finishm 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.document +5 -0
- data/.gitmodules +3 -0
- data/.rspec +1 -0
- data/Gemfile +31 -0
- data/LICENSE.txt +20 -0
- data/README.md +59 -0
- data/Rakefile +51 -0
- data/VERSION +1 -0
- data/bin/assembly_visualiser +106 -0
- data/bin/check_primer_combinations.rb +73 -0
- data/bin/contig_joiner.rb +244 -0
- data/bin/contigs_against_assembly.rb +153 -0
- data/bin/finishm +143 -0
- data/bin/finishm_assembler +55 -0
- data/bin/finishm_gap_closer.rb +241 -0
- data/bin/kmer_abundance_file_tool.rb +49 -0
- data/bin/kmer_pattern_to_assembly.rb +377 -0
- data/bin/kmer_profile_finder.rb +92 -0
- data/bin/kmers_count_parse.d +52 -0
- data/bin/kmers_count_tabulate.d +123 -0
- data/bin/kmers_count_tabulate.rb +84 -0
- data/bin/pcr_result_parser.rb +108 -0
- data/bin/primer_finder.rb +119 -0
- data/bin/read_selection_by_kmer.d +174 -0
- data/bin/scaffold_by_pattern.rb +119 -0
- data/bin/scaffold_connection_possibilities_to_knowns.rb +193 -0
- data/bin/scaffold_end_coverages.rb +69 -0
- data/bin/trail_validator.rb +84 -0
- data/ext/mkrf_conf.rb +56 -0
- data/ext/src/Makefile +140 -0
- data/ext/src/src/allocArray.c +305 -0
- data/ext/src/src/allocArray.h +86 -0
- data/ext/src/src/autoOpen.c +107 -0
- data/ext/src/src/autoOpen.h +18 -0
- data/ext/src/src/binarySequences.c +813 -0
- data/ext/src/src/binarySequences.h +125 -0
- data/ext/src/src/concatenatedGraph.c +233 -0
- data/ext/src/src/concatenatedGraph.h +30 -0
- data/ext/src/src/concatenatedPreGraph.c +262 -0
- data/ext/src/src/concatenatedPreGraph.h +29 -0
- data/ext/src/src/correctedGraph.c +2643 -0
- data/ext/src/src/correctedGraph.h +32 -0
- data/ext/src/src/dfib.c +509 -0
- data/ext/src/src/dfib.h +69 -0
- data/ext/src/src/dfibHeap.c +89 -0
- data/ext/src/src/dfibHeap.h +39 -0
- data/ext/src/src/dfibpriv.h +105 -0
- data/ext/src/src/fib.c +628 -0
- data/ext/src/src/fib.h +78 -0
- data/ext/src/src/fibHeap.c +79 -0
- data/ext/src/src/fibHeap.h +41 -0
- data/ext/src/src/fibpriv.h +110 -0
- data/ext/src/src/globals.h +154 -0
- data/ext/src/src/graph.c +3932 -0
- data/ext/src/src/graph.h +233 -0
- data/ext/src/src/graphReConstruction.c +1472 -0
- data/ext/src/src/graphReConstruction.h +30 -0
- data/ext/src/src/graphStats.c +2167 -0
- data/ext/src/src/graphStats.h +72 -0
- data/ext/src/src/graphStructures.h +52 -0
- data/ext/src/src/kmer.c +652 -0
- data/ext/src/src/kmer.h +73 -0
- data/ext/src/src/kmerOccurenceTable.c +236 -0
- data/ext/src/src/kmerOccurenceTable.h +44 -0
- data/ext/src/src/kseq.h +223 -0
- data/ext/src/src/locallyCorrectedGraph.c +557 -0
- data/ext/src/src/locallyCorrectedGraph.h +40 -0
- data/ext/src/src/passageMarker.c +677 -0
- data/ext/src/src/passageMarker.h +137 -0
- data/ext/src/src/preGraph.c +1717 -0
- data/ext/src/src/preGraph.h +106 -0
- data/ext/src/src/preGraphConstruction.c +990 -0
- data/ext/src/src/preGraphConstruction.h +26 -0
- data/ext/src/src/probe_node_finder.c +84 -0
- data/ext/src/src/probe_node_finder.h +6 -0
- data/ext/src/src/readCoherentGraph.c +557 -0
- data/ext/src/src/readCoherentGraph.h +30 -0
- data/ext/src/src/readSet.c +1734 -0
- data/ext/src/src/readSet.h +67 -0
- data/ext/src/src/readToNode.c +218 -0
- data/ext/src/src/readToNode.h +35 -0
- data/ext/src/src/recycleBin.c +199 -0
- data/ext/src/src/recycleBin.h +58 -0
- data/ext/src/src/roadMap.c +342 -0
- data/ext/src/src/roadMap.h +65 -0
- data/ext/src/src/run.c +318 -0
- data/ext/src/src/run.h +52 -0
- data/ext/src/src/run2.c +744 -0
- data/ext/src/src/runReadToNode.c +29 -0
- data/ext/src/src/scaffold.c +1876 -0
- data/ext/src/src/scaffold.h +64 -0
- data/ext/src/src/shortReadPairs.c +1243 -0
- data/ext/src/src/shortReadPairs.h +32 -0
- data/ext/src/src/splay.c +259 -0
- data/ext/src/src/splay.h +43 -0
- data/ext/src/src/splayTable.c +1315 -0
- data/ext/src/src/splayTable.h +31 -0
- data/ext/src/src/tightString.c +362 -0
- data/ext/src/src/tightString.h +82 -0
- data/ext/src/src/utility.c +199 -0
- data/ext/src/src/utility.h +98 -0
- data/ext/src/third-party/zlib-1.2.3/ChangeLog +855 -0
- data/ext/src/third-party/zlib-1.2.3/FAQ +339 -0
- data/ext/src/third-party/zlib-1.2.3/INDEX +51 -0
- data/ext/src/third-party/zlib-1.2.3/Makefile +154 -0
- data/ext/src/third-party/zlib-1.2.3/Makefile.in +154 -0
- data/ext/src/third-party/zlib-1.2.3/README +125 -0
- data/ext/src/third-party/zlib-1.2.3/adler32.c +149 -0
- data/ext/src/third-party/zlib-1.2.3/adler32.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/algorithm.txt +209 -0
- data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.pup +66 -0
- data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.sas +65 -0
- data/ext/src/third-party/zlib-1.2.3/as400/bndsrc +132 -0
- data/ext/src/third-party/zlib-1.2.3/as400/compile.clp +123 -0
- data/ext/src/third-party/zlib-1.2.3/as400/readme.txt +111 -0
- data/ext/src/third-party/zlib-1.2.3/as400/zlib.inc +331 -0
- data/ext/src/third-party/zlib-1.2.3/compress.c +79 -0
- data/ext/src/third-party/zlib-1.2.3/compress.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/configure +459 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/README.contrib +71 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/buffer_demo.adb +106 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/mtest.adb +156 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/read.adb +156 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/readme.txt +65 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/test.adb +463 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.adb +225 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.ads +114 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.adb +141 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.ads +450 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.adb +701 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.ads +328 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.gpr +20 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm586/README.586 +43 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm586/match.S +364 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm686/README.686 +34 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/asm686/match.S +329 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/Makefile +8 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/README +4 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.c +444 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.h +71 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.pk +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.txt +1 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLib.pas +557 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLibConst.pas +11 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/readme.txt +76 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/delphi/zlibd32.mak +93 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.build +33 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.chm +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.sln +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/AssemblyInfo.cs +58 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/ChecksumImpl.cs +202 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CircularBuffer.cs +83 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CodecBase.cs +198 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Deflater.cs +106 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.cs +288 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.csproj +141 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/GZipStream.cs +301 