OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
- ValidParameters validParameter;
+ ValidParameters validParameter("chimera.slayer");
map<string,string>::iterator it;
//check to make sure all parameters are valid for command
if (inputDir == "not found"){ inputDir = ""; }
else {
string path;
- it = parameters.find("fasta");
- //user has given a template file
- if(it != parameters.end()){
- path = hasPath(it->second);
- //if the user has not given a path then, add inputdir. else leave path alone.
- if (path == "") { parameters["fasta"] = inputDir + it->second; }
- }
-
it = parameters.find("template");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["template"] = inputDir + it->second; }
}
//check for required parameters
- fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true; }
+ fastafile = validParameter.validFile(parameters, "fasta", false);
+ if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true; }
+ else {
+ m->splitAtDash(fastafile, fastaFileNames);
+
+ //go through files and make sure they are good, if not, then disregard them
+ for (int i = 0; i < fastaFileNames.size(); i++) {
+ if (inputDir != "") {
+ string path = m->hasPath(fastaFileNames[i]);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { fastaFileNames[i] = inputDir + fastaFileNames[i]; }
+ }
+
+ int ableToOpen;
+ ifstream in;
+
+ ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
+
+ //if you can't open it, try default location
+ if (ableToOpen == 1) {
+ if (m->getDefaultPath() != "") { //default path is set
+ string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
+ m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
+ ableToOpen = m->openInputFile(tryPath, in, "noerror");
+ fastaFileNames[i] = tryPath;
+ }
+ }
+ in.close();
+
+ if (ableToOpen == 1) {
+ m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
+ //erase from file list
+ fastaFileNames.erase(fastaFileNames.begin()+i);
+ i--;
+ }
+ }
+
+ //make sure there is at least one valid file left
+ if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
+ }
//if the user changes the output directory command factory will send this info to us in the output parameter
- outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
- outputDir = "";
- outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
- }
+ outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
templatefile = validParameter.validFile(parameters, "template", true);
if (templatefile == "not open") { abort = true; }
temp = validParameter.validFile(parameters, "minbs", false); if (temp == "not found") { temp = "90"; }
convert(temp, minBS);
- temp = validParameter.validFile(parameters, "minsnp", false); if (temp == "not found") { temp = "10"; }
+ temp = validParameter.validFile(parameters, "minsnp", false); if (temp == "not found") { temp = "100"; }
convert(temp, minSNP);
temp = validParameter.validFile(parameters, "parents", false); if (temp == "not found") { temp = "3"; }
convert(temp, parents);
temp = validParameter.validFile(parameters, "realign", false); if (temp == "not found") { temp = "f"; }
- realign = isTrue(temp);
+ realign = m->isTrue(temp);
search = validParameter.validFile(parameters, "search", false); if (search == "not found") { search = "distance"; }
m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
m->mothurOut("The chimera.slayer command parameters are fasta, template, processors, ksize, window, match, mismatch, divergence. minsim, mincov, minbs, minsnp, parents, search, iters, increment and numwanted.\n"); //realign,
m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
+ m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n");
m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
m->mothurOut("The processors parameter allows you to specify how many processors you would like to use. The default is 1. \n");
#ifdef USE_MPI
m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
- m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 10) \n");
+ m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 100) \n");
m->mothurOut("The search parameter allows you to specify search method for finding the closest parent. Choices are distance, blast, and kmer, default distance. \n");
- //m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false. Found to make results worse. \n");
+ m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false. \n");
m->mothurOut("The chimera.slayer command should be in the following format: \n");
m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
if (abort == true) { return 0; }
- int start = time(NULL);
-
- chimera = new ChimeraSlayer(fastafile, templatefile, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
-
- string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "slayer.chimeras";
- string accnosFileName = outputDir + getRootName(getSimpleName(fastafile)) + "slayer.accnos";
- bool hasAccnos = true;
-
- if (m->control_pressed) { delete chimera; return 0; }
-
- if (chimera->getUnaligned()) {
- m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
- delete chimera;
- return 0;
- }
- templateSeqsLength = chimera->getLength();
+ for (int s = 0; s < fastaFileNames.size(); s++) {
+
+ m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
- #ifdef USE_MPI
- int pid, end, numSeqsPerProcessor;
- int tag = 2001;
- vector<long> MPIPos;
- MPIWroteAccnos = false;
+ int start = time(NULL);
- MPI_Status status;
- MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
- MPI_Comm_size(MPI_COMM_WORLD, &processors);
-
- MPI_File inMPI;
- MPI_File outMPI;
- MPI_File outMPIAccnos;
+ chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
- int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
- int inMode=MPI_MODE_RDONLY;
+ if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]); }//if user entered a file with a path then preserve it
+ string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras";
+ string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.accnos";
- //char* outFilename = new char[outputFileName.length()];
- //memcpy(outFilename, outputFileName.c_str(), outputFileName.length());
+ if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } return 0; }
- char outFilename[1024];
- strcpy(outFilename, outputFileName.c_str());
-
- //char* outAccnosFilename = new char[accnosFileName.length()];
- //memcpy(outAccnosFilename, accnosFileName.c_str(), accnosFileName.length());
-
