X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=classifyseqscommand.cpp;h=0504e6635802c3f6e3e186e4568675906774b3ff;hb=91a27e0483827c06c21c4fe89558923bbfe86573;hp=9d5a625308f8970888833ec7e74ec4b2e2ec32dd;hpb=86c838c428a9e7d26f902f5492738241fa72c4e7;p=mothur.git diff --git a/classifyseqscommand.cpp b/classifyseqscommand.cpp index 9d5a625..0504e66 100644 --- a/classifyseqscommand.cpp +++ b/classifyseqscommand.cpp @@ -8,11 +8,7 @@ */ #include "classifyseqscommand.h" -#include "sequence.hpp" -#include "bayesian.h" -#include "phylotree.h" -#include "phylosummary.h" -#include "knn.h" + //********************************************************************************************************************** @@ -31,9 +27,11 @@ vector ClassifySeqsCommand::setParameters(){ CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pmismatch); CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "",false,false); parameters.push_back(pgapopen); CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend); + //CommandParameter pflip("flip", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pflip); CommandParameter pcutoff("cutoff", "Number", "", "0", "", "", "",false,true); parameters.push_back(pcutoff); CommandParameter pprobs("probs", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pprobs); CommandParameter piters("iters", "Number", "", "100", "", "", "",false,true); parameters.push_back(piters); + CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave); CommandParameter pnumwanted("numwanted", "Number", "", "10", "", "", "",false,true); parameters.push_back(pnumwanted); CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir); CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir); @@ -63,6 +61,7 @@ string ClassifySeqsCommand::getHelpString(){ #ifdef USE_MPI helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n"; #endif + helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f."; helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n"; helpString += "The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n"; helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n"; @@ -71,6 +70,7 @@ string ClassifySeqsCommand::getHelpString(){ helpString += "The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy. The default is 0.\n"; helpString += "The probs parameter shuts off the bootstrapping results for the bayesian method. The default is true, meaning you want the bootstrapping to be shown.\n"; helpString += "The iters parameter allows you to specify how many iterations to do when calculating the bootstrap confidence score for your taxonomy with the bayesian method. The default is 100.\n"; + //helpString += "The flip parameter allows you shut off mothur's The default is T.\n"; helpString += "The classify.seqs command should be in the following format: \n"; helpString += "classify.seqs(reference=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n"; helpString += "Example classify.seqs(fasta=amazon.fasta, reference=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n"; @@ -91,6 +91,7 @@ ClassifySeqsCommand::ClassifySeqsCommand(){ setParameters(); vector tempOutNames; outputTypes["taxonomy"] = tempOutNames; + outputTypes["accnos"] = tempOutNames; outputTypes["taxsummary"] = tempOutNames; outputTypes["matchdist"] = tempOutNames; } @@ -103,9 +104,11 @@ ClassifySeqsCommand::ClassifySeqsCommand(){ ClassifySeqsCommand::ClassifySeqsCommand(string option) { try { abort = false; calledHelp = false; + rdb = ReferenceDB::getInstance(); //allow user to run help if(option == "help") { help(); abort = true; calledHelp = true; } + else if(option == "citation") { citation(); abort = true; calledHelp = true;} else { vector myArray = setParameters(); @@ -126,6 +129,7 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { outputTypes["taxonomy"] = tempOutNames; outputTypes["taxsummary"] = tempOutNames; outputTypes["matchdist"] = tempOutNames; + outputTypes["accnos"] = tempOutNames; //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 = ""; } @@ -160,16 +164,6 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { } } - //check for required parameters - templateFileName = validParameter.validFile(parameters, "reference", true); - if (templateFileName == "not found") { - m->mothurOut("reference is a required parameter for the classify.seqs command."); - m->mothurOutEndLine(); - abort = true; - } - else if (templateFileName == "not open") { abort = true; } - - fastaFileName = validParameter.validFile(parameters, "fasta", false); if (fastaFileName == "not found") { //if there is a current fasta file, use it @@ -238,6 +232,8 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { //erase from file list fastaFileNames.erase(fastaFileNames.begin()+i); i--; + }else { + m->setFastaFile(fastaFileNames[i]); } } @@ -247,16 +243,6 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; } } - - taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true); - if (taxonomyFileName == "not found") { - m->mothurOut("taxonomy is a required parameter for the classify.seqs