X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=blastdb.cpp;h=2eced7e14dc6fa42e0147747fc622bfa088250f7;hp=e442fa229b49e08d5c7b80495d3c8a10d75d9430;hb=050a3ff02473a3d4c0980964e1a9ebe52e55d6b8;hpb=02909d6cae9963ba00dc746969a370fa8ca934fc diff --git a/blastdb.cpp b/blastdb.cpp index e442fa2..2eced7e 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -14,90 +14,393 @@ /**************************************************************************************************/ -BlastDB::BlastDB(string fastaFileName, float gO, float gE, float m, float mM) : Database(fastaFileName), -gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) { - - globaldata = GlobalData::getInstance(); - - cout << "Generating the temporary BLAST database...\t"; cout.flush(); - - int randNumber = rand(); - dbFileName = toString(randNumber) + ".template.unaligned.fasta"; - queryFileName = toString(randNumber) + ".candidate.unaligned.fasta"; - blastFileName = toString(randNumber) + ".blast"; +BlastDB::BlastDB(string tag, float gO, float gE, float mm, float mM, string b, int tid) : Database(), +gapOpen(gO), gapExtend(gE), match(mm), misMatch(mM) { + try { + count = 0; + path = b; + threadID = tid; - - ofstream unalignedFastaFile; - openOutputFile(dbFileName, unalignedFastaFile); - - for(int i=0;i' << i << endl; // sequences, which will be input to formatdb - unalignedFastaFile << templateSequences[i].getUnaligned() << endl; + int randNumber = rand(); + //int randNumber = 12345; + string pid = ""; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + pid += getpid(); +#else + pid += toString(threadID); +#endif + + + dbFileName = tag + pid + toString(randNumber) + ".template.unaligned.fasta"; + queryFileName = tag + pid + toString(randNumber) + ".candidate.unaligned.fasta"; + blastFileName = tag + pid + toString(randNumber) + ".blast"; + + //make sure blast exists in the write place + if (path == "") { + 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'))); + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + path += "blast/bin/"; +#else + path += "blast\\bin\\"; +#endif + } + + + string formatdbCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + formatdbCommand = path + "formatdb"; // format the database, -o option gives us the ability +#else + formatdbCommand = path + "formatdb.exe"; +#endif + + //test to make sure formatdb exists + ifstream in; + formatdbCommand = m->getFullPathName(formatdbCommand); + int ableToOpen = m->openInputFile(formatdbCommand, in, "no error"); in.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + formatdbCommand + " file does not exist. mothur requires formatdb.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + + string blastCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + blastCommand = path + "blastall"; // format the database, -o option gives us the ability +#else + blastCommand = path + "blastall.exe"; + //wrap entire string in "" + //blastCommand = "\"" + blastCommand + "\""; +#endif + + //test to make sure formatdb exists + ifstream in2; + blastCommand = m->getFullPathName(blastCommand); + ableToOpen = m->openInputFile(blastCommand, in2, "no error"); in2.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + blastCommand + " file does not exist. mothur requires blastall.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + + + string megablastCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + megablastCommand = path + "megablast"; // format the database, -o option gives us the ability +#else + megablastCommand = path + "megablast.exe"; +#endif + + //test to make sure formatdb exists + ifstream in3; + megablastCommand = m->getFullPathName(megablastCommand); + ableToOpen = m->openInputFile(megablastCommand, in3, "no error"); in3.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + megablastCommand + " file does not exist. mothur requires megablast.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + } - unalignedFastaFile.close(); - - 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 - cout << "DONE." << endl << endl; cout.flush(); - emptySequence = Sequence(); - emptySequence.setName("no_match"); - emptySequence.setUnaligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX"); - emptySequence.setAligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX"); - + catch(exception& e) { + m->errorOut(e, "BlastDB", "BlastDB"); + exit(1); + } +} +/**************************************************************************************************/ +BlastDB::BlastDB(string b, int tid) : Database() { + try { + count = 0; + + path = b; + threadID = tid; + + //make sure blast exists in the write place + if (path == "") { + 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'))); + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + path += "blast/bin/"; +#else + path += "blast\\bin\\"; +#endif + } + + int randNumber = rand(); + string pid = ""; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + pid += getpid(); +#else + pid += toString(threadID); +#endif + + dbFileName = pid + toString(randNumber) + ".template.unaligned.fasta"; + queryFileName = pid + toString(randNumber) + ".candidate.unaligned.fasta"; + blastFileName = pid + toString(randNumber) + ".blast"; + + string formatdbCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + formatdbCommand = path + "formatdb"; // format the database, -o option gives us the ability +#else + formatdbCommand = path + "formatdb.exe"; + //wrap entire string in "" + //formatdbCommand = "\"" + formatdbCommand + "\""; +#endif + + //test to make sure formatdb exists + ifstream in; + formatdbCommand = m->getFullPathName(formatdbCommand); + int ableToOpen = m->openInputFile(formatdbCommand, in, "no error"); in.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + formatdbCommand + " file does not exist. mothur requires formatdb.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + + string blastCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + blastCommand = path + "blastall"; // format the database, -o option gives us the ability +#else + blastCommand = path + "blastall.exe"; + //wrap entire string in "" + //blastCommand = "\"" + blastCommand + "\""; +#endif + + //test to make sure formatdb exists + ifstream in2; + blastCommand = m->getFullPathName(blastCommand); + ableToOpen = m->openInputFile(blastCommand, in2, "no error"); in2.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + blastCommand + " file does not exist. mothur requires blastall.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + + + string megablastCommand; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + megablastCommand = path + "megablast"; // format the database, -o option gives us the ability +#else + megablastCommand = path + "megablast.exe"; + //wrap entire string in "" + //megablastCommand = "\"" + megablastCommand + "\""; +#endif + + //test to make sure formatdb exists + ifstream in3; + megablastCommand = m->getFullPathName(megablastCommand); + ableToOpen = m->openInputFile(megablastCommand, in3, "no error"); in3.close(); + if(ableToOpen == 1) { m->mothurOut("[ERROR]: " + megablastCommand + " file does not exist. mothur requires megablast.exe."); m->mothurOutEndLine(); m->control_pressed = true; } + + + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "BlastDB"); + exit(1); + } } /**************************************************************************************************/ BlastDB::~BlastDB(){ - //for (int i = 0; i < templateSequences.size(); i++) { delete templateSequences[i]; } - //templateSequences.clear(); - //delete emptySequence; - - 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{ + m->mothurRemove(queryFileName); // let's clean stuff up and remove the temp files + m->mothurRemove(dbFileName); // let's clean stuff up and remove the temp files + m->mothurRemove((dbFileName+".nsq")); // let's clean stuff up and remove the temp files + m->mothurRemove((dbFileName+".nsi")); // let's clean stuff up and remove the temp files + m->mothurRemove((dbFileName+".nsd")); // let's clean stuff up and remove the temp files + m->mothurRemove((dbFileName+".nin")); // let's clean stuff up and remove the temp files + m->mothurRemove((dbFileName+".nhr")); // let's clean stuff up and remove the temp files + m->mothurRemove(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; + int randNumber = rand(); + string pid = scrubName(seq->getName()); + + m->openOutputFile((queryFileName+pid+toString(randNumber)), 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; + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + + blastCommand = path + "blastall -p blastn -d " + dbFileName + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n); + blastCommand += (" -i " + (queryFileName+pid+toString(randNumber)) + " -o " + blastFileName+pid+toString(randNumber)); + #else + blastCommand = "\"" + path + "blastall\" -p blastn -d " + "\"" + dbFileName + "\"" + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n); + blastCommand += (" -i " + (queryFileName+pid+toString(randNumber)) + " -o " + blastFileName+pid+toString(randNumber)); + //wrap entire string in "" + blastCommand = "\"" + blastCommand + "\""; + #endif + + system(blastCommand.c_str()); + + ifstream m8FileHandle; + m->openInputFile(blastFileName+pid+toString(randNumber), 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(); + m->mothurRemove((queryFileName+pid+toString(randNumber))); + m->mothurRemove((blastFileName+pid+toString(randNumber))); + + return topMatches; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "findClosestSequences"); + exit(1); + } +} /**************************************************************************************************/ +//assumes you have added all the template sequences using the addSequence function and run generateDB. +vector BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) { + try{ + vector topMatches; + float numBases, mismatch, gap, startQuery, endQuery, startRef, endRef, score; + Scores.clear(); + + ofstream queryFile; + int randNumber = rand(); + string pid = scrubName(seq->getName()); + + m->openOutputFile((queryFileName+pid+toString(randNumber)), 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. +//7000004128189528left 0 100 66 0 0 1 66 61 126 1e-31 131 + string blastCommand; + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + blastCommand = path + "megablast -e 1e-10 -d " + dbFileName + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn + blastCommand += (" -i " + (queryFileName+pid+toString(randNumber)) + " -o " + blastFileName+pid+toString(randNumber)); + #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+toString(randNumber)) + " -o " + blastFileName+toString(randNumber)); + + blastCommand = "\"" + path + "megablast\" -e 1e-10 -d " + "\"" + dbFileName + "\"" + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn + blastCommand += (" -i " + (queryFileName+pid+toString(randNumber)) + " -o " + blastFileName+pid+toString(randNumber)); + //wrap entire string in "" + blastCommand = "\"" + blastCommand + "\""; -Sequence BlastDB::findClosestSequence(Sequence* candidate){ + #endif + system(blastCommand.c_str()); - ofstream queryFile; - openOutputFile(queryFileName, queryFile); - queryFile << '>' << candidate->getName() << endl; - queryFile << candidate->getUnaligned() << endl; - queryFile.close(); + ifstream m8FileHandle; + m->openInputFile(blastFileName+pid+toString(randNumber), m8FileHandle, "no error"); + string dummy, eScore; + int templateAccession; + m->gobble(m8FileHandle); + + while(!m8FileHandle.eof()){ + 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; + + //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(); + m->mothurRemove((queryFileName+pid+toString(randNumber))); + m->mothurRemove((blastFileName+pid+toString(randNumber))); +//cout << "\n" ; + return topMatches; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "findClosestMegaBlast"); + exit(1); + } +} +/**************************************************************************************************/ +void BlastDB::addSequence(Sequence seq) { + try { -// 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"; - blastCommand += (" -i " + queryFileName + " -o " + blastFileName); - system(blastCommand.c_str()); + ofstream unalignedFastaFile; + m->openOutputFileAppend(dbFileName, unalignedFastaFile); - ifstream m8FileHandle; - openInputFile(blastFileName, m8FileHandle); + // generating a fasta file with unaligned template + unalignedFastaFile << '>' << count << endl; // sequences, which will be input to formatdb + unalignedFastaFile << seq.getUnaligned() << endl; + unalignedFastaFile.close(); - string dummy; - int templateAccession; - gobble(m8FileHandle); - if(!m8FileHandle.eof()){ - m8FileHandle >> dummy >> templateAccession >> searchScore; - m8FileHandle.close(); - return templateSequences[templateAccession]; + count++; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "addSequence"); + exit(1); + } +} +/**************************************************************************************************/ +void BlastDB::generateDB() { + try { + + //m->mothurOut("Generating the temporary BLAST database...\t"); cout.flush(); + + string formatdbCommand; + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + formatdbCommand = path + "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; // format the database, -o option gives us the ability + + formatdbCommand = "\"" + path + "formatdb\" -p F -o T -i " + "\"" + dbFileName + "\""; + //wrap entire string in "" + formatdbCommand = "\"" + formatdbCommand + "\""; + #endif + //cout << formatdbCommand << endl; + 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(); } - else{ - searchScore = 0.00; - return emptySequence; + catch(exception& e) { + m->errorOut(e, "BlastDB", "generateDB"); + exit(1); } - m8FileHandle.close(); } +/**************************************************************************************************/ +string BlastDB::scrubName(string seqName) { + try { + + string cleanName = ""; + + for (int i = 0; i < seqName.length(); i++) { + if (isalnum(seqName[i])) { cleanName += seqName[i]; } + else { cleanName += "_"; } + } + + return cleanName; + } + catch(exception& e) { + m->errorOut(e, "BlastDB", "scrubName"); + exit(1); + } +} +/**************************************************************************************************/ /**************************************************************************************************/ +