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Inflater.cs +105 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/UnitTests.cs +274 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/LICENSE_1_0.txt +23 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/readme.txt +58 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/README +1 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.c +608 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.h +37 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inffix9.h +107 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inflate9.h +47 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.c +323 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.h +55 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffas86.c +1157 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffast.S +1368 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/test.cpp +24 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.cpp +329 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.h +128 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream.h +307 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream_test.cpp +25 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/README +35 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/TODO +17 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/test.cc +50 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.cc +479 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.h +466 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masm686/match.asm +413 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/bld_ml64.bat +2 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.asm +513 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffas8664.c +186 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.asm +392 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/readme.txt +28 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/bld_ml32.bat +2 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.asm +972 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32c.c +62 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.asm +1083 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.obj +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/mkasm.bat +3 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/readme.txt +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ChangeLogUnzip +67 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/Makefile +25 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/crypt.h +132 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.c +177 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.h +75 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.c +270 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.h +21 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/miniunz.c +585 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/minizip.c +420 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.c +281 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.h +31 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.c +1598 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.h +354 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.c +1219 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.h +235 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/example.pas +599 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/readme.txt +76 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibd32.mak +93 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibpas.pas +236 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/Makefile +8 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/README +63 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.c +837 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.h +31 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/puff/zeros.raw +0 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.c +275 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.txt +10 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile +14 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile.msc +17 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/untgz/untgz.c +674 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/readme.txt +73 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/miniunz.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/minizip.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/testzlib.vcproj +126 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlib.rc +32 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibstat.vcproj +246 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.def +92 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.sln +78 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.vcproj +445 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/miniunz.vcproj +566 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/minizip.vcproj +563 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlib.vcproj +948 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlibdll.vcproj +567 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlib.rc +32 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibstat.vcproj +870 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.def +92 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.sln +144 -0
- data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.vcproj +1219 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.c +423 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.h +441 -0
- data/ext/src/third-party/zlib-1.2.3/crc32.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.c +1736 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.h +331 -0
- data/ext/src/third-party/zlib-1.2.3/deflate.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/example +0 -0
- data/ext/src/third-party/zlib-1.2.3/example.c +565 -0
- data/ext/src/third-party/zlib-1.2.3/examples/README.examples +42 -0
- data/ext/src/third-party/zlib-1.2.3/examples/fitblk.c +233 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gun.c +693 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzappend.c +500 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzjoin.c +448 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzlog.c +413 -0
- data/ext/src/third-party/zlib-1.2.3/examples/gzlog.h +58 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zlib_how.html +523 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zpipe.c +191 -0
- data/ext/src/third-party/zlib-1.2.3/examples/zran.c +404 -0
- data/ext/src/third-party/zlib-1.2.3/gzio.c +1026 -0
- data/ext/src/third-party/zlib-1.2.3/gzio.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/infback.c +623 -0
- data/ext/src/third-party/zlib-1.2.3/infback.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.c +318 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.h +11 -0
- data/ext/src/third-party/zlib-1.2.3/inffast.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inffixed.h +94 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.c +1368 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.h +115 -0
- data/ext/src/third-party/zlib-1.2.3/inflate.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.c +329 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.h +55 -0
- data/ext/src/third-party/zlib-1.2.3/inftrees.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/libz.a +0 -0
- data/ext/src/third-party/zlib-1.2.3/make_vms.com +461 -0
- data/ext/src/third-party/zlib-1.2.3/minigzip +0 -0
