-TEHDHIIYY-STATTGNDEVF-FW',
410
+ 'M-AQQERLLHW-AVGNPNDEQLY-HW',
411
+ 'M-SKKQKVFYF-CASKADEQHIH-LW',
412
+ 'M-TNNNQMMHY-STASSSQHRMF-QW',
413
+ 'M-AEHDHIIYW-AVGTTGKKKFY-YW'
414
+ #* ::::::::: ........... *
415
+ ]
416
+ assert_equal('* ::::::::: ........... *', a.match_line_amino)
417
+ end
418
+
419
+ def test_match_line_nuc
420
+ a = A[ 'aaa', 'aa-','aac', 'at-' ]
421
+ assert_equal('* ', a.match_line_nuc)
422
+ end
423
+
424
+ def test_match_line
425
+ a = A[
426
+ Sequence::AA.new('MNSA'),
427
+ Sequence::AA.new('MHTL'),
428
+ Sequence::AA.new('MQNV'),
429
+ Sequence::AA.new('MKKW')
430
+ ]
431
+ assert_equal('*:. ', a.match_line)
432
+ assert_equal('*:. ', a.match_line(:type => :aa))
433
+ assert_equal('* ', a.match_line(:type => :na))
434
+ end
435
+
436
+ def test_convert_match
437
+ a = A[
438
+ 'aaaa',
439
+ 'accc',
440
+ 'acac',
441
+ 'actc'
442
+ ]
443
+ a.convert_match
444
+ assert_equal(A[ 'aaaa', '.ccc', '.c.c', '.ctc' ], a)
445
+ end
446
+
447
+ def test_convert_unmatch
448
+ a = A[ 'aaaa', '.ccc', '.c.c', '.ctc' ]
449
+ a.convert_unmatch
450
+ assert_equal(A[ 'aaaa', 'accc', 'acac', 'actc' ], a)
451
+ end
452
+
453
+ def test_alignment_normalize!
454
+ a = A[ 'a', 'atg', 'atgc', '' ]
455
+ a.alignment_normalize!
456
+ assert_equal(A[ 'a---', 'atg-', 'atgc', '----'], a)
457
+ end
458
+
459
+ def test_alignment_rstrip!
460
+ a = A[ '--aaa--', '--t-t--', '---g---', '--t' ]
461
+ assert(a.alignment_rstrip!)
462
+ assert_equal(A[ '--aaa', '--t-t', '---g-', '--t' ], a)
463
+ end
464
+
465
+ def test_alignment_rstrip_nil
466
+ a = A[ 'aa', '-a', 'a-' ]
467
+ assert_nil(a.alignment_rstrip!)
468
+ assert_equal(A[ 'aa', '-a', 'a-' ], a)
469
+ end
470
+
471
+ def test_alignment_lstrip!
472
+ a = A[ '--aaa--', '--t-t--', '---g---', '--t' ]
473
+ assert(a.alignment_lstrip!)
474
+ assert_equal(A[ 'aaa--', 't-t--', '-g---', 't' ], a)
475
+ end
476
+
477
+ def test_alignment_lstrip_nil
478
+ a = A[ 'aa', '-a', 'a-' ]
479
+ assert_nil(a.alignment_lstrip!)
480
+ assert_equal(A[ 'aa', '-a', 'a-' ], a)
481
+ end
482
+
483
+ def test_alignment_strip!
484
+ a = A[ '--aaa--', '--t-t--', '---g---', '--t' ]
485
+ assert(a.alignment_strip!)
486
+ assert_equal(A[ 'aaa', 't-t', '-g-', 't' ], a)
487
+ end
488
+
489
+ def test_alignment_strip_nil
490
+ a = A[ 'aa', '-a', 'a-' ]
491
+ assert_nil(a.alignment_strip!)
492
+ assert_equal(A[ 'aa', '-a', 'a-' ], a)
493
+ end
494
+
495
+ def test_remove_all_gaps!
