changed how we count sequences in a fastafile to allow for '>' in sequence names

[mothur.git] / blastdb.cpp

diff --git a/blastdb.cpp b/blastdb.cpp
index b139f3eb7a47bcc86f0920bdc0fc22bad368816a..760824dbb23f9b2a9d680ace769ccc1c6ffab96c 100644 (file)
--- a/blastdb.cpp
+++ b/blastdb.cpp
@@ -25,6 +25,19 @@ gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) {
        queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
        blastFileName = toString(randNumber) + ".blast";
 
+}
+/**************************************************************************************************/
+
+BlastDB::BlastDB() : Database() {
+       
+       globaldata = GlobalData::getInstance();
+       count = 0;
+
+       int randNumber = rand();
+       dbFileName = toString(randNumber) + ".template.unaligned.fasta";
+       queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
+       blastFileName = toString(randNumber) + ".blast";
+
 }
 
 /**************************************************************************************************/
@@ -41,21 +54,22 @@ vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
                vector<int> topMatches;
                
                ofstream queryFile;
-               openOutputFile(queryFileName, queryFile);
+               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);
+               openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
                
                string dummy;
                int templateAccession;
@@ -71,11 +85,13 @@ vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
                        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 +103,7 @@ vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
                vector<int> topMatches;
                
                ofstream queryFile;
-               openOutputFile(queryFileName, queryFile);
+               openOutputFile((queryFileName+seq->getName()), queryFile);
                queryFile << '>' << seq->getName() << endl;
                queryFile << seq->getUnaligned() << endl;
                queryFile.close();
@@ -97,11 +113,11 @@ vector<int> 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);
+               openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
        
                string dummy;
                int templateAccession;
@@ -115,13 +131,16 @@ vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
                        
                        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);
        }
 }
@@ -136,11 +155,11 @@ void BlastDB::addSequence(Sequence seq) {
                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 +167,7 @@ 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 +175,14 @@ 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);
        }
 }
+/**************************************************************************************************/
 
 /**************************************************************************************************/