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 +1,40 @@
1
+ /*
2
+ Copyright 2007, 2008 Daniel Zerbino (zerbino@ebi.ac.uk)
3
+
4
+ This file is part of Velvet.
5
+
6
+ Velvet is free software; you can redistribute it and/or modify
7
+ it under the terms of the GNU General Public License as published by
8
+ the Free Software Foundation; either version 2 of the License, or
9
+ (at your option) any later version.
10
+
11
+ Velvet is distributed in the hope that it will be useful,
12
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
13
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14
+ GNU General Public License for more details.
15
+
16
+ You should have received a copy of the GNU General Public License
17
+ along with Velvet; if not, write to the Free Software
18
+ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
19
+
20
+ */
21
+ #ifndef _LOCALLYCORRECTEDGRAPH_H_
22
+ #define _LOCALLYCORRECTEDGRAPH_H_
23
+
24
+ struct nodeList_st {
25
+ Node *node;
26
+ NodeList *previous;
27
+ NodeList *next;
28
+ } ATTRIBUTE_PACKED;
29
+
30
+ void prepareGraphForLocalCorrections(Graph * graph);
31
+
32
+ void correctGraphLocally(Node * start);
33
+
34
+ void deactivateLocalCorrectionSettings();
35
+
36
+ // Black arts:
37
+ void setLocalMaxReadLength(int value);
38
+ void setLocalMaxGaps(int value);
39
+ void setLocalMaxDivergence(double value);
40
+ #endif
@@ -0,0 +1,677 @@
1
+ /*
2
+ Copyright 2007, 2008 Daniel Zerbino (zerbino@ebi.ac.uk)
3
+
4
+ This file is part of Velvet.
5
+
6
+ Velvet is free software; you can redistribute it and/or modify
7
+ it under the terms of the GNU General Public License as published by
8
+ the Free Software Foundation; either version 2 of the License, or
9
+ (at your option) any later version.
10
+
11
+ Velvet is distributed in the hope that it will be useful,
12
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
13
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14
+ GNU General Public License for more details.
15
+
16
+ You should have received a copy of the GNU General Public License
17
+ along with Velvet; if not, write to the Free Software
18
+ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
19
+
20
+ */
21
+ #include <stdlib.h>
22
+ #include <stdio.h>
23
+ #ifdef _OPENMP
24
+ #include <omp.h>
25
+ #endif
26
+
27
+ #include "globals.h"
28
+ #include "allocArray.h"
29
+ #include "graph.h"
30
+ #include "recycleBin.h"
31
+ #include "passageMarker.h"
32
+ #include "tightString.h"
33
+ #include "utility.h"
34
+
35
+ typedef struct passage_st PassageMarker;
36
+
37
+ struct passage_st {
38
+ struct node_st *node;
39
+ PassageMarkerI nextInNode;
40
+ PassageMarkerI previousInNode;
41
+ PassageMarkerI twinMarker;
42
+ PassageMarkerI nextInSequence;
43
+ IDnum start;
44
+ IDnum finishOffset;
45
+ IDnum sequenceID;
46
+ boolean status;
47
+ } ATTRIBUTE_PACKED;
48
+
49
+ static AllocArray *markerMemory = NULL;
50
+ DECLARE_FAST_ACCESSORS (PM, PassageMarker, markerMemory)
51
+
52
+ static RecycleBin *listMemory = NULL;
53
