int FilterSeqsCommand::filterSequences() {
try {
numSeqs = 0;
for (int s = 0; s < fastafileNames.size(); s++) {
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafileNames[s]));
string filteredFasta = getOutputFileName("fasta", variables);
vector<unsigned long long> positions;
if (savedPositions.size() != 0) { positions = savedPositions[s]; }
else {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
positions = m->divideFile(fastafileNames[s], processors);
#else
if(processors != 1){
int numFastaSeqs = 0;
positions = m->setFilePosFasta(fastafileNames[s], numFastaSeqs);
if (numFastaSeqs < processors) { processors = numFastaSeqs; }
}
#endif
}
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//vector<unsigned long long> positions = m->divideFile(fastafileNames[s], processors);
for (int i = 0; i < (positions.size()-1); i++) {
lines.push_back(new linePair(positions[i], positions[(i+1)]));
}
if(processors == 1){
int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
}else{
int numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s], filteredFasta);
numSeqs += numFastaSeqs;
}
if (m->control_pressed) { return 1; }
#else
if(processors == 1){
lines.push_back(new linePair(0, 1000));
int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
}else {
int numFastaSeqs = positions.size()-1;
//positions = m->setFilePosFasta(fastafileNames[s], numFastaSeqs);
//figure out how many sequences you have to process
int numSeqsPerProcessor = numFastaSeqs / processors;
for (int i = 0; i < processors; i++) {
int startIndex = i * numSeqsPerProcessor;
if(i == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; }
lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
}
numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s], filteredFasta);
numSeqs += numFastaSeqs;
}
if (m->control_pressed) { return 1; }
#endif
outputNames.push_back(filteredFasta); outputTypes["fasta"].push_back(filteredFasta);
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "FilterSeqsCommand", "filterSequences");
exit(1);
}
}
int FilterSeqsCommand::filterSequences() {
try {
numSeqs = 0;
for (int s = 0; s < fastafileNames.size(); s++) {
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafileNames[s]));
string filteredFasta = getOutputFileName("fasta", variables);
#ifdef USE_MPI
int pid, numSeqsPerProcessor, num;
int tag = 2001;
vector<unsigned long long>MPIPos;
MPI_Status status;
MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
MPI_File outMPI;
MPI_File inMPI;
int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
int inMode=MPI_MODE_RDONLY;
char outFilename[1024];
strcpy(outFilename, filteredFasta.c_str());
char inFileName[1024];
strcpy(inFileName, fastafileNames[s].c_str());
MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
if (pid == 0) { //you are the root process
MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
numSeqs += num;
//send file positions to all processes
for(int i = 1; i < processors; i++) {
MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
}
//figure out how many sequences you have to do
numSeqsPerProcessor = num / processors;
int startIndex = pid * numSeqsPerProcessor;
if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
//do your part
driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
//wait on chidren
for(int i = 1; i < processors; i++) {
char buf[5];
MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
}
}else { //you are a child process
MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
MPIPos.resize(num+1);
numSeqs += num;
MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
//figure out how many sequences you have to align
numSeqsPerProcessor = num / processors;
int startIndex = pid * numSeqsPerProcessor;
if(pid == (processors - 1)){ numSeqsPerProcessor = num - pid * numSeqsPerProcessor; }
//align your part
driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
char buf[5];
strcpy(buf, "done");
//tell parent you are done.
MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
}
MPI_File_close(&outMPI);
MPI_File_close(&inMPI);
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
vector<unsigned long long> positions;
if (savedPositions.size() != 0) { positions = savedPositions[s]; }
else {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
positions = m->divideFile(fastafileNames[s], processors);
#else
if(processors != 1){
int numFastaSeqs = 0;
positions = m->setFilePosFasta(fastafileNames[s], numFastaSeqs);
if (positions.size() < processors) { processors = positions.size(); }
}
#endif
}
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//vector<unsigned long long> positions = m->divideFile(fastafileNames[s], processors);
for (int i = 0; i < (positions.size()-1); i++) {
lines.push_back(new linePair(positions[i], positions[(i+1)]));
}
if(processors == 1){
int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
}else{
int numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s], filteredFasta);
numSeqs += numFastaSeqs;
}
if (m->control_pressed) { return 1; }
#else
if(processors == 1){
lines.push_back(new linePair(0, 1000));
int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
}else {
int numFastaSeqs = positions.size()-1;
//positions = m->setFilePosFasta(fastafileNames[s], numFastaSeqs);
//figure out how many sequences you have to process
int numSeqsPerProcessor = numFastaSeqs / processors;
for (int i = 0; i < processors; i++) {
int startIndex = i * numSeqsPerProcessor;
if(i == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; }
lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
}
numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s], filteredFasta);
numSeqs += numFastaSeqs;
}
if (m->control_pressed) { return 1; }
#endif
#endif
outputNames.push_back(filteredFasta); outputTypes["fasta"].push_back(filteredFasta);
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "FilterSeqsCommand", "filterSequences");
exit(1);
}
}