- data/ext/src/third-party/zlib-1.2.3/minigzip.c +322 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.bor +109 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.dj2 +104 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.emx +69 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.msc +106 -0
- data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.tc +94 -0
- data/ext/src/third-party/zlib-1.2.3/old/Makefile.riscos +151 -0
- data/ext/src/third-party/zlib-1.2.3/old/README +3 -0
- data/ext/src/third-party/zlib-1.2.3/old/descrip.mms +48 -0
- data/ext/src/third-party/zlib-1.2.3/old/os2/Makefile.os2 +136 -0
- data/ext/src/third-party/zlib-1.2.3/old/os2/zlib.def +51 -0
- data/ext/src/third-party/zlib-1.2.3/old/visual-basic.txt +160 -0
- data/ext/src/third-party/zlib-1.2.3/old/zlib.html +971 -0
- data/ext/src/third-party/zlib-1.2.3/projects/README.projects +41 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/README.txt +73 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/example.dsp +278 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/minigzip.dsp +278 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsp +609 -0
- data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsw +59 -0
- data/ext/src/third-party/zlib-1.2.3/qnx/package.qpg +141 -0
- data/ext/src/third-party/zlib-1.2.3/trees.c +1219 -0
- data/ext/src/third-party/zlib-1.2.3/trees.h +128 -0
- data/ext/src/third-party/zlib-1.2.3/trees.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/uncompr.c +61 -0
- data/ext/src/third-party/zlib-1.2.3/uncompr.o +0 -0
- data/ext/src/third-party/zlib-1.2.3/win32/DLL_FAQ.txt +397 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.bor +107 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.emx +69 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.gcc +141 -0
- data/ext/src/third-party/zlib-1.2.3/win32/Makefile.msc +126 -0
- data/ext/src/third-party/zlib-1.2.3/win32/VisualC.txt +3 -0
- data/ext/src/third-party/zlib-1.2.3/win32/zlib.def +60 -0
- data/ext/src/third-party/zlib-1.2.3/win32/zlib1.rc +39 -0
- data/ext/src/third-party/zlib-1.2.3/zconf.h +332 -0
- data/ext/src/third-party/zlib-1.2.3/zconf.in.h +332 -0
- data/ext/src/third-party/zlib-1.2.3/zlib.3 +159 -0
- data/ext/src/third-party/zlib-1.2.3/zlib.h +1357 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.c +318 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.h +269 -0
- data/ext/src/third-party/zlib-1.2.3/zutil.o +0 -0
- data/lib/assembly/a_b_visualiser.rb +169 -0
- data/lib/assembly/acyclic_connection_finder.rb +81 -0
- data/lib/assembly/all_orfs.rb +615 -0
- data/lib/assembly/bad_format_writer.rb +46 -0
- data/lib/assembly/bam_probe_read_selector.rb +48 -0
- data/lib/assembly/bubbly_assembler.rb +842 -0
- data/lib/assembly/c_probe_node_finder.rb +38 -0
- data/lib/assembly/connection_interpreter.rb +350 -0
- data/lib/assembly/contig_printer.rb +400 -0
- data/lib/assembly/coverage_based_graph_filter.rb +68 -0
- data/lib/assembly/depth_first_search.rb +63 -0
- data/lib/assembly/dijkstra.rb +216 -0
- data/lib/assembly/fluffer.rb +253 -0
- data/lib/assembly/graph_explorer.rb +85 -0
- data/lib/assembly/graph_generator.rb +315 -0
- data/lib/assembly/height_finder.rb +355 -0
- data/lib/assembly/hybrid_velvet_graph.rb +70 -0
- data/lib/assembly/input_genome.rb +182 -0
- data/lib/assembly/kmer_coverage_based_path_filter.rb +65 -0
- data/lib/assembly/node_finder.rb +171 -0
- data/lib/assembly/oriented_node_trail.rb +507 -0
- data/lib/assembly/paired_end_assembler.rb +53 -0
- data/lib/assembly/paired_end_neighbour_finder.rb +176 -0
- data/lib/assembly/probed_graph.rb +105 -0
- data/lib/assembly/read_input.rb +79 -0
- data/lib/assembly/read_to_node.rb +37 -0
- data/lib/assembly/scaffold_breaker.rb +126 -0
- data/lib/assembly/sequence_hasher.rb +71 -0
- data/lib/assembly/single_coherent_paths_between_nodes.rb +533 -0
- data/lib/assembly/single_coherent_wanderer.rb +261 -0
- data/lib/assembly/single_ended_assembler.rb +441 -0
- data/lib/assembly/velvet_c_binding.rb +54 -0
- data/lib/assembly/velvet_graph_sequence_extractor.rb +123 -0
- data/lib/external/VERSION +1 -0
- data/lib/finishm/assemble.rb +224 -0
- data/lib/finishm/explore.rb +217 -0
- data/lib/finishm/finisher.rb +303 -0
- data/lib/finishm/fluff.rb +122 -0
- data/lib/finishm/gapfiller.rb +325 -0
- data/lib/finishm/orfs_finder.rb +88 -0
- data/lib/finishm/path_counter.rb +90 -0
- data/lib/finishm/primers.rb +425 -0
- data/lib/finishm/primers_check.rb +176 -0
- data/lib/finishm/roundup.rb +344 -0
- data/lib/finishm/sequence.rb +142 -0
- data/lib/finishm/visualise.rb +430 -0
- data/lib/finishm/wander.rb +270 -0
- data/lib/kmer_abundance_pattern.rb +79 -0
- data/lib/kmer_multi_abundance_file.rb +48 -0
- data/lib/oligo_designer.rb +88 -0
- data/lib/priner.rb +66 -0
- data/spec/acyclic_connection_finder_spec.rb +551 -0
- data/spec/all_orfs_spec.rb +443 -0
- data/spec/assemble_spec.rb +186 -0
- data/spec/bubbly_assembler_spec.rb +707 -0
- data/spec/c_node_finder_spec.rb +58 -0
- data/spec/connection_interpreter_spec.rb +284 -0
- data/spec/contig_printer_spec.rb +291 -0
- data/spec/coverage_based_graph_filter_spec.rb +102 -0
- data/spec/data/6_3e4e5e6e.1vANME.bam +0 -0
- data/spec/data/6_3e4e5e6e.1vANME.bam.bai +0 -0
- data/spec/data/acyclic_connection_finder/1/probes.fa +5 -0
- data/spec/data/acyclic_connection_finder/1/random1.fa +2 -0
- data/spec/data/acyclic_connection_finder/1/random1.sammy.fa.gz +0 -0
- data/spec/data/acyclic_connection_finder/1/random2.fa +2 -0
- data/spec/data/acyclic_connection_finder/1/random2.sammy.fa.gz +0 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.fa +39 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.slightly_changed.fa +39 -0
- data/spec/data/assembly/1_simple_bubble_uneven_coverage/reads_combined.fa.gz +0 -0
- data/spec/data/assembly_visualiser/Contig_6_1_to_250.fa.kmers31 +220 -0
- data/spec/data/assembly_visualiser/Contig_7_1_to_250.fa.kmers31 +220 -0
- data/spec/data/assembly_visualiser/Graph +46 -0
- data/spec/data/assembly_visualiser/start_kmers1 +2 -0
- data/spec/data/bands.csv +1 -0
- data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq +0 -0
- data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq.names +544 -0
- data/spec/data/c_probe_node_finder/1/Graph2 +668 -0
- data/spec/data/c_probe_node_finder/1/LastGraph +668 -0
- data/spec/data/c_probe_node_finder/1/Log +756 -0
- data/spec/data/c_probe_node_finder/1/PreGraph +11 -0
- data/spec/data/c_probe_node_finder/1/Roadmaps +2009 -0
- data/spec/data/c_probe_node_finder/1/contigs.fa +29 -0
- data/spec/data/c_probe_node_finder/1/stats.txt +6 -0
- data/spec/data/contig_printer/1/HOWTO_RECREATE +17 -0
- data/spec/data/contig_printer/1/contigs.fa +4 -0
- data/spec/data/contig_printer/1/seq.fa +2408 -0
- data/spec/data/contig_printer/1/seq.fa.svg +153 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Graph2 +2953 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/LastGraph +2953 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Log +21 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/PreGraph +27 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Roadmaps +5182 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/Sequences +3612 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/contigs.fa +36 -0
- data/spec/data/contig_printer/1/seq.fa.velvet/stats.txt +14 -0
- data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam +0 -0
- data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam.bai +0 -0
- data/spec/data/contig_printer/1/seq.node12.fa +4 -0
- data/spec/data/contig_printer/1/seq1_1to550.fa +2 -0
- data/spec/data/contig_printer/1/seq2_1to550.fa +2 -0
- data/spec/data/contig_printer/1/seq2_1to550.fa.fai +1 -0