command."); - m->mothurOutEndLine(); - abort = true; - } - else if (taxonomyFileName == "not open") { abort = true; } - - namefile = validParameter.validFile(parameters, "name", false); if (namefile == "not found") { namefile = ""; } @@ -318,6 +304,8 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { //erase from file list namefileNames.erase(namefileNames.begin()+i); i--; + }else { + m->setNameFile(namefileNames[i]); } } } @@ -374,6 +362,8 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { //erase from file list groupfileNames.erase(groupfileNames.begin()+i); i--; + }else { + m->setGroupFile(groupfileNames[i]); } } } @@ -388,46 +378,94 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { // ...at some point should added some additional type checking... string temp; temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; } - convert(temp, kmerSize); + m->mothurConvert(temp, kmerSize); temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); } m->setProcessors(temp); - convert(temp, processors); + m->mothurConvert(temp, processors); + + temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; } + save = m->isTrue(temp); + rdb->save = save; + if (save) { //clear out old references + rdb->clearMemory(); + } + + //this has to go after save so that if the user sets save=t and provides no reference we abort + templateFileName = validParameter.validFile(parameters, "reference", true); + if (templateFileName == "not found") { + //check for saved reference sequences + if (rdb->referenceSeqs.size() != 0) { + templateFileName = "saved"; + }else { + m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the classify.seqs command."); + m->mothurOutEndLine(); + abort = true; + } + }else if (templateFileName == "not open") { abort = true; } + else { if (save) { rdb->setSavedReference(templateFileName); } } + + //this has to go after save so that if the user sets save=t and provides no reference we abort + taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true); + if (taxonomyFileName == "not found") { + //check for saved reference sequences + if (rdb->wordGenusProb.size() != 0) { + taxonomyFileName = "saved"; + }else { + m->mothurOut("[ERROR]: You don't have any saved taxonomy information and the taxonomy parameter is a required for the classify.seqs command."); + m->mothurOutEndLine(); + abort = true; + } + }else if (taxonomyFileName == "not open") { abort = true; } + else { if (save) { rdb->setSavedTaxonomy(taxonomyFileName); } } search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; } method = validParameter.validFile(parameters, "method", false); if (method == "not found"){ method = "bayesian"; } temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; } - convert(temp, match); + m->mothurConvert(temp, match); temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; } - convert(temp, misMatch); + m->mothurConvert(temp, misMatch); temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; } - convert(temp, gapOpen); + m->mothurConvert(temp, gapOpen); temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; } - convert(temp, gapExtend); + m->mothurConvert(temp, gapExtend); temp = validParameter.validFile(parameters, "numwanted", false); if (temp == "not found"){ temp = "10"; } - convert(temp, numWanted); + m->mothurConvert(temp, numWanted); temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found"){ temp = "0"; } - convert(temp, cutoff); + m->mothurConvert(temp, cutoff); temp = validParameter.validFile(parameters, "probs", false); if (temp == "not found"){ temp = "true"; } probs = m->isTrue(temp); + //temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "T"; } + //flip = m->isTrue(temp); + flip = true; + temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "100"; } - convert(temp, iters); - + m->mothurConvert(temp, iters); if ((method == "bayesian") && (search != "kmer")) { m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine(); search = "kmer"; } + + if (!abort) { + if (namefileNames.size() == 0){ + if (fastaFileNames.size() != 0) { + vector files; files.push_back(fastaFileNames[fastaFileNames.size()-1]); + parser.getNameFile(files); + } + } + } + } } @@ -449,28 +487,31 @@ int ClassifySeqsCommand::execute(){ try { if (abort == true) { if (calledHelp) { return 0; } return 2; } - if(method == "bayesian"){ classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters); } - else if(method == "knn"){ classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted); } + if(method == "bayesian"){ classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip); } + else if(method == "knn"){ classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted, rand()); } else { m->mothurOut(search + " is not a valid method option. I will run the command using bayesian."); m->mothurOutEndLine(); - classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters); + classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip); } if (m->control_pressed) { delete classify; return 0; } - for (int s = 0; s < fastaFileNames.size(); s++) { m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine(); - string RippedTaxName = m->getRootName(m->getSimpleName(taxonomyFileName)); + string baseTName = taxonomyFileName; + if (taxonomyFileName == "saved") {baseTName = rdb->getSavedTaxonomy(); } + + string RippedTaxName = m->getRootName(m->getSimpleName(baseTName)); RippedTaxName = m->getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1)); if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); } RippedTaxName += "."; if (outputDir == "") { outputDir += m->hasPath(fastaFileNames[s]); } string newTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "taxonomy"; + string newaccnosFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "flip.accnos"; string tempTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "taxonomy.temp"; string taxSummary = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "tax.summary"; @@ -480,6 +521,7 @@ int ClassifySeqsCommand::execute(){ } outputNames.push_back(newTaxonomyFile); outputTypes["taxonomy"].push_back(newTaxonomyFile); + outputNames.push_back(newaccnosFile); outputTypes["accnos"].push_back(newaccnosFile); outputNames.push_back(taxSummary); outputTypes["taxsummary"].push_back(taxSummary); int start = time(NULL); @@ -489,7 +531,7 @@ int ClassifySeqsCommand::execute(){ #ifdef USE_MPI int pid, numSeqsPerProcessor; int tag = 2001; - vector MPIPos; + vector MPIPos; MPI_Status status; MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are @@ -498,6 +540,7 @@ int ClassifySeqsCommand::execute(){ MPI_File inMPI; MPI_File outMPINewTax; MPI_File outMPITempTax; + MPI_File outMPIAcc; int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY; int inMode=MPI_MODE_RDONLY; @@ -507,6 +550,9 @@ int ClassifySeqsCommand::execute(){ char outTempTax[1024]; strcpy(outTempTax, tempTaxonomyFile.c_str()); + + char outAcc[1024]; + strcpy(outAcc, newaccnosFile.c_str()); char inFileName[1024]; strcpy(inFileName, fastaFileNames[s].c_str()); @@ -514,8 +560,9 @@ int ClassifySeqsCommand::execute(){ MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax); MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax); + MPI_File_open(MPI_COMM_WORLD, outAcc, outMode, MPI_INFO_NULL, &outMPIAcc); - if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); delete classify; return 0; } if (pid == 0) { //you are the root process @@ -534,9 +581,9 @@ int ClassifySeqsCommand::execute(){ //align your part - driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos); + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos); - if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; } for (int i = 1; i < processors; i++) { int done; @@ -554,9 +601,9 @@ int ClassifySeqsCommand::execute(){ //align your part - driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos); + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos); - if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); delete classify; return 0; } int done = 0; MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD); @@ -566,35 +613,45 @@ int ClassifySeqsCommand::execute(){ MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); + MPI_File_close(&outMPIAcc); MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case #else - vector positions = m->divideFile(fastaFileNames[s], processors); + vector positions; +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + 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)])); } +#else + if (processors == 1) { + lines.push_back(new linePair(0, 1000)); + }else { + positions = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs); + if (positions.size() < processors) { processors = positions.size(); } - for (int i = 0; i < (positions.size()-1); i++) { - lines.push_back(new linePair(positions[i], positions[(i+1)])); - } - - #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - if(processors == 1){ - numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); + //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)); + } } - else{ - processIDS.resize(0); - - numFastaSeqs = createProcesses(newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); - +#endif + if(processors == 1){ + numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]); + }else{ + numFastaSeqs = createProcesses(newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]); } - #else - numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); - #endif #endif + + if (!m->isBlank(newaccnosFile)) { m->mothurOutEndLine(); m->mothurOut("[WARNING]: mothur suspects some of your sequences may be reversed, please check " + newaccnosFile + " for the list of the sequences."); m->mothurOutEndLine(); } m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine(); start = time(NULL); - + + #ifdef USE_MPI if (pid == 0) { //this part does not need to be paralellized @@ -629,9 +686,9 @@ int ClassifySeqsCommand::execute(){ string group = ""; if (groupfile != "") { group = groupfileNames[s]; } - PhyloSummary taxaSum(taxonomyFileName, group); + PhyloSummary taxaSum(baseTName, group); - if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; } if (namefile == "") { taxaSum.summarize(tempTaxonomyFile); } else { @@ -658,9 +715,9 @@ int ClassifySeqsCommand::execute(){ } in.close(); } - remove(tempTaxonomyFile.c_str()); + m->mothurRemove(tempTaxonomyFile); - if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; } //print summary file ofstream outTaxTree; @@ -682,7 +739,7 @@ int ClassifySeqsCommand::execute(){ //read taxfile - this reading and rewriting is done to preserve the confidence scores. string name, taxon; while (!inTax.eof()) { - if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } remove(unclass.c_str()); delete classify; return 0; } + if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->mothurRemove(unclass); delete classify; return 0; } inTax >> name >> taxon; m->gobble(inTax); @@ -693,7 +750,7 @@ int ClassifySeqsCommand::execute(){ inTax.close(); outTax.close(); - remove(newTaxonomyFile.c_str()); + m->mothurRemove(newTaxonomyFile); rename(unclass.c_str(), newTaxonomyFile.c_str()); m->mothurOutEndLine(); @@ -716,7 +773,14 @@ int ClassifySeqsCommand::execute(){ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); } } + current = ""; + itTypes = outputTypes.find("accnos"); + if (itTypes != outputTypes.end()) { + if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); } + } + delete classify; + return 0; } catch(exception& e) { @@ -756,11 +820,14 @@ string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) { /**************************************************************************************************/ -int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string filename) { +int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string accnos, string filename) { try { -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - int process = 1; + int num = 0; + processIDS.clear(); + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + int process = 1; //loop through and create all the processes you want while (process != processors) { @@ -770,8 +837,8 @@ int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, 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){ - num = driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", filename); - + num = driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", filename); + //pass numSeqs to parent ofstream out; string tempFile = filename + toString(getpid()) + ".num.temp"; @@ -788,7 +855,7 @@ int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, } //parent does its part - num = driver(lines[0], taxFileName, tempTaxFile, filename); + num = driver(lines[0], taxFileName, tempTaxFile, accnos, filename); //force parent to wait until all the processes are done for (int i=0;iopenInputFile(tempFile, in); if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; } - in.close(); remove(tempFile.c_str()); + in.close(); m->mothurRemove(m->getFullPathName(tempFile)); + } +#else + ////////////////////////////////////////////////////////////////////////////////////////////////////// + //Windows version shared memory, so be careful when passing variables through the alignData struct. + //Above fork() will clone, so memory is separate, but that's not the case with windows, + ////////////////////////////////////////////////////////////////////////////////////////////////////// + + vector pDataArray; + DWORD dwThreadIdArray[processors-1]; + HANDLE hThreadArray[processors-1]; + + //Create processor worker threads. + for( int i=0; istart, lines[i]->end, match, misMatch, gapOpen, gapExtend, cutoff, i, flip); + pDataArray.push_back(tempclass); + + //MySeqSumThreadFunction is in header. It must be global or static to work with the threads. + //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier + hThreadArray[i] = CreateThread(NULL, 0, MyClassThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]); + } + //parent does its part + num = driver(lines[processors-1], taxFileName + toString(processors-1) + ".temp", tempTaxFile + toString(processors-1) + ".temp", accnos + toString(processors-1) + ".temp", filename); + processIDS.push_back((processors-1)); + + //Wait until all threads have terminated. + WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE); + + //Close all thread handles and free memory allocations. + for(int i=0; i < pDataArray.size(); i++){ + num += pDataArray[i]->count; + CloseHandle(hThreadArray[i]); + delete pDataArray[i]; + } + + #endif + vector nonBlankAccnosFiles; + if (!