X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=blastdb.cpp;h=979d507c23d61b4537d62a1a4736721624616fb2;hb=74d3aaf0f85080ea3cfb1f1871e706bab9f48392;hp=b139f3eb7a47bcc86f0920bdc0fc22bad368816a;hpb=9ada98592a54c82d08f3d46c9b1d8c3e472a922d;p=mothur.git diff --git a/blastdb.cpp b/blastdb.cpp index b139f3e..979d507 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -26,13 +26,41 @@ gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) { blastFileName = toString(randNumber) + ".blast"; } +/**************************************************************************************************/ + +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. @@ -41,25 +69,26 @@ 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 + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);; - blastCommand += (" -i " + queryFileName + " -o " + blastFileName); + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); 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; @@ -67,15 +96,17 @@ 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); } @@ -87,7 +118,7 @@ vector BlastDB::findClosestMegaBlast(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(); @@ -97,15 +128,15 @@ vector BlastDB::findClosestMegaBlast(Sequence* seq, int n) { // 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 + " -o " + blastFileName); + blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName()); 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; @@ -113,15 +144,18 @@ vector BlastDB::findClosestMegaBlast(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); +//cout << templateAccession << endl; } m8FileHandle.close(); - + remove((queryFileName+seq->getName()).c_str()); + remove((blastFileName+seq->getName()).c_str()); +//cout << "\n\n" ; return topMatches; } catch(exception& e) { - errorOut(e, "BlastDB", "findClosest"); + m->errorOut(e, "BlastDB", "findClosest"); exit(1); } } @@ -130,17 +164,17 @@ 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); } } @@ -148,7 +182,7 @@ 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'))); @@ -156,13 +190,14 @@ void BlastDB::generateDB() { 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 - //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); } } +/**************************************************************************************************/ /**************************************************************************************************/