496
+ a = A[ '--aaa--', '--t-t--', '---g---', '--t' ]
497
+ assert(a.remove_all_gaps!)
498
+ assert_equal(A[ 'aaa', 'tt', 'g', 't' ], a)
499
+ end
500
+
501
+ # test of alignment_slice.
502
+ # Please also refer alignment_window.
503
+ def test_alignment_slice
504
+ a = A[ 'a', 'at', 'atgca', 'atg', '' ]
505
+ assert_equal(Alignment::SequenceArray[ '', 't', 'tgc', 'tg', nil ],
506
+ a.alignment_slice(1, 3))
507
+ end
508
+
509
+ def test_alignment_subseq
510
+ a = A[
511
+ Sequence::NA.new('a'),
512
+ Sequence::NA.new('at'),
513
+ Sequence::NA.new('atgca'),
514
+ Sequence::NA.new('atg'),
515
+ Sequence::NA.new('')
516
+ ]
517
+ assert_equal(Alignment::SequenceArray[
518
+ Sequence::NA.new(''),
519
+ Sequence::NA.new('t'),
520
+ Sequence::NA.new('tgc'),
521
+ Sequence::NA.new('tg'),
522
+ nil
523
+ ], a.alignment_subseq(2,4))
524
+ end
525
+
526
+ def test_alignment_concat
527
+ a = A[ 'aaa', 'c', 'gg', 't' ]
528
+ a.alignment_concat(A[ 'ttt', 'gg', 'aa', 'cc', 'aa' ])
529
+ assert_equal(A[ 'aaattt', 'cgg', 'ggaa', 'tcc' ], a)
530
+ a.alignment_concat([ 'c', 't' ])
531
+ assert_equal(A[ 'aaatttc', 'cggt', 'ggaa', 'tcc' ], a)
532
+ end
533
+ end #class TestAlignmentEnumerableExtension
534
+
535
+ class TestAlignmentOutput < Test::Unit::TestCase
536
+ def setup
537
+ @obj = Object.new
538
+ @obj.extend(Alignment::Output)
539
+ end
540
+
541
+ def test_clustal_have_same_name_true
542
+ assert_equal([ 0, 1 ], @obj.instance_eval {
543
+ __clustal_have_same_name?([ 'ATP ATG', 'ATP ATA', 'BBB' ]) })
544
+ end
545
+
546
+ def test_have_same_name_false
547
+ assert_equal(false, @obj.instance_eval {
548
+ __clustal_have_same_name?([ 'GTP ATG', 'ATP ATA', 'BBB' ]) })
549
+ end
550
+
551
+ def test_avoid_same_name
552
+ assert_equal([ 'ATP_ATG', 'ATP_ATA', 'BBB' ],
553
+ @obj.instance_eval {
554
+ __clustal_avoid_same_name([ 'ATP ATG', 'ATP ATA', 'BBB' ]) })
555
+ end
556
+
557
+ def test_avoid_same_name_numbering
558
+ assert_equal([ '0_ATP', '1_ATP', '2_BBB' ],
559
+ @obj.instance_eval {
560
+ __clustal_avoid_same_name([ 'ATP', 'ATP', 'BBB' ]) })
561
+ end
562
+
563
+ end #class TestAlignmentOutput
564
+
565
+
566
+ class TestAlignment < Test::Unit::TestCase
567
+
568
+ # testing helper method
569
+ def build_na_alignment(*sequences)
570
+ sequences.inject(Alignment.new) { |alignment, sequence| alignment << Sequence::NA.new(sequence) }
571
+ end
572
+ private :build_na_alignment
573
+
574
+ def test_equals
575
+ alignment1 = Alignment.new([Sequence::NA.new("agct"), Sequence::NA.new("tagc")])
576
+ alignment2 = Alignment.new([Sequence::NA.new("agct"), Sequence::NA.new("tagc")])
577
+ assert_equal(alignment1, alignment2)
578
+ end
579
+
580
+ # Alignment#store
581
+
582
+ def test_store_cannot_override_key
583
+ alignment = Alignment.new
584
+ alignment.store("Cat DNA", Sequence::NA.new("cat"))
585
+ alignment.store("Cat DNA", Sequence::NA.new("gcat"))
586
+ assert_equal("cat", alignment["Cat DNA"])
587
+ end
588
+
589
+ def test_store_with_nil_key_uses_next_number_for_key
