finishm 0.0.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (554) hide show
  1. checksums.yaml +7 -0
  2. data/.document +5 -0
  3. data/.gitmodules +3 -0
  4. data/.rspec +1 -0
  5. data/Gemfile +31 -0
  6. data/LICENSE.txt +20 -0
  7. data/README.md +59 -0
  8. data/Rakefile +51 -0
  9. data/VERSION +1 -0
  10. data/bin/assembly_visualiser +106 -0
  11. data/bin/check_primer_combinations.rb +73 -0
  12. data/bin/contig_joiner.rb +244 -0
  13. data/bin/contigs_against_assembly.rb +153 -0
  14. data/bin/finishm +143 -0
  15. data/bin/finishm_assembler +55 -0
  16. data/bin/finishm_gap_closer.rb +241 -0
  17. data/bin/kmer_abundance_file_tool.rb +49 -0
  18. data/bin/kmer_pattern_to_assembly.rb +377 -0
  19. data/bin/kmer_profile_finder.rb +92 -0
  20. data/bin/kmers_count_parse.d +52 -0
  21. data/bin/kmers_count_tabulate.d +123 -0
  22. data/bin/kmers_count_tabulate.rb +84 -0
  23. data/bin/pcr_result_parser.rb +108 -0
  24. data/bin/primer_finder.rb +119 -0
  25. data/bin/read_selection_by_kmer.d +174 -0
  26. data/bin/scaffold_by_pattern.rb +119 -0
  27. data/bin/scaffold_connection_possibilities_to_knowns.rb +193 -0
  28. data/bin/scaffold_end_coverages.rb +69 -0
  29. data/bin/trail_validator.rb +84 -0
  30. data/ext/mkrf_conf.rb +56 -0
  31. data/ext/src/Makefile +140 -0
  32. data/ext/src/src/allocArray.c +305 -0
  33. data/ext/src/src/allocArray.h +86 -0
  34. data/ext/src/src/autoOpen.c +107 -0
  35. data/ext/src/src/autoOpen.h +18 -0
  36. data/ext/src/src/binarySequences.c +813 -0
  37. data/ext/src/src/binarySequences.h +125 -0
  38. data/ext/src/src/concatenatedGraph.c +233 -0
  39. data/ext/src/src/concatenatedGraph.h +30 -0
  40. data/ext/src/src/concatenatedPreGraph.c +262 -0
  41. data/ext/src/src/concatenatedPreGraph.h +29 -0
  42. data/ext/src/src/correctedGraph.c +2643 -0
  43. data/ext/src/src/correctedGraph.h +32 -0
  44. data/ext/src/src/dfib.c +509 -0
  45. data/ext/src/src/dfib.h +69 -0
  46. data/ext/src/src/dfibHeap.c +89 -0
  47. data/ext/src/src/dfibHeap.h +39 -0
  48. data/ext/src/src/dfibpriv.h +105 -0
  49. data/ext/src/src/fib.c +628 -0
  50. data/ext/src/src/fib.h +78 -0
  51. data/ext/src/src/fibHeap.c +79 -0
  52. data/ext/src/src/fibHeap.h +41 -0
  53. data/ext/src/src/fibpriv.h +110 -0
  54. data/ext/src/src/globals.h +154 -0
  55. data/ext/src/src/graph.c +3932 -0
  56. data/ext/src/src/graph.h +233 -0
  57. data/ext/src/src/graphReConstruction.c +1472 -0
  58. data/ext/src/src/graphReConstruction.h +30 -0
  59. data/ext/src/src/graphStats.c +2167 -0
  60. data/ext/src/src/graphStats.h +72 -0
  61. data/ext/src/src/graphStructures.h +52 -0
  62. data/ext/src/src/kmer.c +652 -0
  63. data/ext/src/src/kmer.h +73 -0
  64. data/ext/src/src/kmerOccurenceTable.c +236 -0
  65. data/ext/src/src/kmerOccurenceTable.h +44 -0
  66. data/ext/src/src/kseq.h +223 -0
  67. data/ext/src/src/locallyCorrectedGraph.c +557 -0
  68. data/ext/src/src/locallyCorrectedGraph.h +40 -0
  69. data/ext/src/src/passageMarker.c +677 -0
  70. data/ext/src/src/passageMarker.h +137 -0
  71. data/ext/src/src/preGraph.c +1717 -0
  72. data/ext/src/src/preGraph.h +106 -0
  73. data/ext/src/src/preGraphConstruction.c +990 -0
  74. data/ext/src/src/preGraphConstruction.h +26 -0
  75. data/ext/src/src/probe_node_finder.c +84 -0
  76. data/ext/src/src/probe_node_finder.h +6 -0
  77. data/ext/src/src/readCoherentGraph.c +557 -0
  78. data/ext/src/src/readCoherentGraph.h +30 -0
  79. data/ext/src/src/readSet.c +1734 -0
  80. data/ext/src/src/readSet.h +67 -0
  81. data/ext/src/src/readToNode.c +218 -0
  82. data/ext/src/src/readToNode.h +35 -0
  83. data/ext/src/src/recycleBin.c +199 -0
  84. data/ext/src/src/recycleBin.h +58 -0
  85. data/ext/src/src/roadMap.c +342 -0
  86. data/ext/src/src/roadMap.h +65 -0
  87. data/ext/src/src/run.c +318 -0
  88. data/ext/src/src/run.h +52 -0
  89. data/ext/src/src/run2.c +744 -0
  90. data/ext/src/src/runReadToNode.c +29 -0
  91. data/ext/src/src/scaffold.c +1876 -0
  92. data/ext/src/src/scaffold.h +64 -0
  93. data/ext/src/src/shortReadPairs.c +1243 -0
  94. data/ext/src/src/shortReadPairs.h +32 -0
  95. data/ext/src/src/splay.c +259 -0
  96. data/ext/src/src/splay.h +43 -0
  97. data/ext/src/src/splayTable.c +1315 -0
  98. data/ext/src/src/splayTable.h +31 -0
  99. data/ext/src/src/tightString.c +362 -0
  100. data/ext/src/src/tightString.h +82 -0
  101. data/ext/src/src/utility.c +199 -0
  102. data/ext/src/src/utility.h +98 -0
  103. data/ext/src/third-party/zlib-1.2.3/ChangeLog +855 -0
  104. data/ext/src/third-party/zlib-1.2.3/FAQ +339 -0
  105. data/ext/src/third-party/zlib-1.2.3/INDEX +51 -0
  106. data/ext/src/third-party/zlib-1.2.3/Makefile +154 -0
  107. data/ext/src/third-party/zlib-1.2.3/Makefile.in +154 -0
  108. data/ext/src/third-party/zlib-1.2.3/README +125 -0
  109. data/ext/src/third-party/zlib-1.2.3/adler32.c +149 -0
  110. data/ext/src/third-party/zlib-1.2.3/adler32.o +0 -0
  111. data/ext/src/third-party/zlib-1.2.3/algorithm.txt +209 -0
  112. data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.pup +66 -0
  113. data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.sas +65 -0
  114. data/ext/src/third-party/zlib-1.2.3/as400/bndsrc +132 -0
  115. data/ext/src/third-party/zlib-1.2.3/as400/compile.clp +123 -0
  116. data/ext/src/third-party/zlib-1.2.3/as400/readme.txt +111 -0
  117. data/ext/src/third-party/zlib-1.2.3/as400/zlib.inc +331 -0
  118. data/ext/src/third-party/zlib-1.2.3/compress.c +79 -0
  119. data/ext/src/third-party/zlib-1.2.3/compress.o +0 -0
  120. data/ext/src/third-party/zlib-1.2.3/configure +459 -0
  121. data/ext/src/third-party/zlib-1.2.3/contrib/README.contrib +71 -0
  122. data/ext/src/third-party/zlib-1.2.3/contrib/ada/buffer_demo.adb +106 -0
  123. data/ext/src/third-party/zlib-1.2.3/contrib/ada/mtest.adb +156 -0
  124. data/ext/src/third-party/zlib-1.2.3/contrib/ada/read.adb +156 -0
  125. data/ext/src/third-party/zlib-1.2.3/contrib/ada/readme.txt +65 -0
  126. data/ext/src/third-party/zlib-1.2.3/contrib/ada/test.adb +463 -0
  127. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.adb +225 -0
  128. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.ads +114 -0
  129. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.adb +141 -0
  130. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.ads +450 -0
  131. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.adb +701 -0
  132. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.ads +328 -0
  133. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.gpr +20 -0
  134. data/ext/src/third-party/zlib-1.2.3/contrib/asm586/README.586 +43 -0
  135. data/ext/src/third-party/zlib-1.2.3/contrib/asm586/match.S +364 -0
  136. data/ext/src/third-party/zlib-1.2.3/contrib/asm686/README.686 +34 -0
  137. data/ext/src/third-party/zlib-1.2.3/contrib/asm686/match.S +329 -0
  138. data/ext/src/third-party/zlib-1.2.3/contrib/blast/Makefile +8 -0
