X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=blastdb.cpp;h=b1a7b48d1108d747c1423eb0f8bad19e3ceede42;hb=1e8d08e96f4fe99604a6b3502568de464bf60891;hp=2d171391ca24ff1bfa43146de6ac5385de6ed46f;hpb=257f86c2d7b91c2530b7fc98c781d97f250a2a97;p=mothur.git diff --git a/blastdb.cpp b/blastdb.cpp index 2d17139..b1a7b48 100644 --- a/blastdb.cpp +++ b/blastdb.cpp @@ -77,8 +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; @@ -123,14 +132,23 @@ vector BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) { queryFile << '>' << seq->getName() << endl; queryFile << seq->getUnaligned() << endl; queryFile.close(); - cout << seq->getUnaligned() << endl; +// 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) + 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 - 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()); ifstream m8FileHandle; @@ -142,7 +160,7 @@ vector BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) { 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; +// 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; } } // @@ -157,7 +175,7 @@ vector BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) { m8FileHandle.close(); remove((queryFileName+seq->getName()).c_str()); remove((blastFileName+seq->getName()).c_str()); -cout << "\n" ; +//cout << "\n" ; return topMatches; } catch(exception& e) { @@ -195,7 +213,15 @@ void BlastDB::generateDB() { 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();