(m->isBlank(accnos))) { nonBlankAccnosFiles.push_back(accnos); } + else { m->mothurRemove(accnos); } //remove so other files can be renamed to it + for(int i=0;iisBlank(accnos + toString(processIDS[i]) + ".temp"))) { + nonBlankAccnosFiles.push_back(accnos + toString(processIDS[i]) + ".temp"); + }else { m->mothurRemove((accnos + toString(processIDS[i]) + ".temp")); } + + m->mothurRemove((m->getFullPathName(taxFileName) + toString(processIDS[i]) + ".temp")); + m->mothurRemove((m->getFullPathName(tempTaxFile) + toString(processIDS[i]) + ".temp")); } + //append accnos files + if (nonBlankAccnosFiles.size() != 0) { + rename(nonBlankAccnosFiles[0].c_str(), accnos.c_str()); + + for (int h=1; h < nonBlankAccnosFiles.size(); h++) { + appendTaxFiles(nonBlankAccnosFiles[h], accnos); + m->mothurRemove(nonBlankAccnosFiles[h]); + } + }else { //recreate the accnosfile if needed + ofstream out; + m->openOutputFile(accnos, out); + out.close(); + } + return num; -#endif + } catch(exception& e) { m->errorOut(e, "ClassifySeqsCommand", "createProcesses"); @@ -845,13 +973,16 @@ void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) { //********************************************************************************************************************** -int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempTFName, string filename){ +int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempTFName, string accnos, string filename){ try { ofstream outTax; m->openOutputFile(taxFName, outTax); ofstream outTaxSimple; m->openOutputFile(tempTFName, outTaxSimple); + + ofstream outAcc; + m->openOutputFile(accnos, outAcc); ifstream inFASTA; m->openInputFile(filename, inFASTA); @@ -864,7 +995,11 @@ int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempT int count = 0; while (!done) { - if (m->control_pressed) { return 0; } + if (m->control_pressed) { + inFASTA.close(); + outTax.close(); + outTaxSimple.close(); + outAcc.close(); return 0; } Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA); @@ -873,23 +1008,26 @@ int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempT taxonomy = classify->getTaxonomy(candidateSeq); if (m->control_pressed) { delete candidateSeq; return 0; } - - if (taxonomy != "bad seq") { - //output confidence scores or not - if (probs) { - outTax << candidateSeq->getName() << '\t' << taxonomy << endl; - }else{ - outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl; - } - - outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl; + + if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); } + + //output confidence scores or not + if (probs) { + outTax << candidateSeq->getName() << '\t' << taxonomy << endl; + }else{ + outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl; } + + if (classify->getFlipped()) { outAcc << candidateSeq->getName() << endl; } + + outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl; + count++; } delete candidateSeq; - #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - unsigned long int pos = inFASTA.tellg(); + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix) + unsigned long long pos = inFASTA.tellg(); if ((pos == -1) || (pos >= filePos->end)) { break; } #else if (inFASTA.eof()) { break; } @@ -905,6 +1043,7 @@ int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempT inFASTA.close(); outTax.close(); outTaxSimple.close(); + outAcc.close(); return count; } @@ -915,10 +1054,11 @@ int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempT } //********************************************************************************************************************** #ifdef USE_MPI -int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, vector& MPIPos){ +int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, MPI_File& accFile, vector& MPIPos){ try { MPI_Status statusNew; MPI_Status statusTemp; + MPI_Status statusAcc; MPI_Status status; int pid; @@ -946,29 +1086,40 @@ int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File if (candidateSeq->getName() != "") { taxonomy = classify->getTaxonomy(candidateSeq); - if (taxonomy != "bad seq") { - //output confidence scores or not - if (probs) { - outputString = candidateSeq->getName() + "\t" + taxonomy + "\n"; - }else{ - outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n"; - } - - int length = outputString.length(); - char* buf2 = new char[length]; - memcpy(buf2, outputString.c_str(), length); + if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); } - MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew); - delete buf2; - + //output confidence scores or not + if (probs) { + outputString = candidateSeq->getName() + "\t" + taxonomy + "\n"; + }else{ outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n"; - length = outputString.length(); - char* buf = new char[length]; - memcpy(buf, outputString.c_str(), length); + } + + int length = outputString.length(); + char* buf2 = new char[length]; + memcpy(buf2, outputString.c_str(), length); - MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp); - delete buf; + MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew); + delete buf2; + + outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n"; + length = outputString.length(); + char* buf = new char[length]; + memcpy(buf, outputString.c_str(), length); + + MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp); + delete buf; + + if (classify->getFlipped()) { + outputString = candidateSeq->getName() + "\n"; + length = outputString.length(); + char* buf3 = new char[length]; + memcpy(buf3, outputString.c_str(), length); + + MPI_File_write_shared(accFile, buf3, length, MPI_CHAR, &statusAcc); + delete buf3; } + } delete candidateSeq;