X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=blastdb.cpp;h=70349e50f0264b68fba76c4f04ebbddd0c8c96f6;hb=a150db3c2bfd7b76420048a1e94ebe397f2c6045;hp=50a0401b2cb8780765ba56a3ba2c025380c4e212;hpb=5a1e62397b91f57d0d3aff635891df04b8999a88;p=mothur.git diff --git a/blastdb.cpp b/blastdb.cpp index 50a0401..70349e5 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -14,25 +14,51 @@ /**************************************************************************************************/ -BlastDB::BlastDB(float gO, float gE, float m, float mM) : Database(), +BlastDB::BlastDB(string tag, float gO, float gE, float m, float mM) : Database(), gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) { - globaldata = GlobalData::getInstance(); count = 0; int randNumber = rand(); - dbFileName = toString(randNumber) + ".template.unaligned.fasta"; - queryFileName = toString(randNumber) + ".candidate.unaligned.fasta"; - blastFileName = toString(randNumber) + ".blast"; + dbFileName = tag + toString(randNumber) + ".template.unaligned.fasta"; + queryFileName = tag + toString(randNumber) + ".candidate.unaligned.fasta"; + blastFileName = tag + toString(randNumber) + ".blast"; } +/**************************************************************************************************/ + +BlastDB::BlastDB() : Database() { + try { + count = 0; + + int randNumber = rand(); + dbFileName = toString(randNumber) + ".template.unaligned.fasta"; + queryFileName = toString(randNumber) + ".candidate.unaligned.fasta"; + blastFileName = toString(randNumber) + ".blast"; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "BlastDB"); + exit(1); + } +} /**************************************************************************************************/ BlastDB::~BlastDB(){ - remove(queryFileName.c_str()); // let's clean stuff up and remove the temp files - remove(dbFileName.c_str()); // let's clean stuff up and remove the temp files - remove(blastFileName.c_str()); // let's clean stuff up and remove the temp files + try{ + remove(queryFileName.c_str()); // let's clean stuff up and remove the temp files + remove(dbFileName.c_str()); // let's clean stuff up and remove the temp files + remove((dbFileName+".nsq").c_str()); // let's clean stuff up and remove the temp files + remove((dbFileName+".nsi").c_str()); // let's clean stuff up and remove the temp files + remove((dbFileName+".nsd").c_str()); // let's clean stuff up and remove the temp files + remove((dbFileName+".nin").c_str()); // let's clean stuff up and remove the temp files + remove((dbFileName+".nhr").c_str()); // let's clean stuff up and remove the temp files + remove(blastFileName.c_str()); // let's clean stuff up and remove the temp files + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "~BlastDB"); + exit(1); + } } /**************************************************************************************************/ //assumes you have added all the template sequences using the addSequence function and run generateDB. @@ -41,25 +67,33 @@ vector BlastDB::findClosestSequences(Sequence* seq, int n) { vector topMatches; ofstream queryFile; - openOutputFile(queryFileName, queryFile); + m->openOutputFile((queryFileName+seq->getName()), queryFile); queryFile << '>' << seq->getName() << endl; queryFile << seq->getUnaligned() << endl; queryFile.close(); + // the goal here is to quickly survey the database to find the closest match. To do this we are using the default // wordsize used in megablast. I'm sure we're sacrificing accuracy for speed, but anyother way would take way too // long. With this setting, it seems comparable in speed to the suffix tree approach. - string blastCommand = path + "blast/bin/blastall -p blastn -d " + dbFileName + " -b 1 -m 8 -W 28 -v " + toString(n); - blastCommand += (" -i " + queryFileName + " -o " + blastFileName); + string blastCommand; + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + + blastCommand = path + "blast/bin/blastall -p blastn -d " + dbFileName + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);; + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); + #else + blastCommand = "\"" + path + "blast\\bin\\blastall\" -p blastn -d " + dbFileName + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);; + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); + #endif system(blastCommand.c_str()); ifstream m8FileHandle; - openInputFile(blastFileName, m8FileHandle); + m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error"); string dummy; int templateAccession; - gobble(m8FileHandle); + m->gobble(m8FileHandle); while(!m8FileHandle.eof()){ m8FileHandle >> dummy >> templateAccession >> searchScore; @@ -67,65 +101,80 @@ vector BlastDB::findClosestSequences(Sequence* seq, int n) { //get rest of junk in line while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; } } - gobble(m8FileHandle); + m->gobble(m8FileHandle); topMatches.push_back(templateAccession); } m8FileHandle.close(); - + remove((queryFileName+seq->getName()).c_str()); + remove((blastFileName+seq->getName()).c_str()); + return topMatches; } catch(exception& e) { - errorOut(e, "BlastDB", "findClosestSequences"); + m->errorOut(e, "BlastDB", "findClosestSequences"); exit(1); } } /**************************************************************************************************/ //assumes you have added all the template sequences using the addSequence function and run generateDB. -map