+ static const int LISTBLOCKSIZE = 10000;
54
+
55
+ PassageMarkerI allocatePassageMarker()
56
+ {
57
+ if (markerMemory == NULL)
58
+ markerMemory =
59
+ newAllocArray (sizeof(PassageMarker), "PassageMarker");
60
+
61
+ return allocArrayAllocate (markerMemory);
62
+ }
63
+
64
+ static void deallocatePassageMarker(PassageMarkerI marker)
65
+ {
66
+ allocArrayFree(markerMemory, marker);
67
+ }
68
+
69
+ PassageMarkerList *allocatePassageMarkerList()
70
+ {
71
+ if (listMemory == NULL)
72
+ listMemory =
73
+ newRecycleBin(sizeof(PassageMarkerList),
74
+ LISTBLOCKSIZE);
75
+
76
+ return (PassageMarkerList *) allocatePointer(listMemory);
77
+ }
78
+
79
+ void deallocatePassageMarkerList(PassageMarkerList * marker)
80
+ {
81
+ deallocatePointer(listMemory, marker);
82
+ }
83
+
84
+ void setNextInSequence(PassageMarkerI previous, PassageMarkerI next)
85
+ {
86
+ if (previous == NULL_IDX)
87
+ return;
88
+
89
+ PM_FI2P (previous)->nextInSequence = next;
90
+ }
91
+
92
+ void extractPassageMarker(PassageMarkerI marker)
93
+ {
94
+ PassageMarker *twin;
95
+ PassageMarker *markerVal;
96
+
97
+ if (marker == NULL_IDX)
98
+ return;
99
+
100
+ markerVal = PM_FI2P (marker);
101
+ if (markerVal->node == NULL_IDX)
102
+ return;
103
+
104
+ if (markerVal->previousInNode == marker)
105
+ setMarker(markerVal->node, markerVal->nextInNode);
106
+ else
107
+ setNextInNode(markerVal->previousInNode, markerVal->nextInNode);
108
+
109
+ markerVal->previousInNode = NULL_IDX;
110
+ markerVal->nextInNode = NULL_IDX;
111
+ markerVal->node = NULL_IDX;
112
+
113
+ twin = PM_FI2P (markerVal->twinMarker);
114
+ twin->nextInNode = NULL_IDX;
115
+ twin->previousInNode = NULL_IDX;
116
+ twin->node = NULL_IDX;
117
+ }
118
+
119
+ void destroyPassageMarker(PassageMarkerI marker)
120
+ {
121
+ PassageMarker *markerVal;
122
+ PassageMarker *twinVal;
123
+ PassageMarkerI twin;
124
+
125
+ if (marker == NULL_IDX)
126
+ return;
127
+
128
+ markerVal = PM_FI2P (marker);
129
+ twin = markerVal->twinMarker;
130
+ extractPassageMarker(marker);
131
+
132
+ if (markerVal->nextInSequence != NULL_IDX
133
+ && PM_FI2P (PM_FI2P (markerVal->nextInSequence)->twinMarker)->nextInSequence == twin)
134
+ PM_FI2P (PM_FI2P (markerVal->nextInSequence)->twinMarker)->nextInSequence = NULL_IDX;
135
+
136
+ twinVal = PM_FI2P (twin);
137
+ if (twinVal->nextInSequence != NULL_IDX
138
+ && PM_FI2P (PM_FI2P (twinVal->nextInSequence)->twinMarker)->nextInSequence == marker)
139
+ PM_FI2P (PM_FI2P (twinVal->nextInSequence)->twinMarker)->nextInSequence = NULL_IDX;
140
+
141
+ deallocatePassageMarker(twin);
142
+ deallocatePassageMarker(marker);
143
+
144
+ //velvetLog("Done destroying passage marker\n");
145
+ }
146
+
147
+ void destroyAllPassageMarkers()
148
+ {
149
+ if (markerMemory != NULL)
150
+ destroyAllocArray(markerMemory);
151
+ if (listMemory != NULL)
152
+ destroyRecycleBin(listMemory);
153
+ }
154
+
155
+
156
+ void setPreviousInSequence(PassageMarkerI previous, PassageMarkerI next)
157
+ {
158
+ if (next == NULL_IDX)
159
+ return;
160
+ else if (previous == NULL_IDX)