590
+ alignment = Alignment.new
591
+ alignment.store(nil, Sequence::NA.new("cat"))
592
+ alignment.store(nil, Sequence::NA.new("gat"))
593
+ alignment.store(nil, Sequence::NA.new("tat"))
594
+ assert_equal({0=>"cat",1=>"gat",2=>"tat"}, alignment.to_hash)
595
+ end
596
+
597
+ def test_store_with_default_keys_and_user_defined_keys
598
+ alignment = Alignment.new
599
+ alignment.store("cat key", Sequence::NA.new("cat"))
600
+ alignment.store(nil, Sequence::NA.new("cag"))
601
+ alignment.store("gat key", Sequence::NA.new("gat"))
602
+ alignment.store(nil, Sequence::NA.new("gag"))
603
+ assert_equal({"gat key"=>"gat",1=>"cag",3=>"gag","cat key"=>"cat"}, alignment.to_hash)
604
+ end
605
+
606
+ # Test append operator
607
+
608
+ def test_seqclass_when_sequence_used
609
+ alignment = Alignment.new
610
+ alignment << Sequence::NA.new("cat")
611
+ assert_equal({0=>"cat"}, alignment.to_hash)
612
+ end
613
+
614
+ # Test seqclass
615
+
616
+ def test_seqclass_when_sequence_used_no_seqclass_set
617
+ alignment = Alignment.new
618
+ alignment << Sequence::NA.new("cat")
619
+ assert_equal(Sequence::NA, alignment.seqclass)
620
+ end
621
+
622
+ def test_seqclass_String_seq_not_present_no_seqclass_set
623
+ alignment = Alignment.new
624
+ alignment << nil
625
+ assert_equal(String, alignment.seqclass)
626
+ end
627
+
628
+ def test_seqclass_when_seqclass_set
629
+ alignment = Alignment.new
630
+ alignment.seqclass = Fixnum
631
+ alignment << "this doesn't really make sense"
632
+ assert_equal(Fixnum, alignment.seqclass)
633
+ end
634
+
635
+ # Alignment#gap_char
636
+
637
+ def test_default_gap_char
638
+ alignment = Alignment.new
639
+ assert_equal("-", alignment.gap_char)
640
+ end
641
+
642
+ def test_set_and_get_gap_char
643
+ alignment = Alignment.new
644
+ alignment.gap_char = "+"
645
+ assert_equal("+", alignment.gap_char)
646
+ end
647
+
648
+ # Alignment#gap_regexp
649
+
650
+ def test_default_gap_regexp_matches_default_gap_char
651
+ alignment = Alignment.new
652
+ assert(alignment.gap_regexp.match(alignment.gap_char))
653
+ end
654
+
655
+ # Alignment#missing_char
656
+
657
+ def test_default_missing_char
658
+ alignment = Alignment.new
659
+ assert_equal("?", alignment.missing_char)
660
+ end
661
+
662
+ # Alignment#seq_length
663
+
664
+ def test_seq_length_when_one_sequence
665
+ alignment = build_na_alignment("agt")
666
+ assert_equal(3, alignment.seq_length)
667
+ end
668
+
669
+ def test_seq_length_is_max_seq_length
670
+ alignment = build_na_alignment("agt", "agtaa", "agta")
671
+ assert_equal(5, alignment.seq_length)
672
+ end
673
+
674
+ # Alignment#each_site
675
+
676
+ def test_each_site_equal_length
677
+ alignment = build_na_alignment("acg", "gta")
678
+ expected_sites = [["a", "g"], ["c", "t"], ["g", "a"]]
679
+ alignment.each_site do |site|
680
+ assert_equal expected_sites.shift, site, "site ##{3-expected_sites.size} wrong"
681
+ end
682
+ end
683
+
684
+ def test_each_site_unequal_length
685
