adding current file class

[mothur.git] / blastdb.cpp

diff --git a/blastdb.cpp b/blastdb.cpp
index 35a19f485816c87db85b71c74ca138f00595b180..bc92e1c5f083831a88ec2d10e7a0fe94c68bb6b7 100644 (file)
--- a/blastdb.cpp
+++ b/blastdb.cpp
@@ -29,23 +29,38 @@ gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) {
 /**************************************************************************************************/
 
 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";
+       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.
@@ -54,25 +69,26 @@ vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
                vector<int> 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;
@@ -80,11 +96,13 @@ vector<int> 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) {
@@ -100,7 +118,7 @@ vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
                vector<int> topMatches;
                
                ofstream queryFile;
-               openOutputFile(queryFileName, queryFile);
+               m->openOutputFile((queryFileName+seq->getName()), queryFile);
                queryFile << '>' << seq->getName() << endl;
                queryFile << seq->getUnaligned() << endl;
                queryFile.close();
@@ -110,27 +128,30 @@ 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, "no error");
+               m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
        
                string dummy;
                int templateAccession;
-               gobble(m8FileHandle);
+               m->gobble(m8FileHandle);
                
                while(!m8FileHandle.eof()){
                        m8FileHandle >> dummy >> templateAccession >> searchScore;
+                       //cout << templateAccession << '\t' << searchScore << endl;
                        
                        //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;
        }
@@ -144,7 +165,7 @@ 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