161
+ PM_FI2P (PM_FI2P (next)->twinMarker)->nextInSequence = NULL_IDX;
162
+ else
163
+ PM_FI2P (PM_FI2P (next)->twinMarker)->nextInSequence = PM_FI2P (previous)->twinMarker;
164
+ }
165
+
166
+ void disconnectNextPassageMarker(PassageMarkerI marker, Graph * graph)
167
+ {
168
+ PassageMarkerI middle = getNextInSequence(marker);
169
+ PassageMarkerI next = getNextInSequence(middle);
170
+
171
+ setPreviousInSequence(marker, next);
172
+ concatenatePassageMarkers(marker, middle);
173
+ setNextInSequence(middle, NULL_IDX);
174
+ setPreviousInSequence(NULL_IDX, middle);
175
+ }
176
+
177
+ void deleteNextPassageMarker(PassageMarkerI marker, Graph * graph)
178
+ {
179
+ PassageMarkerI middle = getNextInSequence(marker);
180
+ PassageMarkerI next = getNextInSequence(middle);
181
+
182
+ setPreviousInSequence(marker, next);
183
+ setNextInSequence(marker, next);
184
+ setNextInSequence(middle, NULL_IDX);
185
+ setPreviousInSequence(NULL_IDX, middle);
186
+ }
187
+
188
+ PassageMarkerI getNextInNode(PassageMarkerI marker)
189
+ {
190
+ if (marker == NULL_IDX)
191
+ return NULL_IDX;
192
+
193
+ return PM_FI2P (marker)->nextInNode;
194
+ }
195
+
196
+ void setNextInNode(PassageMarkerI marker, PassageMarkerI next)
197
+ {
198
+ PassageMarker *markerVal;
199
+
200
+ // DEBUG
201
+ if (next == marker || next == getTwinMarker(marker))
202
+ abort();
203
+
204
+ if (marker == NULL_IDX)
205
+ return;
206
+
207
+ markerVal = PM_FI2P (marker);
208
+ if (next == NULL_IDX) {
209
+ markerVal->nextInNode = NULL_IDX;
210
+ PM_FI2P (markerVal->twinMarker)->nextInNode = NULL_IDX;
211
+ } else {
212
+ PassageMarker *nextVal;
213
+
214
+ if (markerVal->twinMarker == NULL_IDX) {
215
+ velvetLog("Dead marker in node %li %li\n",
216
+ (long) getNodeID(getNode(marker)),
217
+ (long) getPassageMarkerSequenceID(marker));
218
+ abort();
219
+ }
220
+ nextVal = PM_FI2P (next);
221
+ markerVal->nextInNode = next;
222
+ PM_FI2P (markerVal->twinMarker)->nextInNode = nextVal->twinMarker;
223
+ nextVal->previousInNode = marker;
224
+ PM_FI2P (nextVal->twinMarker)->previousInNode = markerVal->twinMarker;
225
+ }
226
+ }
227
+
228
+ void setTopOfTheNode(PassageMarkerI marker)
229
+ {
230
+ if (marker == NULL_IDX)
231
+ return;
232
+
233
+ PM_FI2P (marker)->previousInNode = marker;
234
+ }
235
+
236
+ PassageMarkerI getNextInSequence(PassageMarkerI marker)
237
+ {
238
+ if (marker != NULL_IDX)
239
+ {
240
+ PassageMarker *markerVal;
241
+
242
+ markerVal = PM_FI2P (marker);
243
+ if (markerVal->nextInSequence == NULL_IDX)
244
+ return NULL_IDX;
245
+ return markerVal->nextInSequence;
246
+ }
247
+ return NULL_IDX;
248
+ }
249
+
250
+ PassageMarkerI getPreviousInSequence(PassageMarkerI marker)
251
+ {
252
+ PassageMarker *twinVal;
253
+
254
+ if (marker == NULL_IDX)
255
+ return NULL_IDX;
256
+
257
+ twinVal = PM_FI2P (PM_FI2P (marker)->twinMarker);
258
+ if (twinVal->nextInSequence == NULL_IDX)
259
+ return NULL_IDX;
260
+
261
+ return PM_FI2P (twinVal->nextInSequence)->twinMarker;
262
+ }
263
+
264
+ void