- data/spec/data/explore/1/2seqs.sammy.fa +12004 -0
- data/spec/data/explore/1/HOWTO_RECREATE.txt +6 -0
- data/spec/data/explore/1/a.fa +2 -0
- data/spec/data/explore/1/seq1_and_a.fa +3 -0
- data/spec/data/explore/1/seq2.fa +2 -0
- data/spec/data/fluff/1/2seqs.sammy.fa +12004 -0
- data/spec/data/fluff/1/HOWTO_RECREATE.txt +5 -0
- data/spec/data/fluff/1/seq1.fa +2 -0
- data/spec/data/fluff/1/seq2.fa +2 -0
- data/spec/data/gapfilling/1/reads.fa +171 -0
- data/spec/data/gapfilling/1/trail_with_Ns.fa +5 -0
- data/spec/data/gapfilling/1/velvetAssembly/Graph2 +130 -0
- data/spec/data/gapfilling/1/velvetAssembly/LastGraph +130 -0
- data/spec/data/gapfilling/1/velvetAssembly/Log +199 -0
- data/spec/data/gapfilling/1/velvetAssembly/PreGraph +7 -0
- data/spec/data/gapfilling/1/velvetAssembly/Roadmaps +239 -0
- data/spec/data/gapfilling/1/velvetAssembly/Sequences +281 -0
- data/spec/data/gapfilling/1/velvetAssembly/contigs.fa +12 -0
- data/spec/data/gapfilling/1/velvetAssembly/stats.txt +4 -0
- data/spec/data/gapfilling/2/HOWTO_recreate +17 -0
- data/spec/data/gapfilling/2/reference.fa +2 -0
- data/spec/data/gapfilling/2/reference_part1.fa +4 -0
- data/spec/data/gapfilling/2/reference_part2.fa +4 -0
- data/spec/data/gapfilling/2/sammy_reads.fa.gz +0 -0
- data/spec/data/gapfilling/2/with_gaps.fa +4 -0
- data/spec/data/gapfilling/3/HOWTO_recreate +4 -0
- data/spec/data/gapfilling/3/reads.fa.gz +0 -0
- data/spec/data/gapfilling/3/reference_part1.fa +4 -0
- data/spec/data/gapfilling/3/reference_part2.fa +4 -0
- data/spec/data/gapfilling/3/with_gaps.fa +4 -0
- data/spec/data/gapfilling/4/HOWTO_recreate +1 -0
- data/spec/data/gapfilling/4/reads.fa.gz +0 -0
- data/spec/data/gapfilling/5/HOWTO_RECREATE +7 -0
- data/spec/data/gapfilling/5/answer.fna +2 -0
- data/spec/data/gapfilling/5/gappy.fna +2 -0
- data/spec/data/gapfilling/5/reads.fa +17961 -0
- data/spec/data/gapfilling/5/velvet51_3.5/LastGraph +8337 -0
- data/spec/data/gapfilling/5/velvet51_3.5/Sequences +20921 -0
- data/spec/data/gapfilling/6/random1.fa +28 -0
- data/spec/data/gapfilling/6/random2.fa +28 -0
- data/spec/data/gapfilling/6/random_sequence_length_2000 +0 -0
- data/spec/data/gapfilling/6/reads.random1.fa.gz +0 -0
- data/spec/data/gapfilling/6/reads.random2.fa.gz +0 -0
- data/spec/data/gapfilling/6/to_gapfill.fa +22 -0
- data/spec/data/kmer_profile_to_assembly/multiple_abundance_file1.csv +2 -0
- data/spec/data/kmers_count1.csv +2 -0
- data/spec/data/kmers_count2.csv +3 -0
- data/spec/data/out +3 -0
- data/spec/data/positive_latching_pair.fa +2 -0
- data/spec/data/primers.csv +4 -0
- data/spec/data/read_selection_by_kmer/blacklist1.txt +1 -0
- data/spec/data/read_selection_by_kmer/input.fasta +6 -0
- data/spec/data/read_selection_by_kmer/whitelist1.txt +1 -0
- data/spec/data/read_selection_by_kmer/whitelist2.txt +2 -0
- data/spec/data/read_to_node/1_a_graph/HOWTO_RECREATE.txt +2 -0
- data/spec/data/read_to_node/1_a_graph/LastGraph +6695 -0
- data/spec/data/read_to_node/1_a_graph/ReadToNode.bin +0 -0
- data/spec/data/read_to_node/2_no_read256_or_259/HOWTO_RECREATE.txt +3 -0
- data/spec/data/read_to_node/2_no_read256_or_259/LastGraph +6693 -0
- data/spec/data/read_to_node/2_no_read256_or_259/ReadToNode.bin +0 -0
- data/spec/data/read_to_node/3_no_last_read/LastGraph +6694 -0
- data/spec/data/read_to_node/3_no_last_read/ReadToNode.bin +0 -0
- data/spec/data/t/details.txt +5 -0
- data/spec/data/t/details.txt.srt +5 -0
- data/spec/data/t/location.txt +3 -0
- data/spec/data/t/location.txt.srt +3 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/answer.fa +2 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/reads.fa.gz +0 -0
- data/spec/data/tweak/1_gap_then_unscaffolded/scaffolds.fa +6 -0
- data/spec/data/tweak/2_second_genome/answer2.fa +2 -0
- data/spec/data/tweak/2_second_genome/reads.fa.gz +0 -0
- data/spec/data/tweak/3_variant/answer.fa +2 -0
- data/spec/data/tweak/3_variant/lesser_answer.fa +2 -0
- data/spec/data/tweak/3_variant/reads.fa.gz +0 -0
- data/spec/data/tweak/3_variant/with_gaps.fa +2 -0
- data/spec/data/velvet_test_trails/Assem/Graph +17 -0
- data/spec/data/velvet_test_trails/Assem/Graph2 +40 -0
- data/spec/data/velvet_test_trails/Assem/LastGraph +40 -0
- data/spec/data/velvet_test_trails/Assem/Log +35 -0
- data/spec/data/velvet_test_trails/Assem/PreGraph +9 -0
- data/spec/data/velvet_test_trails/Assem/Roadmaps +89 -0
- data/spec/data/velvet_test_trails/Assem/Sequences +50 -0
- data/spec/data/velvet_test_trails/Assem/a.svg +53 -0
- data/spec/data/velvet_test_trails/Assem/contigs.fa +15 -0
- data/spec/data/velvet_test_trails/Assem/stats.txt +5 -0
- data/spec/data/velvet_test_trails/node_fwds.fa +8 -0
- data/spec/data/velvet_test_trails/node_seqs.fa +9 -0
- data/spec/data/velvet_test_trails/nodes_fwd_rev.fa +16 -0
- data/spec/data/velvet_test_trails/read1.fa +2 -0
- data/spec/data/velvet_test_trails/reads.fa +50 -0
- data/spec/data/velvet_test_trails_reverse/Assem/LastGraph +17 -0
- data/spec/data/velvet_test_trails_reverse/Assem/a.svg +53 -0
- data/spec/data/velvet_test_trails_reverse/reads_reversed.fa +10 -0
- data/spec/data/visualise/1/LastGraph +6695 -0
- data/spec/data/visualise/2_paired_end/HOWTO_RECREATE.txt +10 -0
- data/spec/data/visualise/2_paired_end/rand1.fa +2 -0
- data/spec/data/visualise/2_paired_end/rand2.fa +2 -0
- data/spec/data/visualise/2_paired_end/with_gaps.fa +8 -0
- data/spec/data/visualise/2_paired_end/with_gaps.read_pairs.fa.gz +0 -0
- data/spec/data/wander/1/random1.fa +2 -0
- data/spec/data/wander/1/random1.sammy.fa +804 -0
- data/spec/depth_first_search_spec.rb +190 -0
- data/spec/dijkstra_spec.rb +143 -0
- data/spec/explore_spec.rb +29 -0
- data/spec/fluffer_spec.rb +155 -0
- data/spec/gapfiller_spec.rb +107 -0
- data/spec/graph_explorer_spec.rb +475 -0
- data/spec/graph_generator_spec.rb +99 -0
- data/spec/height_finder_spec.rb +306 -0
- data/spec/kmer_abundance_pattern_spec.rb +56 -0
- data/spec/kmer_coverage_based_path_filter_spec.rb +73 -0
- data/spec/kmer_profile_finder_spec.rb +38 -0
- data/spec/kmers_count_tabulate_spec.rb +120 -0
- data/spec/oriented_node_trail_spec.rb +221 -0
- data/spec/paired_end_neighbours_spec.rb +126 -0
- data/spec/paths_between_nodes_spec.rb +349 -0
- data/spec/priner_spec.rb +7 -0
- data/spec/read_input_spec.rb +23 -0
- data/spec/read_selection_by_kmer_spec.rb +166 -0
- data/spec/read_to_node_spec.rb +35 -0
- data/spec/roundup_spec.rb +366 -0
- data/spec/scaffold_breaker_spec.rb +144 -0
- data/spec/sequence_spec.rb +43 -0
- data/spec/single_coherent_paths_between_nodes_spec.rb +492 -0
- data/spec/single_coherent_wanderer_spec.rb +120 -0
- data/spec/single_ended_assembler_spec.rb +398 -0
- data/spec/spec_helper.rb +310 -0
- data/spec/velvet_graph_sequence_extractor_spec.rb +80 -0
- data/spec/visualise_spec.rb +105 -0
- data/spec/wander_spec.rb +119 -0
- data/spec/watch_for_changes.sh +16 -0
- data/validation/fasta_compare.rb +72 -0
- data/validation/gapfill_simulate_perfect.rb +108 -0
- metadata +899 -0
@@ -0,0 +1,176 @@
|
|
1
|
+
require 'optparse'
|
2
|
+
require 'tempfile'
|
3
|
+
|
4
|
+
require 'bio-logger'
|
5
|
+
require 'bio'
|
6
|
+
require 'progressbar'
|
7
|
+
require 'bio-ipcress'
|
8
|
+
$:.unshift File.join(ENV['HOME'],'git','bioruby-primer3','lib')
|
9
|
+
require 'bio-primer3'
|
10
|
+
|
11
|
+
class Bio::FinishM::Primers::Checker
|
12
|
+
include Bio::FinishM::Logging
|
13
|
+
|
14
|
+
def add_options(optparse_object, options)
|
15
|
+
optparse_object.banner = "\nUsage: finishm primers_check --primers <primers_file> [options]
|
16
|
+
|
17
|
+
Check that each pair of primers in the primers fil is compatible.