  139. data/ext/src/third-party/zlib-1.2.3/contrib/blast/README +4 -0
  140. data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.c +444 -0
  141. data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.h +71 -0
  142. data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.pk +0 -0
  143. data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.txt +1 -0
  144. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLib.pas +557 -0
  145. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLibConst.pas +11 -0
  146. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/readme.txt +76 -0
  147. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/zlibd32.mak +93 -0
  148. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.build +33 -0
  149. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.chm +0 -0
  150. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.sln +21 -0
  151. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/AssemblyInfo.cs +58 -0
  152. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/ChecksumImpl.cs +202 -0
  153. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CircularBuffer.cs +83 -0
  154. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CodecBase.cs +198 -0
  155. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Deflater.cs +106 -0
  156. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.cs +288 -0
  157. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.csproj +141 -0
  158. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/GZipStream.cs +301 -0
  159. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Inflater.cs +105 -0
  160. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/UnitTests.cs +274 -0
  161. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/LICENSE_1_0.txt +23 -0
  162. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/readme.txt +58 -0
  163. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/README +1 -0
  164. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.c +608 -0
  165. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.h +37 -0
  166. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inffix9.h +107 -0
  167. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inflate9.h +47 -0
  168. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.c +323 -0
  169. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.h +55 -0
  170. data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffas86.c +1157 -0
  171. data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffast.S +1368 -0
  172. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/test.cpp +24 -0
  173. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.cpp +329 -0
  174. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.h +128 -0
  175. data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream.h +307 -0
  176. data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream_test.cpp +25 -0
  177. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/README +35 -0
  178. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/TODO +17 -0
  179. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/test.cc +50 -0
  180. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.cc +479 -0
  181. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.h +466 -0
  182. data/ext/src/third-party/zlib-1.2.3/contrib/masm686/match.asm +413 -0
  183. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/bld_ml64.bat +2 -0
  184. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.asm +513 -0
  185. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.obj +0 -0
  186. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffas8664.c +186 -0
  187. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.asm +392 -0
  188. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.obj +0 -0
  189. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/readme.txt +28 -0
  190. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/bld_ml32.bat +2 -0
  191. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.asm +972 -0
  192. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.obj +0 -0
  193. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32c.c +62 -0
  194. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.asm +1083 -0
  195. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.obj +0 -0
  196. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/mkasm.bat +3 -0
  197. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/readme.txt +21 -0
  198. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ChangeLogUnzip +67 -0
  199. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/Makefile +25 -0
  200. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/crypt.h +132 -0
  201. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.c +177 -0
  202. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.h +75 -0
  203. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.c +270 -0
  204. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.h +21 -0
  205. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/miniunz.c +585 -0
  206. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/minizip.c +420 -0
  207. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.c +281 -0
  208. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.h +31 -0
  209. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.c +1598 -0
  210. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.h +354 -0
  211. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.c +1219 -0
  212. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.h +235 -0
  213. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/example.pas +599 -0
  214. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/readme.txt +76 -0
  215. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibd32.mak +93 -0
  216. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibpas.pas +236 -0
  217. data/ext/src/third-party/zlib-1.2.3/contrib/puff/Makefile +8 -0
  218. data/ext/src/third-party/zlib-1.2.3/contrib/puff/README +63 -0
  219. data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.c +837 -0
  220. data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.h +31 -0
  221. data/ext/src/third-party/zlib-1.2.3/contrib/puff/zeros.raw +0 -0
  222. data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.c +275 -0
  223. data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.txt +10 -0
  224. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile +14 -0
  225. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile.msc +17 -0
  226. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/untgz.c +674 -0
  227. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/readme.txt +73 -0
  228. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/miniunz.vcproj +126 -0
  229. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/minizip.vcproj +126 -0
  230. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/testzlib.vcproj +126 -0
  231. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlib.rc +32 -0