265
+ connectPassageMarkers(PassageMarkerI previous, PassageMarkerI next,
266
+ Graph * graph)
267
+ {
268
+ if (previous != NULL_IDX)
269
+ setNextInSequence(previous, next);
270
+
271
+ if (next != NULL_IDX)
272
+ setPreviousInSequence(previous, next);
273
+ }
274
+
275
+ char *readPassageMarker(PassageMarkerI marker)
276
+ {
277
+ PassageMarker *markerVal;
278
+ char *s = mallocOrExit(100, char);
279
+
280
+ if (marker == NULL_IDX)
281
+ return s;
282
+
283
+ markerVal = PM_FI2P (marker);
284
+ sprintf(s, "MARKER %ld (%lld -> %lld):", (long) markerVal->sequenceID,
285
+ (long long) markerVal->start, (long long) getPassageMarkerFinish(marker));
286
+
287
+ if (getPreviousInSequence(marker) == NULL_IDX)
288
+ sprintf(s, "%s START -> %ld", s,
289
+ (long) getNodeID(getNode(marker)));
290
+ else
291
+ sprintf(s, "%s %ld -> %ld", s,
292
+ (long) getNodeID(getNode(getPreviousInSequence(marker))),
293
+ (long) getNodeID(getNode(marker)));
294
+
295
+ if (getNextInSequence(marker) == NULL_IDX)
296
+ sprintf(s, "%s -> FINISH", s);
297
+ else
298
+ sprintf(s, "%s -> %ld ", s,
299
+ (long) getNodeID(getNode(getNextInSequence(marker))));
300
+
301
+ return s;
302
+ }
303
+
304
+ PassageMarkerI addPassageMarker(IDnum sequenceID, Coordinate start,
305
+ Node * node)
306
+ {
307
+ PassageMarkerI marker = allocatePassageMarker();
308
+ PassageMarkerI twinMarker = allocatePassageMarker();
309
+ PassageMarker *markerVal;
310
+ PassageMarker *twinVal;
311
+
312
+ markerVal = PM_FI2P (marker);
313
+ twinVal = PM_FI2P (twinMarker);
314
+
315
+ markerVal->sequenceID = sequenceID;
316
+ markerVal->start = start;
317
+ markerVal->node = node;
318
+ markerVal->nextInSequence = NULL_IDX;
319
+ markerVal->finishOffset = 0;
320
+ markerVal->twinMarker = twinMarker;
321
+ markerVal->status = false;
322
+
323
+ twinVal->sequenceID = -sequenceID;
324
+ twinVal->start = start + getNodeLength(node);
325
+ twinVal->node = getTwinNode(node);
326
+ twinVal->nextInSequence = NULL_IDX;
327
+ twinVal->finishOffset = 0;
328
+ twinVal->twinMarker = marker;
329
+ twinVal->status = false;
330
+
331
+ setNextInNode(marker, getMarker(node));
332
+ setMarker(node, marker);
333
+
334
+ return marker;
335
+ }
336
+
337
+ PassageMarkerList *copyPassageMarkerList(PassageMarkerList * list)
338
+ {
339
+ PassageMarkerList *copy;
340
+ PassageMarkerList *result = NULL;
341
+ PassageMarkerList *pointer;
342
+
343
+ if (list == NULL)
344
+ return NULL;
345
+
346
+ for (pointer = list; pointer != NULL; pointer = pointer->next) {
347
+ copy = allocatePassageMarkerList();
348
+ copy->marker = pointer->marker;
349
+ copy->next = result;
350
+ result = copy;
351
+ }
352
+
353
+ return result;
354
+ }
355
+
356
+ void incrementFinishOffset(PassageMarkerI marker, Coordinate offset)
357
+ {
358
+ PM_FI2P (marker)->finishOffset += offset;
359
+ }
360
+
361
+ void incrementStartOffset(PassageMarkerI marker, Coordinate offset)
362
+ {
363
+ PM_FI2P (PM_FI2P (marker)->twinMarker)->finishOffset += offset;
364
+ }
365
+
366