|
18
|
+
\n\n"
|
19
|
+
|
20
|
+
options.merge!({
|
21
|
+
:logger => 'stderr',
|
22
|
+
:logger_trace_level => 'info',
|
23
|
+
:melting_temperature_optimum => 56,
|
24
|
+
:melting_temperature_tolerance => 2,
|
25
|
+
:extra_global_primer3_options => {
|
26
|
+
'PRIMER_MAX_POLY_X' => 4,
|
27
|
+
'PRIMER_EXPLAIN_FLAG' => '1',
|
28
|
+
},
|
29
|
+
:persevere => false,
|
30
|
+
})
|
31
|
+
|
32
|
+
optparse_object.separator "Required arguments:\n\n"
|
33
|
+
optparse_object.on("-p", "--primers PRIMERS_FILE", String, "A file of primers, newline separated [required]") do |arg|
|
34
|
+
options[:primers_file] = arg
|
35
|
+
end
|
36
|
+
|
37
|
+
optparse_object.separator "\nOptional arguments:\n\n"
|
38
|
+
optparse_object.on("--contig-universe FASTA_FILE", String, "All contigs in the mixture [default: unspecified (don't test this)]") do |arg|
|
39
|
+
options[:contig_universe] = arg
|
40
|
+
end
|
41
|
+
optparse_object.on("--optimum-melting-temperature TEMPERATURE", Integer, "Primers aim for this melting temperature [default: default in primer3 (currently #{options[:melting_temperature_optimum]}C)]") do |arg|
|
42
|
+
options[:melting_temperature_optimum] = arg.to_i
|
43
|
+
raise Exception, " has to be greater than 0, found #{arg}" unless options[:melting_temperature_optimum] > 0
|
44
|
+
end
|
45
|
+
optparse_object.on("--primer3-options OPTION_LIST", "Give extra instructions to Primer3 [default <none>]. Acceptable values can be found in the primer3 manual e.g. 'PRIMER_MAX_POLY_X=4;PRIMER_MAX_SIZE=22' will specify those 2 parameters to primer3. Argument names are auto-capitalised so 'primer_max_poly_X=4;primer_max_size=22'is equivalent.") do |arg|
|
46
|
+
options[:extra_global_primer3_options] = {}
|
47
|
+
arg.split(';').each do |a2|
|
48
|
+
splits = a2.split('=')
|
49
|
+
unless splits.length == 2
|
50
|
+
raise "Unexpected format of the --primer3-options flag, specifically couldn't parse this part: '#{a2}'"
|
51
|
+
end
|
52
|
+
options[:extra_global_primer3_options][splits[0].upcase]=splits[1]
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
def validate_options(options, argv)
|
58
|
+
if argv.length != 0
|
59
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
60
|
+
else
|
61
|
+
[
|
62
|
+
:primers_file,
|
63
|
+
].each do |sym|
|
64
|
+
if options[sym].nil?
|
65
|
+
return "No option found to specify #{sym}."
|
66
|
+
end
|
67
|
+
end
|
68
|
+
return nil
|
69
|
+
end
|
70
|
+
end
|
71
|
+
|
72
|
+
def run(options, argv)
|
73
|
+
Bio::Log::CLI.configure('bio-primer3')
|
74
|
+
found_error = false
|
75
|
+
|
76
|
+
# Read the primers in
|
77
|
+
primers = []
|
78
|
+
File.foreach(options[:primers_file]) do |line|
|
79
|
+
line.strip!
|
80
|
+
next if line.empty?
|
81
|
+
unless line.match(/^[atgc]+$/i)
|
82
|
+
raise "Malformed primer sequence '#{line}' found in the file - I'm just after one primer per line, that's all."
|
83
|
+
end
|
84
|
+
primers.push line
|
85
|
+
end
|
86
|
+
log.info "Read in #{primers.length} primers e.g. #{primers[0] }"
|
87
|
+
raise "Need at least 2 primers!" unless primers.length >= 2
|
88
|
+
|
89
|
+
|
90
|
+
primer3_options = options[:extra_global_primer3_options]
|
91
|
+
primer3_options.merge!({
|
92
|
+
'PRIMER_OPT_TM' => options[:melting_temperature_optimum],
|
93
|
+
'PRIMER_MIN_TM' => options[:melting_temperature_optimum]-options[:melting_temperature_tolerance],
|
94
|
+
'PRIMER_MAX_TM' => options[:melting_temperature_optimum]+options[:melting_temperature_tolerance],
|
95
|
+
})
|
96
|
+
|
97
|
+
# Check to make sure there is no incompatibilities primer vs primer:
|
98
|
+
num_compared = check_primer_compatibilities(primers, primer3_options)
|
99
|
+
if num_compared == false
|
100
|
+
log.error "Found at least one incompatible primer set, giving up."
|
101
|
+
exit 1
|
102
|
+
end
|
103
|
+
log.info "Validated #{num_compared} different pairs of primers, they don't seem to conflict with each other at all, according to primer3's check primers thing"
|
104
|
+
|
105
|
+
|
106
|
+
# Check the contig universe
|
107
|
+
if options[:contig_universe]
|
108
|
+
log.info "in-silico PCR: testing the contig universe. This could probably be sped up by only making a single call to ipcress, but oh well."
|
109
|
+
num_universe_hits = check_contig_universe(primers, options[:contig_universe])
|
110
|
+
if num_universe_hits > 0
|
111
|
+
log.warn "Found #{results.length} matches between #{primer1} and #{primer2} in the contig universe, expected none."
|
112
|
+
else
|
113
|
+
log.info "No sequence appears to be amplified with any two primers in the background set of contigs (this is a good thing)."
|
114
|
+
end
|
115
|
+
else
|
116
|
+
log.info "Not checking to see if primers match any other contigs not targeted are picked up by these primers, because no universe of contigs was specified."
|
117
|
+
end
|
118
|
+
|
119
|
+
|
120
|
+
if found_error
|
121
|
+
log.warn "At least one problem detected in the primer checking process"
|
122
|
+
else
|
123
|
+
log.warn "No primer issues detected"
|
124
|
+
end
|
125
|
+
end
|
126
|
+
|
127
|
+
|
128
|
+
|
129
|
+
# Each pair of primers should be compatible
|
130
|
+
def check_primer_compatibilities(primers, primer3_options)
|
131
|
+
# First just make sure that everything is ok here
|
132
|
+
log.info "Double checking to make sure there is no incompatibilities between primer pairs"
|
133
|
+
num_compared = 0
|
134
|
+
primers.to_a.combination(2) do |array|
|
135
|
+
primer1 = array[0]
|
136
|
+
primer2 = array[1]
|
137
|
+
|
138
|
+
compatible, result = Bio::Primer3.test_primer_compatibility(primer1, primer2, primer3_options, :return_result => true)
|
139
|
+
num_compared += 1
|
140
|
+
log.debug "Primer3 returned: #{result.inspect}" if log.debug?