+ alignment = build_na_alignment("ac", "gta")
686
+ expected_sites = [["a", "g"], ["c", "t"], ["-", "a"]]
687
+ alignment.each_site do |site|
688
+ assert_equal expected_sites.shift, site, "site ##{3-expected_sites.size} wrong"
689
+ end
690
+ end
691
+
692
+ #TODO: Lots of stuff needing tests here
693
+
694
+ # Alignment#add_seq
695
+
696
+ def test_add_seq_no_key
697
+ alignment = Alignment.new
698
+ alignment.add_seq("agct")
699
+ assert_equal(String, alignment.seqclass, "wrong class")
700
+ assert_equal({0=>"agct"}, alignment.to_hash, "wrong hash")
701
+ end
702
+
703
+ def test_add_seq_using_seq_with_seq_method
704
+ seq = "agtc"
705
+ class <<seq
706
+ def seq
707
+ Sequence::NA.new(self)
708
+ end
709
+ end
710
+
711
+ alignment = Alignment.new
712
+ alignment.add_seq(seq, "key")
713
+ assert_equal(Sequence::NA, alignment.seqclass, "wrong class")
714
+ assert_equal({"key"=>"agtc"}, alignment.to_hash, "wrong hash")
715
+ end
716
+
717
+ def test_add_seq_using_seq_with_naseq_method
718
+ seq = "agtc"
719
+ class <<seq
720
+ def naseq
721
+ Sequence::NA.new(self)
722
+ end
723
+ end
724
+
725
+ alignment = Alignment.new
726
+ alignment.add_seq(seq, "key")
727
+ assert_equal(Sequence::NA, alignment.seqclass, "wrong class")
728
+ assert_equal({"key"=>"agtc"}, alignment.to_hash, "wrong hash")
729
+ end
730
+
731
+ def test_add_seq_using_seq_with_aaseq_method
732
+ seq = "AVGR"
733
+ class <<seq
734
+ def aaseq
735
+ Sequence::AA.new(self)
736
+ end
737
+ end
738
+
739
+ alignment = Alignment.new
740
+ alignment.add_seq(seq, "key")
741
+ assert_equal(Sequence::AA, alignment.seqclass, "wrong class")
742
+ assert_equal({"key"=>"AVGR"}, alignment.to_hash, "wrong hash")
743
+ end
744
+
745
+ def test_add_seq_using_seq_with_definition_method
746
+ seq = "atgc"
747
+ class <<seq
748
+ def definition
749
+ "this is the key"
750
+ end
751
+ end
752
+
753
+ alignment = Alignment.new
754
+ alignment.add_seq(seq)
755
+ assert_equal({"this is the key"=>"atgc"}, alignment.to_hash, "wrong hash")
756
+ end
757
+
758
+ def test_add_seq_using_seq_with_entry_id_method
759
+ seq = "atgc"
760
+ class <<seq
761
+ def entry_id
762
+ 271828
763
+ end
764
+ end
765
+
766
+ alignment = Alignment.new
767
+ alignment.add_seq(seq)
768
+ assert_equal({271828=>"atgc"}, alignment.to_hash, "wrong hash")
769
+ end
770
+
771
+ # Alignment#consensus_string
772
+
773
+ def test_consensus_string_no_gaps
774
+ alignment = build_na_alignment("agtcgattaa",
775
+ "tttcgatgcc")
776
+ assert_equal("??tcgat???", alignment.consensus_string)
777
+ end
778
+
779
+ def test_consensus_threshold_two_sequences
780
+ alignment = build_na_alignment("agtcgattaa",
781
+ "tttcgatgcc")
782
+ # the threshold is the fraction of sequences in which a symbol must
783
+ # occur at a given position to be considered the consensus symbol
784
+ assert_equal("agtcgattaa", alignment.consensus(0.5))
785
+ assert_equal("??tcgat???", alignment.consensus(0.500000001))
786
+ end
787
+
788
+ def test_consensus_threshold_four_sequences