+ Coordinate getFinishOffset(PassageMarkerI marker)
367
+ {
368
+ return PM_FI2P (marker)->finishOffset;
369
+ }
370
+
371
+ void setFinishOffset(PassageMarkerI marker, Coordinate offset)
372
+ {
373
+ PM_FI2P (marker)->finishOffset = offset;
374
+ }
375
+
376
+ Coordinate getStartOffset(PassageMarkerI marker)
377
+ {
378
+ return PM_FI2P (PM_FI2P (marker)->twinMarker)->finishOffset;
379
+ }
380
+
381
+ void setStartOffset(PassageMarkerI marker, Coordinate offset)
382
+ {
383
+ PM_FI2P (PM_FI2P (marker)->twinMarker)->finishOffset = offset;
384
+ }
385
+
386
+ void transposePassageMarker(PassageMarkerI marker, Node * node)
387
+ {
388
+ PassageMarker *markerVal;
389
+ PassageMarker *twinMarkerVal;
390
+
391
+ markerVal = PM_FI2P (marker);
392
+ twinMarkerVal = PM_FI2P (markerVal->twinMarker);
393
+ insertPassageMarker(marker, node);
394
+ markerVal->node = node;
395
+ insertPassageMarker(markerVal->twinMarker, getTwinNode(node));
396
+ twinMarkerVal->node = getTwinNode(node);
397
+ }
398
+
399
+ PassageMarkerI getTwinMarker(PassageMarkerI marker)
400
+ {
401
+ return PM_FI2P (marker)->twinMarker;
402
+ }
403
+
404
+ IDnum getPassageMarkerSequenceID(PassageMarkerI marker)
405
+ {
406
+ return PM_FI2P (marker)->sequenceID;
407
+ }
408
+
409
+ IDnum getAbsolutePassMarkerSeqID(PassageMarkerI marker)
410
+ {
411
+ IDnum ID = PM_FI2P (marker)->sequenceID;
412
+
413
+ if (ID > 0)
414
+ return ID;
415
+ else
416
+ return -ID;
417
+ }
418
+
419
+ Node *getNode(PassageMarkerI marker)
420
+ {
421
+ if (marker == NULL_IDX)
422
+ return NULL;
423
+
424
+ return PM_FI2P (marker)->node;
425
+ }
426
+
427
+ void concatenatePassageMarkers(PassageMarkerI marker,
428
+ PassageMarkerI next)
429
+ {
430
+ PassageMarker *markerVal;
431
+ PassageMarker *nextVal;
432
+
433
+ if (marker == NULL_IDX || next == NULL_IDX)
434
+ return;
435
+
436
+ markerVal = PM_FI2P (marker);
437
+ nextVal = PM_FI2P (next);
438
+
439
+ markerVal->finishOffset = nextVal->finishOffset;
440
+ PM_FI2P (markerVal->twinMarker)->start = PM_FI2P (nextVal->twinMarker)->start;
441
+ markerVal->nextInSequence = nextVal->nextInSequence;
442
+ }
443
+
444
+ boolean getPassageMarkerStatus(PassageMarkerI marker)
445
+ {
446
+ return PM_FI2P (marker)->status;
447
+ }
448
+
449
+ void setPassageMarkerStatus(PassageMarkerI marker, boolean status)
450
+ {
451
+ PassageMarker *markerVal = PM_FI2P (marker);
452
+
453
+ markerVal->status = status;
454
+ PM_FI2P (markerVal->twinMarker)->status = status;
455
+ }
456
+
457
+ boolean isDestinationToMarker(PassageMarkerI marker, Node * node)
458
+ {
459
+ PassageMarker *markerVal = PM_FI2P (marker);
460
+
461
+ if (markerVal->nextInSequence == NULL_IDX)
462
+ return false;
463
+
464
+ return PM_FI2P (markerVal->nextInSequence)->node == node;
465
+ }
466
+
467
+ boolean isTerminal(PassageMarkerI marker)
468
+ {
469
+ PassageMarker *markerVal;
470
+
471
+ if (marker == NULL_IDX)
472
+ return false;
473
+
474
+ markerVal = PM_FI2P (marker);
475
+ return markerVal->nextInSequence == NULL_IDX;
476
+ }
477
+
478