  232. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibstat.vcproj +246 -0
  233. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.def +92 -0
  234. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.sln +78 -0
  235. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.vcproj +445 -0
  236. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/miniunz.vcproj +566 -0
  237. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/minizip.vcproj +563 -0
  238. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlib.vcproj +948 -0
  239. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlibdll.vcproj +567 -0
  240. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlib.rc +32 -0
  241. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibstat.vcproj +870 -0
  242. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.def +92 -0
  243. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.sln +144 -0
  244. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.vcproj +1219 -0
  245. data/ext/src/third-party/zlib-1.2.3/crc32.c +423 -0
  246. data/ext/src/third-party/zlib-1.2.3/crc32.h +441 -0
  247. data/ext/src/third-party/zlib-1.2.3/crc32.o +0 -0
  248. data/ext/src/third-party/zlib-1.2.3/deflate.c +1736 -0
  249. data/ext/src/third-party/zlib-1.2.3/deflate.h +331 -0
  250. data/ext/src/third-party/zlib-1.2.3/deflate.o +0 -0
  251. data/ext/src/third-party/zlib-1.2.3/example +0 -0
  252. data/ext/src/third-party/zlib-1.2.3/example.c +565 -0
  253. data/ext/src/third-party/zlib-1.2.3/examples/README.examples +42 -0
  254. data/ext/src/third-party/zlib-1.2.3/examples/fitblk.c +233 -0
  255. data/ext/src/third-party/zlib-1.2.3/examples/gun.c +693 -0
  256. data/ext/src/third-party/zlib-1.2.3/examples/gzappend.c +500 -0
  257. data/ext/src/third-party/zlib-1.2.3/examples/gzjoin.c +448 -0
  258. data/ext/src/third-party/zlib-1.2.3/examples/gzlog.c +413 -0
  259. data/ext/src/third-party/zlib-1.2.3/examples/gzlog.h +58 -0
  260. data/ext/src/third-party/zlib-1.2.3/examples/zlib_how.html +523 -0
  261. data/ext/src/third-party/zlib-1.2.3/examples/zpipe.c +191 -0
  262. data/ext/src/third-party/zlib-1.2.3/examples/zran.c +404 -0
  263. data/ext/src/third-party/zlib-1.2.3/gzio.c +1026 -0
  264. data/ext/src/third-party/zlib-1.2.3/gzio.o +0 -0
  265. data/ext/src/third-party/zlib-1.2.3/infback.c +623 -0
  266. data/ext/src/third-party/zlib-1.2.3/infback.o +0 -0
  267. data/ext/src/third-party/zlib-1.2.3/inffast.c +318 -0
  268. data/ext/src/third-party/zlib-1.2.3/inffast.h +11 -0
  269. data/ext/src/third-party/zlib-1.2.3/inffast.o +0 -0
  270. data/ext/src/third-party/zlib-1.2.3/inffixed.h +94 -0
  271. data/ext/src/third-party/zlib-1.2.3/inflate.c +1368 -0
  272. data/ext/src/third-party/zlib-1.2.3/inflate.h +115 -0
  273. data/ext/src/third-party/zlib-1.2.3/inflate.o +0 -0
  274. data/ext/src/third-party/zlib-1.2.3/inftrees.c +329 -0
  275. data/ext/src/third-party/zlib-1.2.3/inftrees.h +55 -0
  276. data/ext/src/third-party/zlib-1.2.3/inftrees.o +0 -0
  277. data/ext/src/third-party/zlib-1.2.3/libz.a +0 -0
  278. data/ext/src/third-party/zlib-1.2.3/make_vms.com +461 -0
  279. data/ext/src/third-party/zlib-1.2.3/minigzip +0 -0
  280. data/ext/src/third-party/zlib-1.2.3/minigzip.c +322 -0
  281. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.bor +109 -0
  282. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.dj2 +104 -0
  283. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.emx +69 -0
  284. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.msc +106 -0
  285. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.tc +94 -0
  286. data/ext/src/third-party/zlib-1.2.3/old/Makefile.riscos +151 -0
  287. data/ext/src/third-party/zlib-1.2.3/old/README +3 -0
  288. data/ext/src/third-party/zlib-1.2.3/old/descrip.mms +48 -0
  289. data/ext/src/third-party/zlib-1.2.3/old/os2/Makefile.os2 +136 -0
  290. data/ext/src/third-party/zlib-1.2.3/old/os2/zlib.def +51 -0
  291. data/ext/src/third-party/zlib-1.2.3/old/visual-basic.txt +160 -0
  292. data/ext/src/third-party/zlib-1.2.3/old/zlib.html +971 -0
  293. data/ext/src/third-party/zlib-1.2.3/projects/README.projects +41 -0
  294. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/README.txt +73 -0
  295. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/example.dsp +278 -0
  296. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/minigzip.dsp +278 -0
  297. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsp +609 -0
  298. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsw +59 -0
  299. data/ext/src/third-party/zlib-1.2.3/qnx/package.qpg +141 -0
  300. data/ext/src/third-party/zlib-1.2.3/trees.c +1219 -0
  301. data/ext/src/third-party/zlib-1.2.3/trees.h +128 -0
  302. data/ext/src/third-party/zlib-1.2.3/trees.o +0 -0
  303. data/ext/src/third-party/zlib-1.2.3/uncompr.c +61 -0
  304. data/ext/src/third-party/zlib-1.2.3/uncompr.o +0 -0
  305. data/ext/src/third-party/zlib-1.2.3/win32/DLL_FAQ.txt +397 -0
  306. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.bor +107 -0
  307. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.emx +69 -0
  308. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.gcc +141 -0
  309. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.msc +126 -0
  310. data/ext/src/third-party/zlib-1.2.3/win32/VisualC.txt +3 -0
  311. data/ext/src/third-party/zlib-1.2.3/win32/zlib.def +60 -0
  312. data/ext/src/third-party/zlib-1.2.3/win32/zlib1.rc +39 -0
  313. data/ext/src/third-party/zlib-1.2.3/zconf.h +332 -0
  314. data/ext/src/third-party/zlib-1.2.3/zconf.in.h +332 -0
  315. data/ext/src/third-party/zlib-1.2.3/zlib.3 +159 -0
  316. data/ext/src/third-party/zlib-1.2.3/zlib.h +1357 -0
  317. data/ext/src/third-party/zlib-1.2.3/zutil.c +318 -0
  318. data/ext/src/third-party/zlib-1.2.3/zutil.h +269 -0
  319. data/ext/src/third-party/zlib-1.2.3/zutil.o +0 -0
  320. data/lib/assembly/a_b_visualiser.rb +169 -0
  321. data/lib/assembly/acyclic_connection_finder.rb +81 -0
  322. data/lib/assembly/all_orfs.rb +615 -0
  323. data/lib/assembly/bad_format_writer.rb +46 -0
  324. data/lib/assembly/bam_probe_read_selector.rb +48 -0
  325. data/lib/assembly/bubbly_assembler.rb +842 -0
  326. data/lib/assembly/c_probe_node_finder.rb +38 -0
  327. data/lib/assembly/connection_interpreter.rb +350 -0
  328. data/lib/assembly/contig_printer.rb +400 -0
  329. data/lib/assembly/coverage_based_graph_filter.rb +68 -0
  330. data/lib/assembly/depth_first_search.rb +63 -0
  331. data/lib/assembly/dijkstra.rb +216 -0
  332. data/lib/assembly/fluffer.rb +253 -0
  333. data/lib/assembly/graph_explorer.rb +85 -0
  334. data/lib/assembly/graph_generator.rb +315 -0
  335. data/lib/assembly/height_finder.rb +355 -0
  336. data/lib/assembly/hybrid_velvet_graph.rb +70 -0
  337. data/lib/assembly/input_genome.rb +182 -0
  338. data/lib/assembly/kmer_coverage_based_path_filter.rb +65 -0
  339. data/lib/assembly/node_finder.rb +171 -0