789
+ alignment = build_na_alignment("agtg",
790
+ "ttag",
791
+ "actc",
792
+ "tatc")
793
+ # ties go to the symbol that occurs in the earliest sequence
794
+ assert_equal("agtg", alignment.consensus(0.25))
795
+ assert_equal("a?tg", alignment.consensus(0.26))
796
+ end
797
+
798
+ def test_consensus_opt_gap_mode
799
+ alignment = build_na_alignment("gt-gt-a",
800
+ "ttcggc-",
801
+ "ttcggc-")
802
+ # using threshold = 0.5, that is a symbol must occur >= half the time in order to be consensus
803
+ # gap_mode -1 means gaps are ignored
804
+ assert_equal("ttcggca", alignment.consensus(0.5, :gap_mode => -1), "gap mode -1")
805
+ # gap_mode 0 means gaps are treated like regular symbols, yielding a gap in the last position
806
+ assert_equal("ttcggc-", alignment.consensus(0.5, :gap_mode => 0), "gap mode 0")
807
+ # gap_mode 1 means gaps take precedence over any other symbol, yielding two more gaps
808
+ assert_equal("tt-gg--", alignment.consensus(0.5, :gap_mode => 1), "gap mode 1")
809
+ end
810
+
811
+ def test_consensus_opt_missing_char
812
+ alignment = build_na_alignment("agtcgattaa",
813
+ "tttcgatgcc")
814
+ assert_equal("**tcgat***", alignment.consensus(1, :missing_char => "*"))
815
+ end
816
+
817
+ # Alignment#consensus_iupac
818
+
819
+ def test_consensus_iupac_no_gaps
820
+ alignment = build_na_alignment("agtcgattaa", "tttcgatgcc")
821
+ assert_equal("wktcgatkmm", alignment.consensus_iupac)
822
+ end
823
+
824
+ def test_consensus_iupac_of_ambiguous_bases
825
+ alignment = build_na_alignment("tmrwsykvhdbnd", "uaaaccgaaacab")
826
+ assert_equal("tmrwsykvhdbnn", alignment.consensus_iupac)
827
+ end
828
+
829
+ def test_consensus_iupac_gap_modes
830
+ alignment = build_na_alignment("a-t", "acc")
831
+ # gap_mode -1 means gaps are ignored
832
+ assert_equal("acy", alignment.consensus_iupac(:gap_mode => -1))
833
+ # gap_mode 0 means gaps are treated as normal characters, yielding a missing symbol
834
+ assert_equal("a?y", alignment.consensus_iupac(:gap_mode => 0))
835
+ # gap_mode 1 means gaps take precedence over everything, yielding a gap
836
+ assert_equal("a-y", alignment.consensus_iupac(:gap_mode => 1))
837
+ end
838
+
839
+ def test_consensus_iupac_yields_correct_ambiguous_bases
840
+ assert_equal "t", build_na_alignment("t", "u").consensus_iupac # not really IUPAC
841
+
842
+ # m = a c
843
+ assert_equal "m", build_na_alignment("a", "c").consensus_iupac, "m #1"
844
+ assert_equal "m", build_na_alignment("m", "c").consensus_iupac, "m #2"
845
+ assert_equal "m", build_na_alignment("a", "m").consensus_iupac, "m #3"
846
+ assert_equal "m", build_na_alignment("m", "a", "c").consensus_iupac, "m #4"
847
+
848
+ # r = a g
849
+ assert_equal "r", build_na_alignment("a", "g").consensus_iupac, "r #1"
850
+ assert_equal "r", build_na_alignment("r", "g").consensus_iupac, "r #2"
851
+ assert_equal "r", build_na_alignment("a", "r").consensus_iupac, "r #3"
852
+ assert_equal "r", build_na_alignment("a", "r", "g").consensus_iupac, "r #4"