+ boolean isInitial(PassageMarkerI marker)
479
+ {
480
+ PassageMarker *markerVal;
481
+
482
+ if (marker == NULL_IDX)
483
+ return false;
484
+
485
+ markerVal = PM_FI2P (marker);
486
+ if (markerVal->twinMarker == NULL_IDX) {
487
+ velvetLog("Unpaired marker seq %ld start %lld node %ld\n",
488
+ (long) markerVal->sequenceID, (long long) markerVal->start,
489
+ (long) getNodeID(markerVal->node));
490
+ velvetLog("SNAFU\n");
491
+ abort();
492
+ }
493
+
494
+ return PM_FI2P (markerVal->twinMarker)->nextInSequence == NULL_IDX;
495
+ }
496
+
497
+ Coordinate getPassageMarkerStart(PassageMarkerI marker)
498
+ {
499
+ return PM_FI2P (marker)->start;
500
+ }
501
+
502
+ void setPassageMarkerStart(PassageMarkerI marker, Coordinate start)
503
+ {
504
+ PM_FI2P (marker)->start = start;
505
+ }
506
+
507
+ Coordinate getPassageMarkerFinish(PassageMarkerI marker)
508
+ {
509
+ PassageMarker *twinMarkerVal;
510
+
511
+ twinMarkerVal = PM_FI2P (PM_FI2P (marker)->twinMarker);
512
+ if (twinMarkerVal->start == -10)
513
+ return -10;
514
+
515
+ return twinMarkerVal->start;
516
+ }
517
+
518
+ void setPassageMarkerFinish(PassageMarkerI marker, Coordinate finish)
519
+ {
520
+ PassageMarker *twinMarkerVal;
521
+
522
+ twinMarkerVal = PM_FI2P (PM_FI2P (marker)->twinMarker);
523
+ if (finish == -10)
524
+ twinMarkerVal->start = -10;
525
+
526
+ twinMarkerVal->start = finish;
527
+ }
528
+
529
+ Coordinate getPassageMarkerLength(PassageMarkerI marker)
530
+ {
531
+ PassageMarker *markerVal;
532
+ PassageMarker *twinMarkerVal;
533
+
534
+ markerVal = PM_FI2P (marker);
535
+ twinMarkerVal = PM_FI2P (markerVal->twinMarker);
536
+ if (markerVal->start == -10 || twinMarkerVal->start == -10)
537
+ return 0;
538
+
539
+ else if (markerVal->sequenceID > 0)
540
+ return twinMarkerVal->start - markerVal->start;
541
+ else
542
+ return markerVal->start - twinMarkerVal->start;
543
+ }
544
+
545
+ int passageMarkerDirection(PassageMarkerI marker)
546
+ {
547
+ if (PM_FI2P (marker)->sequenceID > 0)
548
+ return 1;
549
+ else
550
+ return -1;
551
+ }
552
+
553
+ PassageMarkerI addUncertainPassageMarker(IDnum sequenceID, Node * node)
554
+ {
555
+ PassageMarkerI marker = allocatePassageMarker();
556
+ PassageMarkerI twinMarker = allocatePassageMarker();
557
+ PassageMarker *markerVal = PM_FI2P (marker);
558
+ PassageMarker *twinMarkerVal = PM_FI2P (twinMarker);
559
+
560
+ markerVal->sequenceID = sequenceID;
561
+ markerVal->start = -10;
562
+ markerVal->node = node;
563
+ markerVal->nextInSequence = NULL_IDX;
564
+ markerVal->finishOffset = 0;
565
+ markerVal->twinMarker = twinMarker;
566
+ markerVal->status = false;
567
+
568
+ twinMarkerVal->sequenceID = -sequenceID;
569
+ twinMarkerVal->start = -10;
570
+ if (node == NULL)
571
+ twinMarkerVal->node = NULL;
572
+ else
573
+ twinMarkerVal->node = getTwinNode(node);
574
+ twinMarkerVal->nextInSequence = NULL_IDX;
575
+ twinMarkerVal->finishOffset = 0;
576
+ twinMarkerVal->twinMarker = marker;
577
+ twinMarkerVal->status = false;
578
+
579
+ if (node != NULL) {
580