- char outAccnosFilename[1024];
- strcpy(outAccnosFilename, accnosFileName.c_str());
-
- //char* inFileName = new char[fastafile.length()];
- //memcpy(inFileName, fastafile.c_str(), fastafile.length());
-
- char inFileName[1024];
- strcpy(inFileName, fastafile.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);
- MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
+ if (chimera->getUnaligned()) {
+ m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
+ delete chimera;
+ return 0;
+ }
+ templateSeqsLength = chimera->getLength();
- //delete inFileName;
- //delete outFilename;
- //delete outAccnosFilename;
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); delete chimera; return 0; }
-
- if (pid == 0) { //you are the root process
- m->mothurOutEndLine();
- m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
- m->mothurOutEndLine();
-
- string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
+ #ifdef USE_MPI
+ int pid, end, numSeqsPerProcessor;
+ int tag = 2001;
+ vector<unsigned long int> MPIPos;
- //print header
- int length = outTemp.length();
- char* buf2 = new char[length];
- memcpy(buf2, outTemp.c_str(), length);
-
- MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
- delete buf2;
+ MPI_Status status;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
- MPIPos = setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
+ MPI_File inMPI;
+ MPI_File outMPI;
+ MPI_File outMPIAccnos;
- //send file positions to all processes
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
+ int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+ int inMode=MPI_MODE_RDONLY;
- //figure out how many sequences you have to align
- numSeqsPerProcessor = numSeqs / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- int startIndex = pid * numSeqsPerProcessor;
-
- //align your part
- driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); delete chimera; return 0; }
-
- for (int i = 1; i < processors; i++) {
- bool tempResult;
- MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
- if (tempResult != 0) { MPIWroteAccnos = true; }
- }
- }else{ //you are a child process
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
- MPIPos.resize(numSeqs+1);
- MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
+ char outFilename[1024];
+ strcpy(outFilename, outputFileName.c_str());
- //figure out how many sequences you have to align
- numSeqsPerProcessor = numSeqs / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- int startIndex = pid * numSeqsPerProcessor;
+ char outAccnosFilename[1024];
+ strcpy(outAccnosFilename, accnosFileName.c_str());
- //align your part
- driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); delete chimera; return 0; }
+ char inFileName[1024];
+ strcpy(inFileName, fastaFileNames[s].c_str());
- MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
- }
+ 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);
+ MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } delete chimera; return 0; }
- //close files
- MPI_File_close(&inMPI);
- MPI_File_close(&outMPI);
- MPI_File_close(&outMPIAccnos);
-
- //delete accnos file if blank
- if (pid == 0) {
- if (!MPIWroteAccnos) {
- //MPI_Info info;
- //MPI_File_delete(outAccnosFilename, info);
- hasAccnos = false;
- remove(accnosFileName.c_str());
- }
- }
-
- #else
- ofstream outHeader;
- string tempHeader = outputDir + getRootName(getSimpleName(fastafile)) + "slayer.chimeras.tempHeader";
- openOutputFile(tempHeader, outHeader);
-
- chimera->printHeader(outHeader);
- outHeader.close();
+ if (pid == 0) { //you are the root process
+ m->mothurOutEndLine();
+ m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
+ m->mothurOutEndLine();
- //break up file
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors == 1){
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
-
- lines.push_back(new linePair(0, numSeqs));
-
- driver(lines[0], outputFileName, fastafile, accnosFileName);
-
- if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(tempHeader.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
- }
-
- //delete accnos file if its blank
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
-
- }else{
- vector<int> positions;
- processIDS.resize(0);
-
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
-
- string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
- if (input.length() != 0) {
- if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
+ string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
+
+ //print header
+ int length = outTemp.length();
+ char* buf2 = new char[length];
+ memcpy(buf2, outTemp.c_str(), length);
+
+ MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
+ delete buf2;
+
+ MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
+
+ //send file positions to all processes
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
}
- }
- inFASTA.close();
-
- numSeqs = positions.size();
-
- int numSeqsPerProcessor = numSeqs / processors;
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- for (int i = 0; i < processors; i++) {
- long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
- }
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
+ //do your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } remove(outputFileName.c_str()); remove(accnosFileName.c_str()); delete chimera; return 0; }
+
+ }else{ //you are a child process
+ MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPIPos.resize(numSeqs+1);
+ MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
+
+ //do your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } delete chimera; return 0; }
}
+ //close files
+ MPI_File_close(&inMPI);
+ MPI_File_close(&outMPI);
+ MPI_File_close(&outMPIAccnos);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
- createProcesses(outputFileName, fastafile, accnosFileName);
+ #else
+ ofstream outHeader;
+ string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
+ m->openOutputFile(tempHeader, outHeader);
- rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
-
- //append output files
- for(int i=1;i<processors;i++){
- appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
- remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
- }
+ chimera->printHeader(outHeader);
+ outHeader.close();
+
+ vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
- vector<string> nonBlankAccnosFiles;
- //delete blank accnos files generated with multiple processes