853
+
854
+ # w = a t/u
855
+ assert_equal "w", build_na_alignment("a", "t").consensus_iupac, "w #1"
856
+ assert_equal "w", build_na_alignment("a", "u").consensus_iupac, "w #2"
857
+ assert_equal "w", build_na_alignment("w", "a").consensus_iupac, "w #3"
858
+ assert_equal "w", build_na_alignment("t", "w").consensus_iupac, "w #4"
859
+ assert_equal "w", build_na_alignment("w", "u").consensus_iupac, "w #5"
860
+ assert_equal "w", build_na_alignment("u", "t", "a").consensus_iupac, "w #6"
861
+ assert_equal "w", build_na_alignment("w", "u", "t", "a").consensus_iupac, "w #7"
862
+
863
+ # s = c g
864
+ assert_equal "s", build_na_alignment("c", "g").consensus_iupac, "s #1"
865
+ assert_equal "s", build_na_alignment("s", "g").consensus_iupac, "s #2"
866
+ assert_equal "s", build_na_alignment("c", "s").consensus_iupac, "s #3"
867
+ assert_equal "s", build_na_alignment("c", "s", "g").consensus_iupac, "s #4"
868
+
869
+ # y = c t/u
870
+ assert_equal "y", build_na_alignment("c", "t").consensus_iupac, "y #1"
871
+ assert_equal "y", build_na_alignment("c", "u").consensus_iupac, "y #2"
872
+ assert_equal "y", build_na_alignment("y", "c").consensus_iupac, "y #3"
873
+ assert_equal "y", build_na_alignment("t", "y").consensus_iupac, "y #4"
874
+ assert_equal "y", build_na_alignment("y", "u").consensus_iupac, "y #5"
875
+ assert_equal "y", build_na_alignment("u", "t", "c").consensus_iupac, "y #6"
876
+ assert_equal "y", build_na_alignment("y", "u", "t", "c").consensus_iupac, "y #7"
877
+
878
+ # k = g t/u
879
+ assert_equal "k", build_na_alignment("g", "t").consensus_iupac, "k #1"
880
+ assert_equal "k", build_na_alignment("g", "u").consensus_iupac, "k #2"
881
+ assert_equal "k", build_na_alignment("k", "g").consensus_iupac, "k #3"
882
+ assert_equal "k", build_na_alignment("t", "k").consensus_iupac, "k #4"
883
+ assert_equal "k", build_na_alignment("k", "u").consensus_iupac, "k #5"
884
+ assert_equal "k", build_na_alignment("u", "t", "g").consensus_iupac, "k #6"
885
+ assert_equal "k", build_na_alignment("k", "u", "t", "g").consensus_iupac, "k #7"
886
+
887
+ # v = a c g m r s
888
+ assert_equal "v", build_na_alignment("a", "c", "g").consensus_iupac, "v #1"
889
+ assert_equal "v", build_na_alignment("g", "m").consensus_iupac, "v #2"
890
+ assert_equal "v", build_na_alignment("a", "s").consensus_iupac, "v #3"
891
+ assert_equal "v", build_na_alignment("c", "r").consensus_iupac, "v #4"
892
+ assert_equal "v", build_na_alignment("m", "s").consensus_iupac, "v #5"
893
+ assert_equal "v", build_na_alignment("m", "r").consensus_iupac, "v #6"
894
+ assert_equal "v", build_na_alignment("s", "r").consensus_iupac, "v #7"
895
+ assert_equal "v", build_na_alignment("s", "r", "m").consensus_iupac, "v #8"
896
+ assert_equal "v", build_na_alignment("s", "r", "m", "a", "c", "g").consensus_iupac, "v #9"
897
+ assert_equal "v", build_na_alignment("v", "g").consensus_iupac, "v #10" # alright, enough
898
+
899
+ # b = t/u c g s y k
900
