X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=blastdb.cpp;h=979d507c23d61b4537d62a1a4736721624616fb2;hb=d051154d778b77ba36dc603bbdb3c148a62a8e33;hp=ef8f737ee8bc5507b5c7a24dd71c3a1bd27d80ce;hpb=526a868606faa50caf86e7399f7554c0335b39e5;p=mothur.git diff --git a/blastdb.cpp b/blastdb.cpp index ef8f737..979d507 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -7,8 +7,6 @@ * */ -using namespace std; - #include "database.hpp" #include "sequence.hpp" @@ -16,80 +14,190 @@ using namespace std; /**************************************************************************************************/ -BlastDB::BlastDB(string fastaFileName, float gO, float gE, float m, float mM) : Database(fastaFileName), +BlastDB::BlastDB(float gO, float gE, float m, float mM) : Database(), gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) { - - cout << "Generating the temporary BLAST database...\t"; cout.flush(); + + globaldata = GlobalData::getInstance(); + count = 0; int randNumber = rand(); dbFileName = toString(randNumber) + ".template.unaligned.fasta"; queryFileName = toString(randNumber) + ".candidate.unaligned.fasta"; blastFileName = toString(randNumber) + ".blast"; +} +/**************************************************************************************************/ - ofstream unalignedFastaFile; - openOutputFile(dbFileName, unalignedFastaFile); - - for(int i=0;i' << i << endl; // sequences, which will be input to formatdb - unalignedFastaFile << templateSequences[i]->getUnaligned() << endl; - } - unalignedFastaFile.close(); - - string formatdbCommand = "~/Pipeline/src/cpp/production/blast/bin/formatdb -p F -o T -i " + dbFileName; // format the database, -o option gives us the ability - system(formatdbCommand.c_str()); // to get the right sequence names, i think. -p F - // option tells formatdb that seqs are DNA, not prot - cout << "DONE." << endl << endl; cout.flush(); - emptySequence = new Sequence(); - emptySequence->setName("no_match"); - emptySequence->setUnaligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX"); - emptySequence->setAligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX"); - +BlastDB::BlastDB() : Database() { + try { + globaldata = GlobalData::getInstance(); + 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. +vector BlastDB::findClosestSequences(Sequence* seq, int n) { + try{ + vector topMatches; + + ofstream 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 + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);; + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); + system(blastCommand.c_str()); + + ifstream m8FileHandle; + m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error"); + + string dummy; + int templateAccession; + m->gobble(m8FileHandle); + + while(!m8FileHandle.eof()){ + m8FileHandle >> dummy >> templateAccession >> searchScore; + + //get rest of junk in line + while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; } } + + m->gobble(m8FileHandle); + topMatches.push_back(templateAccession); + } + m8FileHandle.close(); + remove((queryFileName+seq->getName()).c_str()); + remove((blastFileName+seq->getName()).c_str()); -Sequence* BlastDB::findClosestSequence(Sequence* candidate){ + return topMatches; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "findClosestSequences"); + exit(1); + } - ofstream queryFile; - openOutputFile(queryFileName, queryFile); - queryFile << '>' << candidate->getName() << endl; - queryFile << candidate->getUnaligned() << endl; - queryFile.close(); - +} +/**************************************************************************************************/ +//assumes you have added all the template sequences using the addSequence function and run generateDB. +vector BlastDB::findClosestMegaBlast(Sequence* seq, int n) { + try{ + vector topMatches; + + ofstream 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. -// 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/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()); + system(blastCommand.c_str()); - string blastCommand = "~/Pipeline/src/cpp/production/blast/bin/blastall -p blastn -d " + dbFileName + " -b 1 -m 8 -W 28"; - blastCommand += (" -i " + queryFileName + " -o " + blastFileName); - system(blastCommand.c_str()); - - ifstream m8FileHandle; - openInputFile(blastFileName, m8FileHandle); + ifstream m8FileHandle; + m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error"); - string dummy; - int templateAccession; - gobble(m8FileHandle); - if(!m8FileHandle.eof()){ - m8FileHandle >> dummy >> templateAccession >> searchScore; + string dummy; + int templateAccession; + m->gobble(m8FileHandle); + + while(!m8FileHandle.eof()){ + m8FileHandle >> dummy >> templateAccession >> searchScore; + + //get rest of junk in line + while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; } } + + m->gobble(m8FileHandle); + topMatches.push_back(templateAccession); +//cout << templateAccession << endl; + } m8FileHandle.close(); - return templateSequences[templateAccession]; + remove((queryFileName+seq->getName()).c_str()); + remove((blastFileName+seq->getName()).c_str()); +//cout << "\n\n" ; + return topMatches; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "findClosest"); + exit(1); + } +} +/**************************************************************************************************/ +void BlastDB::addSequence(Sequence seq) { + try { + + ofstream 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++; } - else{ - searchScore = 0.00; - return emptySequence; + catch(exception& e) { + m->errorOut(e, "BlastDB", "addSequence"); + exit(1); } } +/**************************************************************************************************/ +void BlastDB::generateDB() { + try { + + //m->mothurOut("Generating the temporary BLAST database...\t"); cout.flush(); + + path = globaldata->argv; + path = path.substr(0, (path.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 + system(formatdbCommand.c_str()); // to get the right sequence names, i think. -p F + // option tells formatdb that seqs are DNA, not prot + //m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine(); cout.flush(); + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "generateDB"); + exit(1); + } +} +/**************************************************************************************************/ /**************************************************************************************************/ +