X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=blastdb.cpp;h=b1a7b48d1108d747c1423eb0f8bad19e3ceede42;hb=1e8d08e96f4fe99604a6b3502568de464bf60891;hp=979d507c23d61b4537d62a1a4736721624616fb2;hpb=260ae19c36cb11a53ddc5a75b5e507f8dd8b31d6;p=mothur.git diff --git a/blastdb.cpp b/blastdb.cpp index 979d507..b1a7b48 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -14,23 +14,21 @@ /**************************************************************************************************/ -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 { - globaldata = GlobalData::getInstance(); count = 0; int randNumber = rand(); @@ -79,8 +77,17 @@ vector BlastDB::findClosestSequences(Sequence* seq, int n) { // 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()); + 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()); + //wrap entire string in "" + blastCommand = "\"" + blastCommand + "\""; + #endif system(blastCommand.c_str()); ifstream m8FileHandle; @@ -113,49 +120,66 @@ vector BlastDB::findClosestSequences(Sequence* seq, int n) { } /**************************************************************************************************/ //assumes you have added all the template sequences using the addSequence function and run generateDB. -vector BlastDB::findClosestMegaBlast(Sequence* seq, int n) { +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; + 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/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()); +//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()); + //wrap entire string in "" + blastCommand = "\"" + blastCommand + "\""; + + #endif + system(blastCommand.c_str()); ifstream m8FileHandle; m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error"); - string dummy; + string dummy, eScore; int templateAccession; m->gobble(m8FileHandle); while(!m8FileHandle.eof()){ - m8FileHandle >> dummy >> templateAccession >> searchScore; + 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; } } - + //while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; }else{ cout << c; } } // + //cout << endl; m->gobble(m8FileHandle); - topMatches.push_back(templateAccession); + if (searchScore >= minPerID) { + topMatches.push_back(templateAccession); + Scores.push_back(searchScore); + } //cout << templateAccession << endl; } m8FileHandle.close(); remove((queryFileName+seq->getName()).c_str()); remove((blastFileName+seq->getName()).c_str()); -//cout << "\n\n" ; +//cout << "\n" ; return topMatches; } catch(exception& e) { - m->errorOut(e, "BlastDB", "findClosest"); + m->errorOut(e, "BlastDB", "findClosestMegaBlast"); exit(1); } } @@ -184,10 +208,20 @@ void BlastDB::generateDB() { //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 + "\""; + //wrap entire string in "" + formatdbCommand = "\"" + formatdbCommand + "\""; + #endif 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();