+ assert_equal "b", build_na_alignment("t", "c", "g").consensus_iupac, "b #1"
901
+ assert_equal "b", build_na_alignment("g", "y").consensus_iupac, "b #2"
902
+ assert_equal "b", build_na_alignment("t", "s").consensus_iupac, "b #3"
903
+ assert_equal "b", build_na_alignment("c", "k").consensus_iupac, "b #4"
904
+ assert_equal "b", build_na_alignment("y", "s").consensus_iupac, "b #5"
905
+ assert_equal "b", build_na_alignment("y", "k").consensus_iupac, "b #6"
906
+ assert_equal "b", build_na_alignment("s", "k").consensus_iupac, "b #7"
907
+ assert_equal "b", build_na_alignment("s", "k", "y").consensus_iupac, "b #8"
908
+ assert_equal "b", build_na_alignment("s", "k", "y", "u", "c", "g").consensus_iupac, "b #9"
909
+ assert_equal "b", build_na_alignment("b", "g").consensus_iupac, "b #10"
910
+
911
+ # h = t/u c a y w m
912
+ assert_equal "h", build_na_alignment("t", "c", "a").consensus_iupac, "h #1"
913
+ assert_equal "h", build_na_alignment("a", "y").consensus_iupac, "h #2"
914
+ assert_equal "h", build_na_alignment("c", "w").consensus_iupac, "h #3"
915
+ assert_equal "h", build_na_alignment("u", "m").consensus_iupac, "h #4"
916
+ assert_equal "h", build_na_alignment("y", "w").consensus_iupac, "h #5"
917
+ assert_equal "h", build_na_alignment("y", "m").consensus_iupac, "h #6"
918
+ assert_equal "h", build_na_alignment("y", "w").consensus_iupac, "h #7"
919
+ assert_equal "h", build_na_alignment("w", "m", "y").consensus_iupac, "h #8"
920
+ assert_equal "h", build_na_alignment("w", "m", "y", "t", "c", "a").consensus_iupac, "h #9"
921
+ assert_equal "h", build_na_alignment("h", "t").consensus_iupac, "h #10"
922
+
923
+ # d = t/u g a r w k
924
+ assert_equal "d", build_na_alignment("t", "g", "a").consensus_iupac, "d #1"
925
+ assert_equal "d", build_na_alignment("r", "t").consensus_iupac, "d #2"
926
+ assert_equal "d", build_na_alignment("w", "g").consensus_iupac, "d #3"
927
+ assert_equal "d", build_na_alignment("k", "a").consensus_iupac, "d #4"
928
+ assert_equal "d", build_na_alignment("k", "r").consensus_iupac, "d #5"
929
+ assert_equal "d", build_na_alignment("k", "w").consensus_iupac, "d #6"
930
+ assert_equal "d", build_na_alignment("r", "w").consensus_iupac, "d #7"
931
+ assert_equal "d", build_na_alignment("r", "w", "k").consensus_iupac, "d #8"
932
+ assert_equal "d", build_na_alignment("k", "r", "w", "t", "g", "a").consensus_iupac, "d #9"
933
+ assert_equal "d", build_na_alignment("d", "t").consensus_iupac, "d #10"
934
+
935
+ # n = anything
936
+ assert_equal "n", build_na_alignment("a", "g", "c", "t").consensus_iupac, "n #1"
937
+ assert_equal "n", build_na_alignment("a", "g", "c", "u").consensus_iupac, "n #2"
938
+ assert_equal "n", build_na_alignment("w", "s").consensus_iupac, "n #3"
939
+ assert_equal "n", build_na_alignment("k", "m").consensus_iupac, "n #4"
940
+ assert_equal "n", build_na_alignment("r", "y").consensus_iupac, "n #5"
941
+ end
942
+
943
+ def test_consensus_iupac_missing_char
944
+ alignment = build_na_alignment("a??", "ac?")