  340. data/lib/assembly/oriented_node_trail.rb +507 -0
  341. data/lib/assembly/paired_end_assembler.rb +53 -0
  342. data/lib/assembly/paired_end_neighbour_finder.rb +176 -0
  343. data/lib/assembly/probed_graph.rb +105 -0
  344. data/lib/assembly/read_input.rb +79 -0
  345. data/lib/assembly/read_to_node.rb +37 -0
  346. data/lib/assembly/scaffold_breaker.rb +126 -0
  347. data/lib/assembly/sequence_hasher.rb +71 -0
  348. data/lib/assembly/single_coherent_paths_between_nodes.rb +533 -0
  349. data/lib/assembly/single_coherent_wanderer.rb +261 -0
  350. data/lib/assembly/single_ended_assembler.rb +441 -0
  351. data/lib/assembly/velvet_c_binding.rb +54 -0
  352. data/lib/assembly/velvet_graph_sequence_extractor.rb +123 -0
  353. data/lib/external/VERSION +1 -0
  354. data/lib/finishm/assemble.rb +224 -0
  355. data/lib/finishm/explore.rb +217 -0
  356. data/lib/finishm/finisher.rb +303 -0
  357. data/lib/finishm/fluff.rb +122 -0
  358. data/lib/finishm/gapfiller.rb +325 -0
  359. data/lib/finishm/orfs_finder.rb +88 -0
  360. data/lib/finishm/path_counter.rb +90 -0
  361. data/lib/finishm/primers.rb +425 -0
  362. data/lib/finishm/primers_check.rb +176 -0
  363. data/lib/finishm/roundup.rb +344 -0
  364. data/lib/finishm/sequence.rb +142 -0
  365. data/lib/finishm/visualise.rb +430 -0
  366. data/lib/finishm/wander.rb +270 -0
  367. data/lib/kmer_abundance_pattern.rb +79 -0
  368. data/lib/kmer_multi_abundance_file.rb +48 -0
  369. data/lib/oligo_designer.rb +88 -0
  370. data/lib/priner.rb +66 -0
  371. data/spec/acyclic_connection_finder_spec.rb +551 -0
  372. data/spec/all_orfs_spec.rb +443 -0
  373. data/spec/assemble_spec.rb +186 -0
  374. data/spec/bubbly_assembler_spec.rb +707 -0
  375. data/spec/c_node_finder_spec.rb +58 -0
  376. data/spec/connection_interpreter_spec.rb +284 -0
  377. data/spec/contig_printer_spec.rb +291 -0
  378. data/spec/coverage_based_graph_filter_spec.rb +102 -0
  379. data/spec/data/6_3e4e5e6e.1vANME.bam +0 -0
  380. data/spec/data/6_3e4e5e6e.1vANME.bam.bai +0 -0
  381. data/spec/data/acyclic_connection_finder/1/probes.fa +5 -0
  382. data/spec/data/acyclic_connection_finder/1/random1.fa +2 -0
  383. data/spec/data/acyclic_connection_finder/1/random1.sammy.fa.gz +0 -0
  384. data/spec/data/acyclic_connection_finder/1/random2.fa +2 -0
  385. data/spec/data/acyclic_connection_finder/1/random2.sammy.fa.gz +0 -0
  386. data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.fa +39 -0
  387. data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.slightly_changed.fa +39 -0
  388. data/spec/data/assembly/1_simple_bubble_uneven_coverage/reads_combined.fa.gz +0 -0
  389. data/spec/data/assembly_visualiser/Contig_6_1_to_250.fa.kmers31 +220 -0
  390. data/spec/data/assembly_visualiser/Contig_7_1_to_250.fa.kmers31 +220 -0
  391. data/spec/data/assembly_visualiser/Graph +46 -0
  392. data/spec/data/assembly_visualiser/start_kmers1 +2 -0
  393. data/spec/data/bands.csv +1 -0
  394. data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq +0 -0
  395. data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq.names +544 -0
  396. data/spec/data/c_probe_node_finder/1/Graph2 +668 -0
  397. data/spec/data/c_probe_node_finder/1/LastGraph +668 -0
  398. data/spec/data/c_probe_node_finder/1/Log +756 -0
  399. data/spec/data/c_probe_node_finder/1/PreGraph +11 -0
  400. data/spec/data/c_probe_node_finder/1/Roadmaps +2009 -0
  401. data/spec/data/c_probe_node_finder/1/contigs.fa +29 -0
  402. data/spec/data/c_probe_node_finder/1/stats.txt +6 -0
  403. data/spec/data/contig_printer/1/HOWTO_RECREATE +17 -0
  404. data/spec/data/contig_printer/1/contigs.fa +4 -0
  405. data/spec/data/contig_printer/1/seq.fa +2408 -0
  406. data/spec/data/contig_printer/1/seq.fa.svg +153 -0
  407. data/spec/data/contig_printer/1/seq.fa.velvet/Graph2 +2953 -0
  408. data/spec/data/contig_printer/1/seq.fa.velvet/LastGraph +2953 -0
  409. data/spec/data/contig_printer/1/seq.fa.velvet/Log +21 -0
  410. data/spec/data/contig_printer/1/seq.fa.velvet/PreGraph +27 -0
  411. data/spec/data/contig_printer/1/seq.fa.velvet/Roadmaps +5182 -0
  412. data/spec/data/contig_printer/1/seq.fa.velvet/Sequences +3612 -0
  413. data/spec/data/contig_printer/1/seq.fa.velvet/contigs.fa +36 -0
  414. data/spec/data/contig_printer/1/seq.fa.velvet/stats.txt +14 -0
  415. data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam +0 -0
  416. data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam.bai +0 -0
  417. data/spec/data/contig_printer/1/seq.node12.fa +4 -0
  418. data/spec/data/contig_printer/1/seq1_1to550.fa +2 -0
  419. data/spec/data/contig_printer/1/seq2_1to550.fa +2 -0
  420. data/spec/data/contig_printer/1/seq2_1to550.fa.fai +1 -0
  421. data/spec/data/explore/1/2seqs.sammy.fa +12004 -0
  422. data/spec/data/explore/1/HOWTO_RECREATE.txt +6 -0
  423. data/spec/data/explore/1/a.fa +2 -0
  424. data/spec/data/explore/1/seq1_and_a.fa +3 -0
  425. data/spec/data/explore/1/seq2.fa +2 -0
  426. data/spec/data/fluff/1/2seqs.sammy.fa +12004 -0
  427. data/spec/data/fluff/1/HOWTO_RECREATE.txt +5 -0
  428. data/spec/data/fluff/1/seq1.fa +2 -0
  429. data/spec/data/fluff/1/seq2.fa +2 -0
  430. data/spec/data/gapfilling/1/reads.fa +171 -0
  431. data/spec/data/gapfilling/1/trail_with_Ns.fa +5 -0
  432. data/spec/data/gapfilling/1/velvetAssembly/Graph2 +130 -0
  433. data/spec/data/gapfilling/1/velvetAssembly/LastGraph +130 -0
  434. data/spec/data/gapfilling/1/velvetAssembly/Log +199 -0
  435. data/spec/data/gapfilling/1/velvetAssembly/PreGraph +7 -0
  436. data/spec/data/gapfilling/1/velvetAssembly/Roadmaps +239 -0
  437. data/spec/data/gapfilling/1/velvetAssembly/Sequences +281 -0
  438. data/spec/data/gapfilling/1/velvetAssembly/contigs.fa +12 -0
  439. data/spec/data/gapfilling/1/velvetAssembly/stats.txt +4 -0
  440. data/spec/data/gapfilling/2/HOWTO_recreate +17 -0
  441. data/spec/data/gapfilling/2/reference.fa +2 -0
  442. data/spec/data/gapfilling/2/reference_part1.fa +4 -0
  443. data/spec/data/gapfilling/2/reference_part2.fa +4 -0
  444. data/spec/data/gapfilling/2/sammy_reads.fa.gz +0 -0
  445. data/spec/data/gapfilling/2/with_gaps.fa +4 -0
  446. data/spec/data/gapfilling/3/HOWTO_recreate +4 -0
  447. data/spec/data/gapfilling/3/reads.fa.gz +0 -0
  448. data/spec/data/gapfilling/3/reference_part1.fa +4 -0
  449. data/spec/data/gapfilling/3/reference_part2.fa +4 -0
  450. data/spec/data/gapfilling/3/with_gaps.fa +4 -0
  451. data/spec/data/gapfilling/4/HOWTO_recreate +1 -0
  452. data/spec/data/gapfilling/4/reads.fa.gz +0 -0
  453. data/spec/data/gapfilling/5/HOWTO_RECREATE +7 -0
  454. data/spec/data/gapfilling/5/answer.fna +2 -0
  455. data/spec/data/gapfilling/5/gappy.fna +2 -0
  456. data/spec/data/gapfilling/5/reads.fa +17961 -0
  457. data/spec/data/gapfilling/5/velvet51_3.5/LastGraph +8337 -0
  458. data/spec/data/gapfilling/5/velvet51_3.5/Sequences +20921 -0
  459. data/spec/data/gapfilling/6/random1.fa +28 -0
