+int FilterSeqsCommand::filterSequences() {
+ try {
+
+ for (int s = 0; s < fastafileNames.size(); s++) {
+
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+
+ string filteredFasta = outputDir + getRootName(getSimpleName(fastafileNames[s])) + "filter.fasta";
+#ifdef USE_MPI
+ int pid, start, end;
+ int tag = 2001;
+
+ 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[filteredFasta.length()];
+ strcpy(outFilename, filteredFasta.c_str());
+
+ char inFileName[fastafileNames[s].length()];
+ 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 (pid == 0) { //you are the root process
+
+ setLines(fastafileNames[s]);
+
+ for (int j = 0; j < lines.size(); j++) { //each process
+ if (j != 0) { //don't send to yourself
+ MPI_Send(&lines[j]->start, 1, MPI_INT, j, tag, MPI_COMM_WORLD); //start position in file
+ MPI_Send(&bufferSizes[j], 1, MPI_INT, j, tag, MPI_COMM_WORLD); //how bytes for the read
+ }
+ }
+
+ //read your peice of file
+ char buf[bufferSizes[0]];
+ MPI_File_read_at(inMPI, lines[0]->start, buf, bufferSizes[0], MPI_CHAR, &status);
+ istringstream iss (buf,istringstream::in);
+
+ //do your part
+ driverMPIRun(iss, outMPI);
+
+ //wait on chidren
+ for(int i = 1; i < processors; i++) {
+ char buf[4];
+ MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
+ }
+
+ }else { //you are a child process
+ //receive your section of file
+ int startPos, numLines, bufferSize;
+ MPI_Recv(&startPos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPI_Recv(&bufferSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+
+ //read your peice of file
+ char buf2[bufferSize];
+ MPI_File_read_at(inMPI, startPos, buf2, bufferSize, MPI_CHAR, &status);
+ istringstream iss (buf2,istringstream::in);
+
+ //do your part
+ driverMPIRun(iss, outMPI);
+
+ char buf[4];
+ strcpy(buf, "done");
+
+ //tell parent you are done.
+ MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
+ }
+
+ MPI_File_close(&outMPI);
+ MPI_File_close(&inMPI);
+
+#else
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ if(processors == 1){
+ ifstream inFASTA;
+ openInputFile(fastafileNames[s], inFASTA);
+ int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ inFASTA.close();
+
+ lines.push_back(new linePair(0, numFastaSeqs));
+
+ driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
+ }else{
+ setLines(fastafileNames[s]);
+ createProcessesRunFilter(filter, fastafileNames[s]);
+
+ rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
+
+ //append fasta files
+ for(int i=1;i<processors;i++){
+ appendAlignFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
+ remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
+ }
+ }
+
+ if (m->control_pressed) { return 1; }
+ #else
+ ifstream inFASTA;
+ openInputFile(fastafileNames[s], inFASTA);
+ int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ inFASTA.close();
+
+ lines.push_back(new linePair(0, numFastaSeqs));
+
+ driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
+
+ if (m->control_pressed) { return 1; }
+ #endif
+#endif
+ outputNames.push_back(filteredFasta);
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "FilterSeqsCommand", "filterSequences");
+ exit(1);
+ }
+}
+/**************************************************************************************/
+int FilterSeqsCommand::driverMPIRun(istringstream& in, MPI_File& outMPI) {
+ try {
+ string outputString = "";
+ int count = 0;
+ MPI_Status status;
+
+ while (!in.eof()) {
+
+ Sequence seq(in); gobble(in);
+
+ if (seq.getName() != "") {
+ string align = seq.getAligned();
+ string filterSeq = "";
+
+ for(int j=0;j<alignmentLength;j++){
+ if(filter[j] == '1'){
+ filterSeq += align[j];
+ }
+ }
+
+ count++;
+ outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
+
+ if(count % 10 == 0){ //output to file
+ //send results to parent
+ int length = outputString.length();
+ char buf[length];
+ strcpy(buf, outputString.c_str());
+
+ MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
+ outputString = "";
+ }
+
+ }
+ }
+
+ if(outputString != ""){ //output to file
+ //send results to parent
+ int length = outputString.length();
+ char buf[length];
+ strcpy(buf, outputString.c_str());
+
+ MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
+ outputString = "";
+ }
+
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
+ exit(1);
+ }
+}
+/**************************************************************************************/
+int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
+ try {
+ ofstream out;
+ openOutputFile(outputFilename, out);
+
+ ifstream in;
+ openInputFile(inputFilename, in);
+
+ in.seekg(line->start);
+
+ for(int i=0;i<line->numSeqs;i++){
+
+ if (m->control_pressed) { in.close(); out.close(); return 0; }
+
+ Sequence seq(in);
+ if (seq.getName() != "") {
+ string align = seq.getAligned();
+ string filterSeq = "";
+
+ for(int j=0;j<alignmentLength;j++){
+ if(filter[j] == '1'){
+ filterSeq += align[j];
+ }
+ }
+
+ out << '>' << seq.getName() << endl << filterSeq << endl;
+ }
+ gobble(in);
+ }
+ out.close();
+ in.close();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+
+int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
+ try {
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ int process = 0;
+ int exitCommand = 1;
+ processIDS.clear();
+
+ //loop through and create all the processes you want
+ while (process != processors) {
+ int pid = fork();
+
+ if (pid > 0) {
+ processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
+ process++;
+ }else if (pid == 0){
+ string filteredFasta = filename + toString(getpid()) + ".temp";
+ driverRunFilter(F, filteredFasta, filename, lines[process]);
+ exit(0);
+ }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
+ }
+
+ //force parent to wait until all the processes are done
+ for (int i=0;i<processors;i++) {
+ int temp = processIDS[i];
+ wait(&temp);
+ }
+
+ return exitCommand;
+#endif
+ }
+ catch(exception& e) {
+ m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
+ exit(1);
+ }
+}
+/**************************************************************************************/