changes for chimera slayer

[mothur.git] / blastdb.cpp

diff --git a/blastdb.cpp b/blastdb.cpp
index 979d507c23d61b4537d62a1a4736721624616fb2..3e1210b46abaa4c01314a8702feea3eded345c11 100644 (file)
--- a/blastdb.cpp
+++ b/blastdb.cpp
@@ -17,7 +17,6 @@
 BlastDB::BlastDB(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();
@@ -30,7 +29,6 @@ gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) {
 
 BlastDB::BlastDB() : Database() {
        try {
-               globaldata = GlobalData::getInstance();
                count = 0;
 
                int randNumber = rand();
@@ -113,11 +111,14 @@ vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
 }
 /**************************************************************************************************/
 //assumes you have added all the template sequences using the addSequence function and run generateDB.
-vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
+vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n, int minPerID) {
        try{
                vector<int> 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;
@@ -126,7 +127,7 @@ vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
                //      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 = 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());
@@ -134,28 +135,32 @@ vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
                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';
                        
                        //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 (score >= 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);
        }
 }
@@ -184,8 +189,10 @@ 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
                system(formatdbCommand.c_str());                                                                //      to get the right sequence names, i think. -p F