945
+ assert_equal("a??", alignment.consensus_iupac())
946
+ end
947
+
948
+ def test_consensus_iupac_missing_char_option
949
+ alignment = build_na_alignment("a**t", "ac**")
950
+ assert_equal("a***", alignment.consensus_iupac(:missing_char => "*"))
951
+ end
952
+
953
+ # Alignment#convert_match
954
+
955
+ def test_convert_match
956
+ alignment = Alignment.new
957
+ alignment << Sequence::NA.new("agtcgattaa")
958
+ alignment << Sequence::NA.new("tttcgatgcc")
959
+ match = alignment.convert_match
960
+ assert_equal(alignment[0], match[0], "first sequence altered")
961
+ assert_equal("tt.....gcc", match[1], "wrong match")
962
+ end
963
+
964
+ # Alignment#convert_unmatch
965
+
966
+ def test_convert_unmatch
967
+ alignment = Alignment.new
968
+ alignment << Sequence::NA.new("agtcgattaa")
969
+ alignment << Sequence::NA.new("tt.....gcc")
970
+ unmatched = alignment.convert_unmatch
971
+ assert_equal("agtcgattaa", unmatched[0], "first changed")
972
+ assert_equal("tttcgatgcc", unmatched[1], "second wrong")
973
+ end
974
+
975
+ def test_convert_unmatch_multiple_sequences
976
+ alignment = Alignment.new
977
+ alignment << Sequence::NA.new("agtcgattaa")
978
+ alignment << Sequence::NA.new("tt.....gcc")
979
+ alignment << Sequence::NA.new("c...c..g.c")
980
+ unmatched = alignment.convert_unmatch
981
+ assert_equal("agtcgattaa", unmatched[0], "first changed")
982
+ assert_equal("tttcgatgcc", unmatched[1], "second wrong")
983
+ assert_equal("cgtccatgac", unmatched[2], "third wrong")
984
+ end
985
+
986
+ def test_convert_unmatch_different_length_sequences_truncates_seq_if_last_matched
987
+ alignment = Alignment.new
988
+ alignment << Sequence::NA.new("agtcgatta")
989
+ alignment << Sequence::NA.new("tt.....gc.")
990
+ unmatched = alignment.convert_unmatch
991
+ assert_equal("agtcgatta", unmatched[0], "first changed")
992
+ assert_equal("tttcgatgc", unmatched[1], "second wrong") #TODO: verify this is correct, and not . at end
993
+ end
994
+
995
+ def test_convert_unmatch_different_match_char
996
+ alignment = Alignment.new
997
+ alignment << Sequence::NA.new("agtcga")
998
+ alignment << Sequence::NA.new("tt====")
999
+ unmatched = alignment.convert_unmatch('=')
1000
+ assert_equal("agtcga", unmatched[0], "first changed")
1001
+ assert_equal("tttcga", unmatched[1], "second wrong")
1002
+ end
1003
+
1004
+ # Alignment#match_line
1005
+
1006
+ def test_match_line_protein
1007
+ alignment = Alignment.new
1008
+ alignment << Sequence::AA.new("AELFMCF")
1009
+ alignment << Sequence::AA.new("AKLVNNF")
1010
+ assert_equal "*:*. *", alignment.match_line
1011
+ end
1012
+
1013
+ #TODO: lots more on the consensus, match, etc.
1014
+
1015
+ # Alignment#normalize
1016
+
1017
+ def test_normalizebang_extends_sequences_with_gaps
1018
+ alignment = build_na_alignment("a", "ag", "agc", "agct")
1019
+ alignment.normalize!
1020
+ assert_equal({0=>"a---",1=>"ag--",2=>"agc-",3=>"agct"}, alignment.to_hash)
1021
+ end
1022
+
1023
+ # Alignment#to_clustal
1024
+ end
1025
+ end