  460. data/spec/data/gapfilling/6/random2.fa +28 -0
  461. data/spec/data/gapfilling/6/random_sequence_length_2000 +0 -0
  462. data/spec/data/gapfilling/6/reads.random1.fa.gz +0 -0
  463. data/spec/data/gapfilling/6/reads.random2.fa.gz +0 -0
  464. data/spec/data/gapfilling/6/to_gapfill.fa +22 -0
  465. data/spec/data/kmer_profile_to_assembly/multiple_abundance_file1.csv +2 -0
  466. data/spec/data/kmers_count1.csv +2 -0
  467. data/spec/data/kmers_count2.csv +3 -0
  468. data/spec/data/out +3 -0
  469. data/spec/data/positive_latching_pair.fa +2 -0
  470. data/spec/data/primers.csv +4 -0
  471. data/spec/data/read_selection_by_kmer/blacklist1.txt +1 -0
  472. data/spec/data/read_selection_by_kmer/input.fasta +6 -0
  473. data/spec/data/read_selection_by_kmer/whitelist1.txt +1 -0
  474. data/spec/data/read_selection_by_kmer/whitelist2.txt +2 -0
  475. data/spec/data/read_to_node/1_a_graph/HOWTO_RECREATE.txt +2 -0
  476. data/spec/data/read_to_node/1_a_graph/LastGraph +6695 -0
  477. data/spec/data/read_to_node/1_a_graph/ReadToNode.bin +0 -0
  478. data/spec/data/read_to_node/2_no_read256_or_259/HOWTO_RECREATE.txt +3 -0
  479. data/spec/data/read_to_node/2_no_read256_or_259/LastGraph +6693 -0
  480. data/spec/data/read_to_node/2_no_read256_or_259/ReadToNode.bin +0 -0
  481. data/spec/data/read_to_node/3_no_last_read/LastGraph +6694 -0
  482. data/spec/data/read_to_node/3_no_last_read/ReadToNode.bin +0 -0
  483. data/spec/data/t/details.txt +5 -0
  484. data/spec/data/t/details.txt.srt +5 -0
  485. data/spec/data/t/location.txt +3 -0
  486. data/spec/data/t/location.txt.srt +3 -0
  487. data/spec/data/tweak/1_gap_then_unscaffolded/answer.fa +2 -0
  488. data/spec/data/tweak/1_gap_then_unscaffolded/reads.fa.gz +0 -0
  489. data/spec/data/tweak/1_gap_then_unscaffolded/scaffolds.fa +6 -0
  490. data/spec/data/tweak/2_second_genome/answer2.fa +2 -0
  491. data/spec/data/tweak/2_second_genome/reads.fa.gz +0 -0
  492. data/spec/data/tweak/3_variant/answer.fa +2 -0
  493. data/spec/data/tweak/3_variant/lesser_answer.fa +2 -0
  494. data/spec/data/tweak/3_variant/reads.fa.gz +0 -0
  495. data/spec/data/tweak/3_variant/with_gaps.fa +2 -0
  496. data/spec/data/velvet_test_trails/Assem/Graph +17 -0
  497. data/spec/data/velvet_test_trails/Assem/Graph2 +40 -0
  498. data/spec/data/velvet_test_trails/Assem/LastGraph +40 -0
  499. data/spec/data/velvet_test_trails/Assem/Log +35 -0
  500. data/spec/data/velvet_test_trails/Assem/PreGraph +9 -0
  501. data/spec/data/velvet_test_trails/Assem/Roadmaps +89 -0
  502. data/spec/data/velvet_test_trails/Assem/Sequences +50 -0
  503. data/spec/data/velvet_test_trails/Assem/a.svg +53 -0
  504. data/spec/data/velvet_test_trails/Assem/contigs.fa +15 -0
  505. data/spec/data/velvet_test_trails/Assem/stats.txt +5 -0
  506. data/spec/data/velvet_test_trails/node_fwds.fa +8 -0
  507. data/spec/data/velvet_test_trails/node_seqs.fa +9 -0
  508. data/spec/data/velvet_test_trails/nodes_fwd_rev.fa +16 -0
  509. data/spec/data/velvet_test_trails/read1.fa +2 -0
  510. data/spec/data/velvet_test_trails/reads.fa +50 -0
  511. data/spec/data/velvet_test_trails_reverse/Assem/LastGraph +17 -0
  512. data/spec/data/velvet_test_trails_reverse/Assem/a.svg +53 -0
  513. data/spec/data/velvet_test_trails_reverse/reads_reversed.fa +10 -0
  514. data/spec/data/visualise/1/LastGraph +6695 -0
  515. data/spec/data/visualise/2_paired_end/HOWTO_RECREATE.txt +10 -0
  516. data/spec/data/visualise/2_paired_end/rand1.fa +2 -0
  517. data/spec/data/visualise/2_paired_end/rand2.fa +2 -0
  518. data/spec/data/visualise/2_paired_end/with_gaps.fa +8 -0
  519. data/spec/data/visualise/2_paired_end/with_gaps.read_pairs.fa.gz +0 -0
  520. data/spec/data/wander/1/random1.fa +2 -0
  521. data/spec/data/wander/1/random1.sammy.fa +804 -0
  522. data/spec/depth_first_search_spec.rb +190 -0
  523. data/spec/dijkstra_spec.rb +143 -0
  524. data/spec/explore_spec.rb +29 -0
  525. data/spec/fluffer_spec.rb +155 -0
  526. data/spec/gapfiller_spec.rb +107 -0
  527. data/spec/graph_explorer_spec.rb +475 -0
  528. data/spec/graph_generator_spec.rb +99 -0
  529. data/spec/height_finder_spec.rb +306 -0
  530. data/spec/kmer_abundance_pattern_spec.rb +56 -0
  531. data/spec/kmer_coverage_based_path_filter_spec.rb +73 -0
  532. data/spec/kmer_profile_finder_spec.rb +38 -0
  533. data/spec/kmers_count_tabulate_spec.rb +120 -0
  534. data/spec/oriented_node_trail_spec.rb +221 -0
  535. data/spec/paired_end_neighbours_spec.rb +126 -0
  536. data/spec/paths_between_nodes_spec.rb +349 -0
  537. data/spec/priner_spec.rb +7 -0
  538. data/spec/read_input_spec.rb +23 -0
  539. data/spec/read_selection_by_kmer_spec.rb +166 -0
  540. data/spec/read_to_node_spec.rb +35 -0
  541. data/spec/roundup_spec.rb +366 -0
  542. data/spec/scaffold_breaker_spec.rb +144 -0
  543. data/spec/sequence_spec.rb +43 -0
  544. data/spec/single_coherent_paths_between_nodes_spec.rb +492 -0
  545. data/spec/single_coherent_wanderer_spec.rb +120 -0
  546. data/spec/single_ended_assembler_spec.rb +398 -0
  547. data/spec/spec_helper.rb +310 -0
  548. data/spec/velvet_graph_sequence_extractor_spec.rb +80 -0
  549. data/spec/visualise_spec.rb +105 -0
  550. data/spec/wander_spec.rb +119 -0
  551. data/spec/watch_for_changes.sh +16 -0
  552. data/validation/fasta_compare.rb +72 -0
  553. data/validation/gapfill_simulate_perfect.rb +108 -0
  554. metadata +899 -0
@@ -0,0 +1,707 @@
1
+ require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
2
+
3
+ Bio::Log::CLI.logger('stderr'); Bio::Log::CLI.trace('debug'); log = Bio::Log::LoggerPlus.new('finishm'); Bio::Log::CLI.configure('finishm')
4
+
5
+ class GraphTesting
6
+ def self.metapath_to_array(metapath)
7
+ to_return = []
8
+ metapath.collect do |node_or_arr|
9
+ if node_or_arr.kind_of?(Bio::AssemblyGraphAlgorithms::BubblyAssembler::Bubble)
10
+ paths = []
11
+ rev = node_or_arr.is_reverse
12
+ node_or_arr.each_path do |path|
13
+ path_in_bubble = rev ? path[1..-1] : path[0...-1]
14
+ paths.push path_in_bubble.collect{|onode| onode.node_id}
15
+ end
16
+ if rev
17
+ to_return.push node_or_arr.converging_oriented_node_settable[0]
18
+ to_return.push paths.sort.collect{|path| path.length == 1 ? path[0] : path }
19
+ else
20
+ to_return.push paths.sort.collect{|path| path.length == 1 ? path[0] : path }
21
+ to_return.push node_or_arr.converging_oriented_node_settable[0]
22
+ end
23
+ elsif node_or_arr.kind_of?(Bio::Velvet::Graph::OrientedNodeTrail::OrientedNode)
24
+ to_return.push node_or_arr.node_id
25
+ else
26
+ raise "Unknown metapath element: #{node_or_arr.inspect}"
27
+ end
28
+ end
29
+ return to_return
30
+ end
31
+
32
+ def self.metapaths_to_arrays(metapaths)
33
+ return metapaths.collect do |m|
34
+ metapath_to_array m
35
+ end
36
+ end
37
+ end
38
+
39
+
40
+ describe "BubblyAssembler" do
41
+ describe 'assemble_from' do
42
+ it 'should assemble something easy' do
43
+ graph, initial_path, terminal = GraphTesting.emit_ss([