|
141
|
+
|
142
|
+
if compatible == false
|
143
|
+
log.warn "Found an incompatibility between #{primer1} and #{primer2}"
|
144
|
+
log.warn "Primer3 output was: #{result.inspect}"
|
145
|
+
return false
|
146
|
+
end
|
147
|
+
end
|
148
|
+
return num_compared
|
149
|
+
end
|
150
|
+
|
151
|
+
|
152
|
+
|
153
|
+
|
154
|
+
# Does any primer pairs amplify anything in a large set of background (universe) contigs?
|
155
|
+
# They shouldn't.
|
156
|
+
def check_contig_universe(primers, contig_universe)
|
157
|
+
num_hits = 0
|
158
|
+
num_compared = 0
|
159
|
+
|
160
|
+
primers.to_a.combination(2) do |array|
|
161
|
+
primer1 = array[0]
|
162
|
+
primer2 = array[1]
|
163
|
+
num_compared += 1
|
164
|
+
|
165
|
+
primer_set = Bio::Ipcress::PrimerSet.new primer1, primer2
|
166
|
+
results = Bio::Ipcress.run primer_set, options[:contig_universe], ipcress_options
|
167
|
+
|
168
|
+
if results.length > 0
|
169
|
+
num_hits += 1
|
170
|
+
log.warn "Found #{results.length} matches between #{primer1} and #{primer2} in the contig universe, expected none."
|
171
|
+
end
|
172
|
+
end
|
173
|
+
|
174
|
+
log.debug "Tested #{num_compared} pairs of primers to see if anything else was hit, warnings above if any did so"
|
175
|
+
end
|
176
|
+
end
|
@@ -0,0 +1,344 @@
|
|
1
|
+
class Bio::FinishM::RoundUp
|
2
|
+
include Bio::FinishM::Logging
|
3
|
+
|
4
|
+
DEFAULT_OPTIONS = {
|
5
|
+
:contig_end_length => 200,
|
6
|
+
:graph_search_leash_length => 20000,
|
7
|
+
:unscaffold_first => false,
|
8
|
+
:recoherence_kmer => 1,
|
9
|
+
:debug => false,
|
10
|
+
:gapfill_only => false,
|
11
|
+
:max_explore_nodes => 10000,
|
12
|
+
:max_gapfill_paths => 10,
|
13
|
+
}
|
14
|
+
|
15
|
+
def add_options(optparse_object, options)
|
16
|
+
optparse_object.banner = "\nUsage: finishm roundup --genomes <genome1.fasta>[,<genome2.fasta>,...] --fastq-gz <reads..> --output-directory <directory>
|
17
|
+
|
18
|
+
Takes one or more genomes and tries to improve their quality by reducing the number of
|
19
|
+
scaffolds and N characters they contain.
|
20
|
+
|
21
|
+
Example:
|
22
|
+
|
23
|
+
finishm roundup --genomes genome1.fasta,genome2.fasta --fastq-gz reads.1.fq.gz,reads.2.fq.gz --output-directory finishm_roundup_results
|
24
|
+
|
25
|
+
That will create a collapsed de-Bruijn graph from reads.1.fq.gz and reads.2.fq.gz, then try to find connections between
|
26
|
+
the starts and the ends of the contigs in genome1.fasta. If any connections between contigs are mutually exclusive,
|
27
|
+
then they are incorporated into scaffolds together, and gapfilling is attempted. The final sequences are output in
|
28
|
+
the finishm_roundup_results directory in FASTA format. The procedure is then repeated for genome2.fasta.
|
29
|
+
|
30
|
+
\n\n"
|
31
|
+
|
32
|
+
options.merge!(DEFAULT_OPTIONS)
|
33
|
+
|
34
|
+
optparse_object.separator "\nRequired arguments:\n\n"
|
35
|
+
optparse_object.on("--genomes FASTA_1[,FASTA_2...]", Array, "fasta files of genomes to be improved [required]") do |arg|
|
36
|
+
options[:assembly_files] = arg
|
37
|
+
end
|
38
|
+
optparse_object.on("--output-directory PATH", "Output results to this directory [required]") do |arg|
|
39
|
+
options[:output_directory] = arg
|
40
|
+
end
|
41
|
+
|
42
|
+
optparse_object.separator "\nThere must be some definition of reads too:\n\n" #TODO improve this help
|
43
|
+
Bio::FinishM::ReadInput.new.add_options(optparse_object, options)
|
44
|
+
|
45
|
+
optparse_object.separator "\nOptional arguments:\n\n"
|
46
|
+
optparse_object.on("--overhang NUM", Integer, "Start assembling this far from the ends of the contigs [default: #{options[:contig_end_length] }]") do |arg|
|
47
|
+
options[:contig_end_length] = arg.to_i
|
48
|
+
end
|
49
|
+
optparse_object.on("--recoherence-kmer NUM", Integer, "Use a kmer longer than the original velvet one, to help remove bubbles and circular paths [default: none]") do |arg|
|
50
|
+
options[:recoherence_kmer] = arg
|
51
|
+
end
|
52
|
+
optparse_object.on("--leash-length NUM", Integer, "Don't explore too far in the graph, only this far and not much more [default: #{options[:graph_search_leash_length] }]") do |arg|
|
53
|
+
options[:graph_search_leash_length] = arg
|
54
|
+
end
|
55
|
+
optparse_object.on("--unscaffold-first", "Break the scaffolds in the contigs file apart, and then wander between the resultant contigs. This option is only relevant to the wander step; gapfilling is attempted on all gaps by default. [default: #{options[:unscaffold_first] }]") do
|
56
|
+
options[:unscaffold_first] = true
|
57
|
+
end
|
58
|
+
optparse_object.on("--gapfill-only", "Don't wander, just gapfill [default: #{options[:gapfill_only] }]") do
|
59
|
+
options[:gapfill_only] = true
|
60
|
+
end
|
61
|
+
optparse_object.on("--max-gapfill-paths NUM", Integer, "When this number of paths is exceeded, don't gapfill, print as Ns [default: #{options[:max_gapfill_paths] }]") do |arg|
|
62
|
+
options[:max_gapfill_paths] = arg
|
63
|
+
end
|
64
|
+
optparse_object.on("--max-explore-nodes NUM", Integer, "Only explore this many nodes. If max is reached, do not make connections. [default: #{options[:max_explore_nodes] }]") do |arg|
|
65
|
+
options[:max_explore_nodes] = arg
|
66
|
+
end
|
67
|
+
optparse_object.on("--debug", "Build the graph, then drop to a pry console. [default: #{options[:debug] }]") do
|
68
|
+
options[:debug] = true
|
69
|
+
end
|
70
|
+
optparse_object.on("--probe NUM",Integer,"debug mode: explore from this probe number (1-based index), gapfill only, no wander. [default: explore from all probes}]") do |arg|
|
71
|
+
options[:interesting_probes] = [arg-1] #convert to 0-based index
|
72
|
+
options[:gapfill_only] = true
|
73
|
+
end
|
74
|
+
#optparse_object.on("--proceed-on-short-contigs", "By default, when overly short contigs are encountered, finishm croaks. This option stops the croaking [default: #{options[:proceed_on_short_contigs] }]") do
|
75
|
+
# options[:proceed_on_short_contigs] = true
|
76
|
+
#end
|
77
|
+
|
78
|
+
Bio::FinishM::GraphGenerator.new.add_options optparse_object, options
|
79
|
+
end
|
80
|
+
|
81
|
+
def validate_options(options, argv)
|
82
|
+
#TODO: give a better description of the error that has occurred
|
83
|
+
#TODO: require reads options
|
84
|
+
if argv.length != 0
|
85
|
+
return "Dangling argument(s) found e.g. #{argv[0] }"
|
86
|
+
else
|
87
|
+
[
|
88
|
+
:assembly_files,
|
89
|
+
:output_directory,
|
90
|
+
].each do |sym|
|
91
|
+
if options[sym].nil?
|
92
|
+
return "No option found to specify #{sym}."