+ setNextInNode(marker, getMarker(node));
581
+ setMarker(node, marker);
582
+ }
583
+
584
+ return marker;
585
+ }
586
+
587
+ PassageMarkerList *newPassageMarkerList(PassageMarkerI marker,
588
+ PassageMarkerList * next)
589
+ {
590
+ PassageMarkerList *list = allocatePassageMarkerList();
591
+ list->marker = marker;
592
+ list->next = next;
593
+ return list;
594
+ }
595
+
596
+ PassageMarkerI newPassageMarker(IDnum seqID, Coordinate start,
597
+ Coordinate finish, Coordinate startOffset,
598
+ Coordinate finishOffset)
599
+ {
600
+ PassageMarkerI marker;
601
+ PassageMarkerI twinMarker;
602
+ PassageMarker *markerVal;
603
+ PassageMarker *twinMarkerVal;
604
+
605
+ #ifdef _OPENMP
606
+ #pragma omp critical
607
+ {
608
+ #endif
609
+ marker = allocatePassageMarker();
610
+ twinMarker = allocatePassageMarker();
611
+ #ifdef _OPENMP
612
+ }
613
+ #endif
614
+ markerVal = PM_FI2P (marker);
615
+ twinMarkerVal = PM_FI2P (twinMarker);
616
+
617
+ // velvetLog("Values %d\t%d\t%d\t%d\t%d\n", seqID, start, finish, startOffset, finishOffset);
618
+
619
+ markerVal->sequenceID = seqID;
620
+ markerVal->node = NULL;
621
+ markerVal->nextInSequence = NULL_IDX;
622
+ markerVal->twinMarker = twinMarker;
623
+ markerVal->nextInNode = NULL_IDX;
624
+ markerVal->status = false;
625
+
626
+ twinMarkerVal->sequenceID = -seqID;
627
+ twinMarkerVal->node = NULL;
628
+ twinMarkerVal->nextInSequence = NULL_IDX;
629
+ twinMarkerVal->twinMarker = marker;
630
+ twinMarkerVal->nextInNode = NULL_IDX;
631
+ twinMarkerVal->status = false;
632
+
633
+ setPassageMarkerStart(marker, start);
634
+ setPassageMarkerFinish(marker, finish);
635
+ setStartOffset(marker, startOffset);
636
+ setFinishOffset(marker, finishOffset);
637
+
638
+ if (getPassageMarkerLength(marker) < 0) {
639
+ velvetLog("Negative marker %ld %lld %lld %lld\n",
640
+ (long) getPassageMarkerSequenceID(marker),
641
+ (long long) getPassageMarkerStart(marker),
642
+ (long long) getPassageMarkerFinish(marker),
643
+ (long long) getPassageMarkerLength(marker));
644
+ abort();
645
+ }
646
+
647
+ return marker;
648
+ }
649
+
650
+ void exportMarker(FILE * outfile, PassageMarkerI marker,
651
+ TightString * sequences, int WORDLENGTH)
652
+ {
653
+ PassageMarker *markerVal = PM_FI2P (marker);
654
+ PassageMarkerI current;
655
+
656
+ if (markerVal->sequenceID > 0) {
657
+ if (!isInitial(marker)) {
658
+ return;
659
+ }
660
+ current = marker;
661
+ } else {
662
+ if (!isTerminal(marker)) {
663
+ return;
664
+ }
665
+ current = markerVal->twinMarker;
666
+ }
667
+
668
+ velvetFprintf(outfile, "SEQ\t%li\n", (long) PM_FI2P (current)->sequenceID);
669
+ for (; current != NULL_IDX; current = PM_FI2P (current)->nextInSequence) {
670
+ velvetFprintf(outfile, "%ld\t%lld\t%lld\t%lld\t%lld",
671
+ (long) getNodeID(PM_FI2P (current)->node), (long long) getStartOffset(current),
672
+ (long long) getPassageMarkerStart(current),
673
+ (long long) getPassageMarkerFinish(current),
674
+ (long long) getFinishOffset(current));
675
+ velvetFprintf(outfile, "\n");
676
+ }
677
+ }