44
+ [1,2],
45
+ [2,3],
46
+ ], 1, 3)
47
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
48
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
49
+ GraphTesting.metapath_to_array(metapath).should == [1,2,3]
50
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..3).to_a
51
+ end
52
+
53
+ it 'should handle the simplest bubble' do
54
+ graph, initial_path, terminal = GraphTesting.emit_ss([
55
+ [1,2],
56
+ [2,4],
57
+ [1,3],
58
+ [3,4],
59
+ ], 1, 4)
60
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
61
+ cartographer.assembly_options[:max_tip_length] = -1
62
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
63
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,3],4]
64
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..4).to_a
65
+ end
66
+
67
+ it 'should handle a simple bubble' do
68
+ graph, initial_path, terminal = GraphTesting.emit_ss([
69
+ [1,2],
70
+ [2,3],
71
+ [1,4],
72
+ [4,3],
73
+ [3,5],
74
+ ], 1, 5)
75
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
76
+ cartographer.assembly_options[:max_tip_length] = -1
77
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
78
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5]
79
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..5).to_a
80
+ end
81
+
82
+ it 'should handle a slightly complex bubble' do
83
+ graph, initial_path, terminal = GraphTesting.emit_ss([
84
+ [1,2],
85
+ [2,3],
86
+ [3,4],
87
+ [4,7],
88
+ [2,5],
89
+ [5,4],
90
+ [5,6],
91
+ [6,7],
92
+ [7,8]
93
+ ], 1, 8)
94
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
95
+ cartographer.assembly_options[:max_tip_length] = -1
96
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
97
+ GraphTesting.metapath_to_array(metapath).should == [1,2,[[3,4],[5,4],[5,6]],7,8]
98
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..8).to_a
99
+ end
100
+
101
+ it 'should deal with fluff in the middle of a bubble' do
102
+ graph, initial_path, terminal = GraphTesting.emit_ss([
103
+ [1,2],
104
+ [2,3],
105
+ [1,4],
106
+ [4,3],
107
+ [3,5],
108
+ [4,99],
109
+ ], 1, 5)
110
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
111
+ cartographer.assembly_options[:max_tip_length] = 11
112
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
113
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5]
114
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..5).to_a+[99]
115
+ end
116
+
117
+ it 'should deal with fluff at the beginning of a bubble' do
118
+ graph, initial_path, terminal = GraphTesting.emit_ss([
119
+ [1,2],
120
+ [2,3],
121
+ [1,99],
122
+ [1,4],
123
+ [4,3],
124
+ [3,5],
125
+ ], 1, 5)
126
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
127
+ cartographer.assembly_options[:max_tip_length] = 11
128
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
129
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5]
130
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..5).to_a+[99]
131
+ end
132
+
133
+ it 'should deal with fluff at the end of a bubble' do
134
+ graph, initial_path, terminal = GraphTesting.emit_ss([
135
+ [1,2],
136
+ [2,3],
137
+ [1,4],
138
+ [4,3],
139
+ [3,5],
140
+ [5,6],
141
+ [3,99]
142
+ ], 1, 6)
143
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
144
+ cartographer.assembly_options[:max_tip_length] = 11
145
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
146
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5,6]
147
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..6).to_a+[99]
148
+ end
149
+
150
+ it 'should be able to handle two bubbles' do
151
+ graph, initial_path, terminal = GraphTesting.emit_ss([
152
+ [1,2],
153
+ [2,3],
154
+ [1,4],
155
+ [4,3],
156
+ [3,5],
157
+
158
+ [5,6],
159
+ [5,7],
160
+ [6,8],
161
+ [8,9],
162
+ [7,8],
163
+ [7,10],
164
+ [10,9],
165
+
166
+ [9,11],
167
+ [11,12],
168
+
169
+ [7,99],
170
+ ], 1, 6)
171
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
172
+ cartographer.assembly_options[:max_tip_length] = 11
173
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
174
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5,[[6,8],[7,8],[7,10]],9,11,12]
175
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..12).to_a + [99]
176
+ end
177
+
178
+ it 'should respect the leash length' do
179
+ graph, initial_path, terminal = GraphTesting.emit_ss([
180
+ [1,2],
181
+ [2,3],
182
+ [1,4],
183
+ [4,3],
184
+
185
+ [3,5],
186
+ [5,6],
187
+ ], 1, 2)
188
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
189
+ cartographer.assembly_options[:max_tip_length] = -1
190
+ cartographer.assembly_options[:max_bubble_length] = 5
191
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
192
+ GraphTesting.metapath_to_array(metapath).should == [1]
193
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == [1]
194
+ end
195
+
196
+ it 'should be able to start from initial paths that include bubbles' do
197
+ graph, initial_path, terminal = GraphTesting.emit_ss([
198
+ [1,2],
199
+ [2,3],
200
+ [1,4],
201
+ [4,3],
202
+ [3,5],
203
+ [5,6],
204
+ [3,99]
205
+ ], 1, 6)
206
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
207
+ cartographer.assembly_options[:max_tip_length] = 11
208
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
209
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5,6]
210
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..6).to_a + [99]
211
+
212
+ metapath2, visited_nodes2 = cartographer.assemble_from(
213
+ metapath[0..1],
214
+ Set.new(visited_nodes.to_a.select{|s| s[0] <= 4 })
215
+ )
216
+ GraphTesting.metapath_to_array(metapath2).should == [1,[2,4],3,5,6]
217
+ visited_nodes2.to_a.collect{|s| s[0]}.sort.should == (1..6).to_a + [99]
218
+ end
219
+
220
+ it 'should respect the node count limits' do
221
+ graph, initial_path, terminal = GraphTesting.emit_ss([
222
+ [1,2],
223
+ [2,3],
224
+ [1,4],
225
+ [4,3], #bubble has 2 nodes
226
+ [3,5],
227
+
228
+ [5,6],
229
+ [5,7],
230
+ [6,8],
231
+ [8,9],
232
+ [7,8],
233
+ [7,10],
234
+ [10,9], #bubble has 5 nodes
235
+
236
+ [9,11],
237
+ [11,12],
238
+
239
+ [7,99],
240
+ ], 1, 6)
241
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
242
+ cartographer.assembly_options[:max_tip_length] = 11
243
+ cartographer.assembly_options[:bubble_node_count_limit] = 99
244
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
245
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5,[[6,8],[7,8],[7,10]],9,11,12]
246
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..12).to_a + [99]
247
+
248
+ cartographer.assembly_options[:bubble_node_count_limit] = 4
249
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
250
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,4],3,5]
251
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..5).to_a + [99] #debatable whether 99 should be included here, but eh
252
+ end
253
+
254
+ it 'should not close bubbles prematurely' do
255
+ # bubble breaker graph
256
+ graph, initial_path, terminal = GraphTesting.emit_ss([
257