BlastDB::findClosest(Sequence* seq, int n) { +vector BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) { try{ - map topMatches; + vector topMatches; + float numBases, mismatch, gap, startQuery, endQuery, startRef, endRef, score; + Scores.clear(); ofstream queryFile; - openOutputFile(queryFileName, queryFile); + + m->openOutputFile((queryFileName+seq->getName()), queryFile); queryFile << '>' << seq->getName() << endl; queryFile << seq->getUnaligned() << endl; queryFile.close(); - +// cout << seq->getUnaligned() << endl; // the goal here is to quickly survey the database to find the closest match. To do this we are using the default // wordsize used in megablast. I'm sure we're sacrificing accuracy for speed, but anyother way would take way too // long. With this setting, it seems comparable in speed to the suffix tree approach. - - string blastCommand = path + "blast/bin/blastall -p blastn -d " + dbFileName + " -m 8 -W 28 -b " + toString(n) + " -v " + toString(n); - blastCommand += (" -i " + queryFileName + " -o " + blastFileName); - system(blastCommand.c_str()); +//7000004128189528left 0 100 66 0 0 1 66 61 126 1e-31 131 + string blastCommand; + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + blastCommand = path + "blast/bin/megablast -e 1e-10 -d " + dbFileName + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); + #else + blastCommand = "\"" + path + "blast\\bin\\megablast\" -e 1e-10 -d " + dbFileName + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); + #endif + system(blastCommand.c_str()); + ifstream m8FileHandle; - openInputFile(blastFileName, m8FileHandle); + m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error"); - string dummy; + string dummy, eScore; int templateAccession; - gobble(m8FileHandle); -//string name = seq->getName(); -//ofstream out; -//openOutputFileAppend(name, out); + m->gobble(m8FileHandle); + while(!m8FileHandle.eof()){ - m8FileHandle >> dummy >> templateAccession >> searchScore; -//out << dummy << '\t' << templateAccession << '\t' << searchScore << endl; - //get rest of junk in line - while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); - //out << c; - if (c == 10 || c == 13){ break; } } + m8FileHandle >> dummy >> templateAccession >> searchScore >> numBases >> mismatch >> gap >> startQuery >> endQuery >> startRef >> endRef >> eScore >> score; +// cout << dummy << '\t' << templateAccession << '\t' << searchScore << '\t' << numBases << '\t' << mismatch << '\t' << gap << '\t' << startQuery << '\t' << endQuery << '\t' << startRef << '\t' << endRef << '\t' << eScore << '\t' << score << endl; - gobble(m8FileHandle); - topMatches[templateAccession] = searchScore; + //get rest of junk in line + //while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; }else{ cout << c; } } // + //cout << endl; + m->gobble(m8FileHandle); + if (searchScore >= minPerID) { + topMatches.push_back(templateAccession); + Scores.push_back(searchScore); + } +//cout << templateAccession << endl; } m8FileHandle.close(); -//out.close(); + remove((queryFileName+seq->getName()).c_str()); + remove((blastFileName+seq->getName()).c_str()); +//cout << "\n" ; return topMatches; } catch(exception& e) { - errorOut(e, "BlastDB", "findClosest"); + m->errorOut(e, "BlastDB", "findClosestMegaBlast"); exit(1); } } @@ -134,17 +183,17 @@ void BlastDB::addSequence(Sequence seq) { try { ofstream unalignedFastaFile; - openOutputFileAppend(dbFileName, unalignedFastaFile); + m->openOutputFileAppend(dbFileName, unalignedFastaFile); // generating a fasta file with unaligned template unalignedFastaFile << '>' << count << endl; // sequences, which will be input to formatdb unalignedFastaFile << seq.getUnaligned() << endl; unalignedFastaFile.close(); - + count++; } catch(exception& e) { - errorOut(e, "BlastDB", "addSequence"); + m->errorOut(e, "BlastDB", "addSequence"); exit(1); } } @@ -152,21 +201,30 @@ void BlastDB::addSequence(Sequence seq) { void BlastDB::generateDB() { try { - //mothurOut("Generating the temporary BLAST database...\t"); cout.flush(); + //m->mothurOut("Generating the temporary BLAST database...\t"); cout.flush(); - path = globaldata->argv; - path = path.substr(0, (path.find_last_of('m'))); + path = m->argv; + string tempPath = path; + for (int i = 0; i < path.length(); i++) { tempPath[i] = tolower(path[i]); } + path = path.substr(0, (tempPath.find_last_of('m'))); - string formatdbCommand = path + "blast/bin/formatdb -p F -o T -i " + dbFileName; // format the database, -o option gives us the ability + string formatdbCommand; + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + formatdbCommand = path + "blast/bin/formatdb -p F -o T -i " + dbFileName; // format the database, -o option gives us the ability + #else + formatdbCommand = "\"" + path + "blast\\bin\\formatdb\" -p F -o T -i " + dbFileName; + #endif system(formatdbCommand.c_str()); // to get the right sequence names, i think. -p F // option tells formatdb that seqs are DNA, not prot - //mothurOut("DONE."); mothurOutEndLine(); mothurOutEndLine(); cout.flush(); + //m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine(); cout.flush(); } catch(exception& e) { - errorOut(e, "BlastDB", "generateDB"); + m->errorOut(e, "BlastDB", "generateDB"); exit(1); } } +/**************************************************************************************************/ /**************************************************************************************************/