|
93
|
+
end
|
94
|
+
end
|
95
|
+
|
96
|
+
#if return nil from here, options all were parsed successfully
|
97
|
+
return Bio::FinishM::ReadInput.new.validate_options(options, [])
|
98
|
+
end
|
99
|
+
end
|
100
|
+
|
101
|
+
def run(options, argv=[])
|
102
|
+
# Make sure output directory is writeable to avoid late croaking
|
103
|
+
output_directory = setup_output_directory options[:output_directory]
|
104
|
+
|
105
|
+
# Gather the probes from each genome supplied
|
106
|
+
genomes = Bio::FinishM::InputGenome.parse_genome_fasta_files(
|
107
|
+
options[:assembly_files],
|
108
|
+
options[:contig_end_length],
|
109
|
+
options
|
110
|
+
)
|
111
|
+
|
112
|
+
# Generate one velvet assembly to rule them all (forging the assembly is hard work..)
|
113
|
+
probe_sequences = genomes.collect{|genome| genome.probe_sequences}.flatten
|
114
|
+
# Generate the graph with the probe sequences in it.
|
115
|
+
read_input = Bio::FinishM::ReadInput.new
|
116
|
+
read_input.parse_options options
|
117
|
+
master_graph = Bio::FinishM::GraphGenerator.new.generate_graph(probe_sequences, read_input, options)
|
118
|
+
|
119
|
+
binding.pry if options[:debug]
|
120
|
+
|
121
|
+
# For each genome, wander, gapfill, then output
|
122
|
+
wanderer = Bio::FinishM::Wanderer.new
|
123
|
+
gapfiller = Bio::FinishM::GapFiller.new
|
124
|
+
printer = Bio::AssemblyGraphAlgorithms::ContigPrinter.new
|
125
|
+
genomes.each do |genome|
|
126
|
+
# wander using just the probes on the ends of the scaffolds
|
127
|
+
connected_scaffolds = nil
|
128
|
+
all_connections = []
|
129
|
+
gaps_filled_in_genome = 0
|
130
|
+
wandered_probe_indices = nil
|
131
|
+
|
132
|
+
File.open(File.join(output_directory, File.basename(genome.filename)+".report.txt"),'w') do |report|
|
133
|
+
report.puts "#{Time.now} FinishM report for roundup run with: #{options.inspect}"
|
134
|
+
|
135
|
+
if options[:gapfill_only]
|
136
|
+
log.info "Skipping wander, gapfilling only"
|
137
|
+
connected_scaffolds = Bio::FinishM::ConnectionInterpreter.new([], (0...genome.scaffolds.length)).scaffolds([])
|
138
|
+
else
|
139
|
+
log.debug "Wandering.."
|
140
|
+
connected_scaffolds, all_connections, wandered_probe_indices = wander_a_genome(wanderer, genome, master_graph, options, report)
|
141
|
+
end
|
142
|
+
|
143
|
+
# Write out all the connections
|
144
|
+
File.open(File.join(output_directory, File.basename(genome.filename)+".connections.csv"),'w') do |con_file|
|
145
|
+
all_connections.each do |connection|
|
146
|
+
con_file.puts connection
|
147
|
+
end
|
148
|
+
end
|
149
|
+
|
150
|
+
output_path = File.join(output_directory, File.basename(genome.filename)+".scaffolds.fasta")
|
151
|
+
variants_path = File.join(output_directory, File.basename(genome.filename)+".at_least_half_completely_wrong.vcf")
|
152
|
+
num_circular_scaffolds = 0
|
153
|
+
|
154
|
+
File.open(output_path, 'w') do |output_file|
|
155
|
+
File.open(variants_path,'w') do |variants_file|
|
156
|
+
variants_file.puts %w(#CHROM POS ID REF ALT QUAL FILTER INFO).join("\t")
|
157
|
+
# gapfill between
|
158
|
+
# (1) interpreted_connections
|
159
|
+
# (2) gaps that were present before above wander
|
160
|
+
connected_scaffolds.each_with_index do |cross_scaffold_connection, connected_scaffold_index|
|
161
|
+
superscaffold_name = "scaffold#{connected_scaffold_index+1}"
|
162
|
+
|
163
|
+
pretend_contig = cross_scaffold_connection.contigs[0]
|
164
|
+
first_scaffold_index = pretend_contig.sequence_index
|
165
|
+
|
166
|
+
# Gapfill contigs within the scaffold on the extreme LHS
|
167
|
+
scaffold_sequence, num_gaps, variants = gapfill_a_scaffold(gapfiller, printer, master_graph, genome, first_scaffold_index, pretend_contig.direction, superscaffold_name, report, variants_file, options)
|
168
|
+
gaps_filled_in_genome += num_gaps
|
169
|
+
|
170
|
+
last_contig = nil
|
171
|
+
cross_scaffold_connection.contigs.each_with_index do |contig, superscaffold_index|
|
172
|
+
unless last_contig.nil? #skip the first contig - it be done
|
173
|
+
last_name = genome.scaffolds[last_contig.sequence_index].name
|
174
|
+
current_name = genome.scaffolds[contig.sequence_index].name
|
175
|
+
log.debug "Connecting #{last_name} and #{current_name}" if log.debug?
|
176
|
+
|
177
|
+
# Ready the contig on the RHS of this join
|
178
|
+
# Gapfill within the scaffold on the RHS of the new gap
|
179
|
+
rhs_sequence, num_gaps, variants = gapfill_a_scaffold(gapfiller, printer, master_graph, genome, contig.sequence_index, contig.direction, superscaffold_name, report, variants_file, options)
|
180
|
+
gaps_filled_in_genome += num_gaps
|
181
|
+
|
182
|
+
# Gapfill across the new gap between scaffolds
|
183
|
+
aconn = gapfiller.gapfill(master_graph,
|
184
|
+
last_contig.direction == true ? genome.last_probe(last_contig.sequence_index).index : genome.first_probe(last_contig.sequence_index).index,
|
185
|
+
contig.direction == true ? genome.first_probe(contig.sequence_index).index : genome.last_probe(contig.sequence_index).index,
|
186
|
+
options
|
187
|
+
)
|
188
|
+
second_sequence = genome.scaffolds[contig.sequence_index].contigs[0].sequence
|
189
|
+
log.debug "Found #{aconn.paths.length} connections between #{last_name} and #{current_name}" if log.debug?
|
190
|
+
if aconn.paths.length == 0
|
191
|
+
# when this occurs, it is due to there being a circuit in the path, so no paths are printed.
|
192
|
+
# (at least for now) TODO: this could be improved.
|
193
|
+
# Just arbitrarily put in 100 N characters, to denote a join, but no gapfill
|
194
|
+
scaffold_sequence = scaffold_sequence+('N'*100)+rhs_sequence
|
195
|
+
else
|
196
|
+
scaffold_sequence, variants = printer.ready_two_contigs_and_connections(
|
197
|
+
master_graph.graph,
|
198
|
+
scaffold_sequence,
|
199
|
+
aconn,
|
200
|
+
rhs_sequence,
|
201
|
+
master_graph.velvet_sequences
|
202
|
+
)
|
203
|
+
# Print variants
|
204
|
+
# TODO: need to change coordinates of variants, particularly when >2 contigs are joined?
|
205
|
+
variants.each do |variant|
|
206
|
+
variant.reference_name = superscaffold_name
|
207
|
+
variants_file.puts variant.vcf(scaffold_sequence)
|
208
|
+
end
|
209
|
+
end
|
210
|
+
end
|
211
|
+
last_contig = contig
|
212
|
+
end
|
213
|
+
|
214
|
+
#Output the scaffold to the output directory
|
215
|
+
descriptor = nil
|
216
|
+
if cross_scaffold_connection.circular?