+ [1,2],
258
+ [2,3],
259
+ [3,4],
260
+ [1,4],
261
+ [3,5],
262
+ [4,5],
263
+ [4,6],
264
+ [5,6],
265
+ [6,7]
266
+ ], 1, 6)
267
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
268
+ cartographer.assembly_options[:max_tip_length] = 11
269
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
270
+ GraphTesting.metapath_to_array(metapath).should == [1,[[2,3,4],[2,3,4,5],[2,3,5],4,[4,5]],6,7]
271
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == (1..7).to_a
272
+ end
273
+
274
+ it 'should handle circuits in bubbles' do
275
+ graph, initial_path, terminal = GraphTesting.emit_ss([
276
+ [1,2],
277
+ [2,4],
278
+ [1,3],
279
+ [3,4],
280
+
281
+ [2,10],
282
+ [10,2],
283
+ ], 1, 4)
284
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
285
+ cartographer.assembly_options[:max_tip_length] = -1
286
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
287
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,[2,10,2],3],4]
288
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == [1,2,3,4,10]
289
+ end
290
+
291
+ it 'should not die on bubbles that have circuits around the converging node' do
292
+ graph, initial_path, terminal = GraphTesting.emit_ss([
293
+ [1,20],
294
+ [20,21],
295
+ [21,4],
296
+ [1,5],
297
+ [5,4],
298
+ [4,3],
299
+ [3,5],
300
+ [3,8],
301
+ [8,9],
302
+ ], 1, 4)
303
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
304
+ cartographer.assembly_options[:max_tip_length] = -1
305
+ graph.nodes[20].ends_of_kmers_of_node = 'T'*100 #make this long so the circuit is discovered first
306
+ graph.nodes[8].ends_of_kmers_of_node = 'T'*100 #make exit from bubble long so it is not discarded as a dead-end
307
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
308
+ GraphTesting.metapath_to_array(metapath).should == [1,[5,[20, 21]],4,3,[[],[5,4,3]],8,9]
309
+ metapath.reference_trail.to_shorthand.should == '1s,5s,4s,3s,5s,4s,3s,8s,9s'
310
+ visited_nodes.to_a.collect{|s| s[0]}.sort.should == [1,3,4,5,8,9,20,21]
311
+ end
312
+
313
+ it 'should take the longest path at the end of the contig' do
314
+ graph, initial_path, terminal = GraphTesting.emit_ss([
315
+ [1,2],
316
+ [1,3],
317
+ ], 1, 4)
318
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
319
+ graph.nodes[3].ends_of_kmers_of_node = 'T'*100 #make this long so it is chosen
320
+ cartographer.assembly_options[:max_tip_length] = 200
321
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
322
+ GraphTesting.metapath_to_array(metapath).should == [1,3]
323
+ end
324
+
325
+ it 'should work on bubbles at the ends of contigs within tip length' do
326
+ graph, initial_path = GraphTesting.emit_ss([
327
+ [1,2],
328
+ [1,3],
329
+ [2,4],
330
+ [3,4],
331
+ ], 1, 2)
332
+ assembler = Bio::AssemblyGraphAlgorithms::SingleEndedAssembler.new graph
333
+ graph.nodes[3].ends_of_kmers_of_node = 'T'*11 #make this long so it is chosen
334
+ assembler.assembly_options[:max_tip_length] = 200
335
+ metapath, visited_nodes = assembler.assemble_from(initial_path)
336
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,3],4] #known bug. Currently it chooses a single path because one side is chosen as the longest short tip
337
+ end
338
+ end
339
+
340
+ describe 'assemble' do
341
+ it 'should be able to get out of the middle of a simple path' do
342
+ graph = GraphTesting.emit([
343
+ [1,2],
344
+ [2,3],
345
+ [4,1],
346
+ ])
347
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
348
+ :max_tip_length => -1,
349
+ :min_contig_size => 0,
350
+ }
351
+ metapaths = cartographer.assemble
352
+ GraphTesting.metapaths_to_arrays(metapaths).should == [[4,1,2,3]]
353
+ end
354
+
355
+ it 'should be able to get out of the middle convergent node when both sides have bubbles' do
356
+ graph = GraphTesting.emit([
357
+ [1,2],
358
+ [2,3],
359
+ [4,1],
360
+
361
+ [11,12],
362
+ [11,13],
363
+ [12,4],
364
+ [13,4],
365
+
366
+ [3,21],
367
+ [3,22],
368
+ [21,23],
369
+ [22,23],
370
+ ])
371
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
372
+ :max_tip_length => -1,
373
+ :min_contig_size => 0,
374
+ }
375
+ metapaths = cartographer.assemble
376
+ GraphTesting.metapaths_to_arrays(metapaths).should == [
377
+ [11,[12,13],4,1,2,3,[21,22],23]
378
+ ]
379
+ end
380
+
381
+ it 'should be able to get out of the middle of a path when it is bubbly' do
382
+ graph = GraphTesting.emit([
383
+ [1,2],
384
+ [2,3],
385
+ [4,1],
386
+ [4,5],
387
+ [5,3],
388
+ [6,4],
389
+ ])
390
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
391
+ :max_tip_length => -1,
392
+ :min_contig_size => 0,
393
+ }
394
+ metapaths = cartographer.assemble
395
+ GraphTesting.metapaths_to_arrays(metapaths).should == [
396
+ [6,4,[5,[1,2]],3]
397
+ ] #This is a known bug. Fixing requires at least medium-size change to the algoruithm.
398
+ end
399
+
400
+ it 'should deal with short tips everywhere' do
401
+ graph = GraphTesting.emit([
402
+ [1,2],
403
+ [2,3],
404
+ [4,1],
405
+
406
+ [11,12],
407
+ [11,13],
408
+ [12,4],
409
+ [13,4],
410
+
411
+ [3,21],
412
+ [3,22],
413
+ [21,23],
414
+ [22,23],
415
+ [23,24],
416
+
417
+ [1,101],
418
+ [2,102],
419
+ [3,103],
420
+ [4,104],
421
+ [11,111],
422
+ [12,112],
423
+ [13,113],
424
+ [21,121],
425
+ [22,122],
426
+ [23,123],
427
+
428
+ [201,1],
429
+ [202,2],
430
+ [203,3],
431
+ [204,4],
432
+ [211,11],
433
+ [212,12],
434
+ [213,13],
435
+ [221,21],
436
+ [222,22],
437
+ [223,23],
438
+ ])
439
+ # make the end nodes longer so can test for tip length
440
+ graph.nodes[11].ends_of_kmers_of_node = 'T'*30
441
+ graph.nodes[23].ends_of_kmers_of_node = 'T'*30
442
+
443
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
444
+ :max_tip_length => 15,
445
+ :min_contig_size => 0,
446
+ }
447
+ metapaths = cartographer.assemble
448
+ GraphTesting.metapaths_to_arrays(metapaths).should == [
449
+ [211,11,[12,13],4,1,2,3,[21,22],23,123]
450
+ ] #known bug
451
+ end
452
+
453
+ it 'should handle reverse circuits in bubbles' do
454
+ graph, initial_path, terminal = GraphTesting.emit_ss([
455
+ [1,2],
456
+ [2,4],
457
+ [1,3],
458
+ [3,4],
459
+
460
+ [10,1],
461
+ [11,1],
462
+ [9,10],
463
+ [9,11],
464
+
465
+ [99,10],
466
+ [10,99],
467
+ ], 1, 4)
468
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph
469
+ cartographer.assembly_options[:max_tip_length] = -1
470
+ cartographer.assembly_options[:min_contig_size] = -1
471
+ metapaths = cartographer.assemble
472
+ GraphTesting.metapaths_to_arrays(metapaths).should == [
473
+ [9,[10,[10,99,10],11],1,[2,3],4]
474
+ ]
475
+ end
476
+
477
+
478
+ it 'should handle reverse circuit bubbles at the end of in bubbles' do
479
+ fail
480
+ end
481
+
482
+ it 'should be able to assemble several contigs' do
483
+ fail
484
+ end
485
+
486
+ it 'should be able to use recoherence to get around an inter-genome repeat' do
487
+ fail
488
+ end
489
+
490
+ it 'should handle when is_short_tip? is given a bubble' do
491
+ fail #why does this even happen?
492
+ # make sure that reverse is also ok with this.
