/**************************************************************************************************/
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.
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;
//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();
-
- string root = dbFileName;
- string temp = dbFileName + ".nsq";
- remove(temp.c_str());
- temp = dbFileName + ".nsi";
- remove(temp.c_str());
-
- temp = dbFileName + ".nsd";
- remove(temp.c_str());
-
- temp = dbFileName + ".nin";
- remove(temp.c_str());
-
- temp = dbFileName + ".nhr";
- remove(temp.c_str());
-
+ remove((queryFileName+seq->getName()).c_str());
+ remove((blastFileName+seq->getName()).c_str());
-
return topMatches;
}
catch(exception& e) {
vector<int> topMatches;
ofstream queryFile;
- openOutputFile(queryFileName, queryFile);
+ m->openOutputFile((queryFileName+seq->getName()), queryFile);
queryFile << '>' << seq->getName() << endl;
queryFile << seq->getUnaligned() << endl;
queryFile.close();
// 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;
//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;
}
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
unalignedFastaFile << seq.getUnaligned() << endl;
unalignedFastaFile.close();
-
+
count++;
}
catch(exception& e) {
exit(1);
}
}
-#ifdef USE_MPI
-/**************************************************************************************************/
-int BlastDB::MPISend(int receiver) {
- try {
-
- //send gapOpen - float
- MPI_Send(&gapOpen, 1, MPI_FLOAT, receiver, 2001, MPI_COMM_WORLD);
-
- //send gapExtend - float
- MPI_Send(&gapExtend, 1, MPI_FLOAT, receiver, 2001, MPI_COMM_WORLD);
-
- //send match - float
- MPI_Send(&match, 1, MPI_FLOAT, receiver, 2001, MPI_COMM_WORLD);
-
- //send mismatch - float
- MPI_Send(&misMatch, 1, MPI_FLOAT, receiver, 2001, MPI_COMM_WORLD);
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "BlastDB", "MPISend");
- exit(1);
- }
-}
-/**************************************************************************************************/
-int BlastDB::MPIRecv(int sender) {
- try {
- MPI_Status status;
-
- //receive gapOpen - float
- MPI_Recv(&gapOpen, 1, MPI_FLOAT, sender, 2001, MPI_COMM_WORLD, &status);
-
- //receive gapExtend - float
- MPI_Recv(&gapExtend, 1, MPI_FLOAT, sender, 2001, MPI_COMM_WORLD, &status);
-
- //receive match - float
- MPI_Recv(&match, 1, MPI_FLOAT, sender, 2001, MPI_COMM_WORLD, &status);
-
- //receive mismatch - float
- MPI_Recv(&misMatch, 1, MPI_FLOAT, sender, 2001, MPI_COMM_WORLD, &status);
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "BlastDB", "MPIRecv");
- exit(1);
- }
-}
-#endif
/**************************************************************************************************/
/**************************************************************************************************/