|
217
|
+
descriptor = 'circular'
|
218
|
+
num_circular_scaffolds += 1
|
219
|
+
else
|
220
|
+
descriptor = 'scaffold'
|
221
|
+
end
|
222
|
+
scaffold_names = cross_scaffold_connection.contigs.collect do |contig|
|
223
|
+
genome.scaffolds[contig.sequence_index].name
|
224
|
+
end
|
225
|
+
output_file.puts ">#{superscaffold_name} #{descriptor} #{scaffold_names.join(':') }"
|
226
|
+
output_file.puts scaffold_sequence
|
227
|
+
end
|
228
|
+
|
229
|
+
num_connected_scaffolds = genome.scaffolds.length - connected_scaffolds.length
|
230
|
+
log.info "Wrote #{connected_scaffolds.length} scaffolds to #{output_path}, after scaffolding #{num_connected_scaffolds} scaffolds together (#{num_circular_scaffolds} were circular) and filling #{gaps_filled_in_genome} gaps."
|
231
|
+
end
|
232
|
+
end
|
233
|
+
end
|
234
|
+
end
|
235
|
+
end
|
236
|
+
|
237
|
+
def setup_output_directory(given_directory)
|
238
|
+
output_directory = File.absolute_path(given_directory)
|
239
|
+
log.debug "Using output directory: #{output_directory}" if log.debug?
|
240
|
+
|
241
|
+
if File.exist?(output_directory)
|
242
|
+
if !File.directory?(output_directory)
|
243
|
+
log.error "Specified --output-directory #{output_directory} exists but is a file and not a directory. Cannot continue."
|
244
|
+
exit 1
|
245
|
+
elsif !File.writable?(output_directory)
|
246
|
+
log.error "Specified --output-directory #{output_directory} is not writeable. Cannot continue."
|
247
|
+
exit 1
|
248
|
+
else
|
249
|
+
log.debug "Already existing output directory #{output_directory} seems usable"
|
250
|
+
end
|
251
|
+
else
|
252
|
+
# Creating a new output directory
|
253
|
+
Dir.mkdir(output_directory)
|
254
|
+
end
|
255
|
+
|
256
|
+
return output_directory
|
257
|
+
end
|
258
|
+
|
259
|
+
def wander_a_genome(wanderer, genome, master_probed_graph, options, report)
|
260
|
+
# Create new finishm_graph with only probes from the ends of the scaffolds of this genome
|
261
|
+
probe_indices = []
|
262
|
+
genome.each_scaffold_end_numbered_probe{|probe| probe_indices.push(probe.number)}
|
263
|
+
genome_graph = master_probed_graph.subgraph(probe_indices)
|
264
|
+
|
265
|
+
num_scaffolds = genome.scaffolds.length
|
266
|
+
|
267
|
+
all_connections = wanderer.probed_graph_to_connections(genome_graph, options)
|
268
|
+
|
269
|
+
interpreter = Bio::FinishM::ConnectionInterpreter.new(all_connections, (0...num_scaffolds))
|
270
|
+
connections = interpreter.doubly_single_contig_connections
|
271
|
+
report.puts "Found #{connections.length} connections between contigs that can be used for scaffolding"
|
272
|
+
unconnected_probes = interpreter.unconnected_probes
|
273
|
+
report.puts "Found #{unconnected_probes.length} contig ends that did not connect to any others"
|
274
|
+
unconnected_probes.each do |probe|
|
275
|
+
report.puts "Did not connect to any other probes: #{probe.inspect}"
|
276
|
+
end
|
277
|
+
|
278
|
+
return interpreter.scaffolds(connections), all_connections, probe_indices
|
279
|
+
end
|
280
|
+
|
281
|
+
def gapfill_a_scaffold(gapfiller, printer, master_graph, genome, scaffold_index, scaffold_direction, superscaffold_name, report, variants_file, options)
|
282
|
+
connections = []
|
283
|
+
genome.each_gap_probe_pair(scaffold_index) do |probe1, probe2|
|
284
|
+
log.info "Gapfilling between probes #{probe1.number+1} and #{probe2.number+1}.."
|
285
|
+
next unless options[:interesting_probes].nil? or
|
286
|
+
options[:interesting_probes].include?(probe1.number) or
|
287
|
+
options[:interesting_probes].include?(probe2.number)
|
288
|
+
connections.push gapfiller.gapfill(master_graph, probe1.index, probe2.index, options)
|
289
|
+
end
|
290
|
+
log.debug "Found #{connections.length} connections" if log.debug?
|
291
|
+
|
292
|
+
all_variants = []
|
293
|
+
num_gapfills = 0
|
294
|
+
scaffold = genome.scaffolds[scaffold_index]
|
295
|
+
gapfilled_sequence = genome.scaffolds[scaffold_index].contigs[0].sequence
|
296
|
+
connections.each_with_index do |aconn, i|
|
297
|
+
rhs_sequence = scaffold.contigs[i+1].sequence
|
298
|
+
gapfilled_sequence, variants, gapfilled = piece_together_gapfill(
|
299
|
+
printer, master_graph, gapfilled_sequence, aconn, rhs_sequence, genome.gap_length(scaffold_index, i),
|
300
|
+
options[:max_gapfill_paths]
|
301
|
+
)
|
302
|
+
if gapfilled
|
303
|
+
num_gapfills += 1
|
304
|
+
variants.each{|v| all_variants << v}
|
305
|
+
to_log = "Filled a gap on genome #{genome.filename}: scaffold #{scaffold.name}: #{scaffold.contigs[i].scaffold_position_end+1}-#{scaffold.contigs[i+1].scaffold_position_start-1}"
|
306
|
+
report.puts to_log
|
307
|
+
log.info to_log
|
308
|
+
end
|
309
|
+
end
|
310
|
+
if scaffold_direction == false
|
311
|
+
gapfilled_sequence = revcom(gapfilled_sequence)
|
312
|
+
all_variants.each do |variant|
|
313
|
+
variant.position = gapfilled_sequence.length - variant.position
|
314
|
+
variant.reverse!
|
315
|
+
end
|
316
|
+
end
|
317
|
+
all_variants.each do |variant|
|
318
|
+
variant.reference_name = superscaffold_name
|
319
|
+
variants_file.puts variant.vcf(gapfilled_sequence)
|
320
|
+
end
|
321
|
+
return gapfilled_sequence, num_gapfills, all_variants
|
322
|
+
end
|
323
|
+
|
324
|
+
def piece_together_gapfill(printer, master_graph, first_sequence, aconn, second_sequence, gap_length, max_gapfill_paths)
|
325
|
+
scaffold_sequence = nil
|
326
|
+
gapfilled = -1
|
327
|
+
if aconn.paths.length == 0 or aconn.paths.length > max_gapfill_paths
|
328
|
+
# No paths found. Just fill with Ns like it was before
|
329
|
+
scaffold_sequence = first_sequence + 'N'*gap_length + second_sequence
|
330
|
+
gapfilled = false
|
331
|
+
else
|
332
|
+
scaffold_sequence, variants = printer.ready_two_contigs_and_connections(
|
333
|
+
master_graph.graph, first_sequence, aconn, second_sequence, master_graph.velvet_sequences
|
334
|
+
)
|
335
|
+
gapfilled = true
|
336
|
+
end
|
337
|
+
scaffold_sequence.gsub!('-','') #remove gaps i.e. where the consensus is a gap
|
338
|
+
return scaffold_sequence, variants, gapfilled
|
339
|
+
end
|
340
|
+
|
341
|
+
def revcom(seq)
|
342
|
+
Bio::Sequence::NA.new(seq).reverse_complement.to_s.upcase
|
343
|
+
end
|
344
|
+
end
|