493
+ end
494
+ end
495
+ end
496
+
497
+
498
+
499
+
500
+ describe 'metapath' do
501
+ it 'should act like an array' do
502
+ metapath = Bio::AssemblyGraphAlgorithms::BubblyAssembler::MetaPath.new
503
+ metapath.length.should == 0
504
+ graph, initial_path, term = GraphTesting.emit_ss([[1,2]],1,2)
505
+ onode = initial_path[0]
506
+ metapath << onode
507
+ metapath.length.should == 1
508
+ metapath[0].should == onode
509
+ end
510
+
511
+ it 'should be able to output a reference contig' do
512
+ graph, initial_path = GraphTesting.emit_ss([
513
+ [1,2],
514
+ [2,3],
515
+ [1,4],
516
+ [4,3],
517
+ [3,5],
518
+ ],1,1)
519
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
520
+ :max_tip_length => -1,
521
+ :min_contig_size => 0,
522
+ }
523
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
524
+ metapath.reference_trail.to_shorthand.should == '1s,2s,3s,5s'
525
+ end
526
+
527
+ it 'should find reference paths when the bubble is a single node insert' do
528
+ graph, initial_path, terminal = GraphTesting.emit_ss([
529
+ [1,2],
530
+ [2,3],
531
+ [1,3],
532
+ [3,4],
533
+ ], 1, 1)
534
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
535
+ :max_tip_length => -1,
536
+ :min_contig_size => 0,
537
+ }
538
+ metapath, v = cartographer.assemble_from initial_path
539
+ metapath.to_shorthand.should == '1s,{2s,3s|3s},4s'
540
+ metapath.reference_trail.to_shorthand.should == '1s,2s,3s,4s'
541
+
542
+
543
+ graph, initial_path, terminal = GraphTesting.emit_ss([
544
+ [1,2],
545
+ [1,3],
546
+ [3,2],
547
+ [2,4],
548
+ ], 1, 1)
549
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
550
+ :max_tip_length => -1,
551
+ :min_contig_size => 0,
552
+ }
553
+ graph.arcs.get_arcs_by_node_id(1,2)
554
+ metapath, v = cartographer.assemble_from initial_path
555
+ metapath.to_shorthand.should == '1s,{2s|3s,2s},4s'
556
+ metapath.reference_trail.to_shorthand.should == '1s,2s,4s'
557
+ end
558
+
559
+ it 'should be able to iterate over variants from the reference contig' do
560
+ fail
561
+ end
562
+
563
+ it 'should be able to write out a reference and VCF format file' do
564
+ fail
565
+ end
566
+ end
567
+
568
+
569
+ describe 'Bubble' do
570
+ it 'should reference trail compare by node ID' do
571
+ graph, initial_path = GraphTesting.emit_ss([
572
+ [1,2],
573
+ [2,3],
574
+ [1,4],
575
+ [4,3],
576
+ ],1,1)
577
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
578
+ :max_tip_length => -1,
579
+ :min_contig_size => 0,
580
+ }
581
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
582
+ bubble = metapath[1]
583
+ bubble.reference_trail.should be_kind_of(Bio::Velvet::Graph::OrientedNodeTrail)
584
+ bubble.reference_trail.to_shorthand.should == '2s,3s'
585
+ end
586
+
587
+ it 'should reference trail compare by coverage' do
588
+ graph, initial_path = GraphTesting.emit_ss([
589
+ [1,2],
590
+ [2,3],
591
+ [1,4],
592
+ [4,3],
593
+ ],1,1)
594
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
595
+ :max_tip_length => -1,
596
+ :min_contig_size => 0,
597
+ }
598
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
599
+ bubble = metapath[1]
600
+ graph.nodes[2].coverages = [10]
601
+ graph.nodes[4].coverages = [20]
602
+ bubble.reference_trail.to_shorthand.should == '4s,3s'
603
+ end
604
+
605
+ it 'should reference trail slightly more complex bubble' do
606
+ graph, initial_path = GraphTesting.emit_ss([
607
+ [2,3],
608
+ [3,4],
609
+ [4,5],
610
+ [2,6],
611
+ [6,4],
612
+ [6,7],
613
+ [7,5],
614
+ ],2,7)
615
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
616
+ :max_tip_length => -1,
617
+ :min_contig_size => 0,
618
+ }
619
+ metapath, visited_nodes = cartographer.assemble_from(initial_path)
620
+ bubble = metapath[1]
621
+ bubble.reference_trail.to_shorthand.should == '3s,4s,5s'
622
+ graph.nodes[4].coverages = [10]
623
+ graph.nodes[7].coverages = [20]
624
+ bubble.reference_trail.to_shorthand.should == '6s,7s,5s'
625
+ end
626
+
627
+ it 'should reference trail when path is reversed' do
628
+ graph, initial_path, terminal = GraphTesting.emit_ss([
629
+ [1,2],
630
+ [2,3],
631
+ [1,4],
632
+ [4,3],
633
+ [3,5],
634
+ ], 1, 5)
635
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
636
+ :max_tip_length => -1,
637
+ :min_contig_size => 0,
638
+ }
639
+ metapath, v = cartographer.assemble_from initial_path
640
+ metapath.to_shorthand.should == '1s,{2s,3s|4s,3s},5s'
641
+ metapath.reverse!
642
+ metapath.to_shorthand.should == "5e,{3e,2e|3e,4e},1e"
643
+ metapath.reference_trail.to_shorthand.should == '5e,3e,2e,1e'
644
+ end
645
+
646
+ it 'should work with circuits found in alpha testing' do
647
+ graph, initial_path, terminal = GraphTesting.emit_ss([
648
+ [1,2],
649
+ [2,3],
650
+ [3,4],
651
+ [4,3],
652
+ [4,5],
653
+ [2,5],
654
+ [5,99],
655
+ [1,6],
656
+ [6,5],
657
+ [5,7],
658
+ [7,8]
659
+ ], 1, 1)
660
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
661
+ :max_tip_length => -1,
662
+ :min_contig_size => 0,
663
+ }
664
+ graph.nodes[7].ends_of_kmers_of_node = 'T'*100 #make exit from bubble long so it is not discarded as a dead-end
665
+ metapath, v = cartographer.assemble_from initial_path
666
+ metapath.to_shorthand.should == '1s,{2s,3s,4s,3s,4s,5s|2s,3s,4s,5s|2s,5s|6s,5s},7s,8s' #Tim - tip bug?
667
+ #metapath.fate.should == Bio::AssemblyGraphAlgorithms::BubblyAssembler::CIRCUIT_WITHIN_BUBBLE_FATE
668
+ end
669
+
670
+ it 'should give a correct coverage' do
671
+ graph, initial_path, terminal = GraphTesting.emit_ss([
672
+ [1,2],
673
+ [2,3],
674
+ [1,3],
675
+ [4,5],
676
+ ],1,1)
677
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
678
+ :max_tip_length => -1,
679
+ :min_contig_size => 0,
680
+ }
681
+ metapath, v = cartographer.assemble_from initial_path
682
+ metapath.coverage.should == 0.5
683
+ end
684
+
685
+ it 'should return each path once' do
686
+ # low-road graph
687
+ graph, initial_path, terminal = GraphTesting.emit_ss([
688
+ [1,2],
689
+ [2,3],
690
+ [3,7],
691
+ [7,8], #extend path out of bubble to avoid begin dropped as a 'long tip'
692
+ [1,4],
693
+ [4,5],
694
+ [5,6],
695
+ [5,2],
696
+ [6,3],
697
+ ], 1, 1)
698
+ cartographer = Bio::AssemblyGraphAlgorithms::BubblyAssembler.new graph, {
699
+ :max_tip_length => -1,
700
+ :min_contig_size => 0,
701
+ }
702
+ metapath, v = cartographer.assemble_from initial_path
703
+ GraphTesting.metapath_to_array(metapath).should == [1,[2,[4,5,2],[4,5,6]],3,7,8]
704
+ metapath.to_shorthand.should == '1s,{2s,3s|4s,5s,2s,3s|4s,5s,6s,3s},7s,8s'
705
+ metapath.reference_trail.to_shorthand.should == '1s,2s,3s,7s,8s'
706
+ end
707
+ end