- for(int i=0;i<processors;i++){
- if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
- nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
- }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str()); }
- }
+ for (int i = 0; i < (positions.size()-1); i++) {
+ lines.push_back(new linePair(positions[i], positions[(i+1)]));
+ }
+
+ //break up file
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ if(processors == 1){
+ numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
+
+ if (m->control_pressed) { remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
+
+ }else{
+ processIDS.resize(0);
+
+ numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
- //append accnos files
- if (nonBlankAccnosFiles.size() != 0) {
- rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
+ rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
+ rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
+
+ //append output files
+ for(int i=1;i<processors;i++){
+ m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
+ remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
+ }
- for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
- appendFiles(nonBlankAccnosFiles[h], accnosFileName);
- remove(nonBlankAccnosFiles[h].c_str());
+ //append output files
+ for(int i=1;i<processors;i++){
+ m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
+ remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
}
- }else{ hasAccnos = false; }
-
- if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
+
+ if (m->control_pressed) { remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
}
- }
-
- #else
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
- lines.push_back(new linePair(0, numSeqs));
-
- driver(lines[0], outputFileName, fastafile, accnosFileName);
-
- if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(tempHeader.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
- }
+ #else
+ numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
+
+ if (m->control_pressed) { remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
+
+ #endif
- //delete accnos file if its blank
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
- #endif
+ m->appendFiles(outputFileName, tempHeader);
- appendFiles(outputFileName, tempHeader);
-
- remove(outputFileName.c_str());
- rename(tempHeader.c_str(), outputFileName.c_str());
-
- #endif
- delete chimera;
+ remove(outputFileName.c_str());
+ rename(tempHeader.c_str(), outputFileName.c_str());
+
+ #endif
+ delete chimera;
+
+
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+
+ outputNames.push_back(outputFileName);
+ outputNames.push_back(accnosFileName);
+
+ m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
+ }
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
- m->mothurOut(outputFileName); m->mothurOutEndLine();
- if (hasAccnos) { m->mothurOut(accnosFileName); m->mothurOutEndLine(); }
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
m->mothurOutEndLine();
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
-
- m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
-
return 0;
}
}
//**********************************************************************************************************************
-int ChimeraSlayerCommand::driver(linePair* line, string outputFName, string filename, string accnos){
+int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
try {
ofstream out;
- openOutputFile(outputFName, out);
+ m->openOutputFile(outputFName, out);
ofstream out2;
- openOutputFile(accnos, out2);
+ m->openOutputFile(accnos, out2);
ifstream inFASTA;
- openInputFile(filename, inFASTA);
+ m->openInputFile(filename, inFASTA);
- inFASTA.seekg(line->start);
-
- for(int i=0;i<line->numSeqs;i++){
+ inFASTA.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ while (!done) {
if (m->control_pressed) { return 1; }
- Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA);
+ Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
//print results
chimera->print(out, out2);
}
+ count++;
}
delete candidateSeq;
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ unsigned long int pos = inFASTA.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (inFASTA.eof()) { break; }
+ #endif
+
//report progress
- if((i+1) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(i+1)); m->mothurOutEndLine(); }
+ if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
}
//report progress
- if((line->numSeqs) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(line->numSeqs)); m->mothurOutEndLine(); }
+ if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
out.close();
out2.close();
inFASTA.close();
- return 0;
+ return count;
}
catch(exception& e) {
m->errorOut(e, "ChimeraSlayerCommand", "driver");
}
//**********************************************************************************************************************
#ifdef USE_MPI
-int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<long>& MPIPos){
+int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
try {
MPI_Status status;
int pid;
delete buf4;
- Sequence* candidateSeq = new Sequence(iss); gobble(iss);
+ Sequence* candidateSeq = new Sequence(iss); m->gobble(iss);
if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
//cout << "about to print" << endl;
//print results
bool isChimeric = chimera->print(outMPI, outAccMPI);
- if (isChimeric) { MPIWroteAccnos = true; }
-
}
}
delete candidateSeq;
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
- // processIDS.resize(0);
+ int num = 0;
//loop through and create all the processes you want
while (process != processors) {
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){
- driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
+ num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
+
+ //pass numSeqs to parent
+ ofstream out;
+ string tempFile = outputFileName + toString(getpid()) + ".num.temp";
+ m->openOutputFile(tempFile, out);
+ out << num << endl;
+ out.close();
+
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
- return 0;
+ for (int i = 0; i < processIDS.size(); i++) {
+ ifstream in;
+ string tempFile = outputFileName + toString(processIDS[i]) + ".num.temp";
+ m->openInputFile(tempFile, in);
+ if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
+ in.close(); remove(tempFile.c_str());
+ }
+
+ return num;
#endif
}
catch(exception& e) {
projects / mothur.git / blobdiff