X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=classifyseqscommand.cpp;h=a95f436dd7ffdf4403d316be77a1f8ab580f21f5;hb=9013e13ecfb2fda3c2664a76f76cc99b8c7fa74c;hp=39dda9ffd5fb29bb872ce4ccb205882d69926c66;hpb=832d53a9dfac6b1795735eec643d8cf627b0d8e3;p=mothur.git diff --git a/classifyseqscommand.cpp b/classifyseqscommand.cpp index 39dda9f..a95f436 100644 --- a/classifyseqscommand.cpp +++ b/classifyseqscommand.cpp @@ -11,11 +11,12 @@ #include "sequence.hpp" #include "bayesian.h" #include "phylotree.h" +#include "phylosummary.h" #include "knn.h" //********************************************************************************************************************** -ClassifySeqsCommand::ClassifySeqsCommand(string option){ +ClassifySeqsCommand::ClassifySeqsCommand(string option) { try { abort = false; @@ -25,56 +26,129 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ else { //valid paramters for this command - string AlignArray[] = {"template","fasta","name","search","ksize","method","processors","taxonomy","match","mismatch","gapopen","gapextend","numwanted","cutoff","probs","iters"}; + string AlignArray[] = {"template","fasta","name","group","search","ksize","method","processors","taxonomy","match","mismatch","gapopen","gapextend","numwanted","cutoff","probs","iters", "outputdir","inputdir"}; vector myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string))); OptionParser parser(option); map parameters = parser.getParameters(); - ValidParameters validParameter; + ValidParameters validParameter("classify.seqs"); + map::iterator it; //check to make sure all parameters are valid for command - for (map::iterator it = parameters.begin(); it != parameters.end(); it++) { + for (it = parameters.begin(); it != parameters.end(); it++) { if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { 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 = ""; } + + //if the user changes the input directory command factory will send this info to us in the output parameter + string inputDir = validParameter.validFile(parameters, "inputdir", false); + if (inputDir == "not found"){ inputDir = ""; } + else { + string path; + it = parameters.find("template"); + //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["template"] = inputDir + it->second; } + } + + it = parameters.find("taxonomy"); + //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["taxonomy"] = inputDir + it->second; } + } + + it = parameters.find("group"); + //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["group"] = inputDir + it->second; } + } + } + //check for required parameters templateFileName = validParameter.validFile(parameters, "template", true); if (templateFileName == "not found") { - mothurOut("template is a required parameter for the classify.seqs command."); - mothurOutEndLine(); + m->mothurOut("template 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") { mothurOut("fasta is a required parameter for the classify.seqs command."); mothurOutEndLine(); abort = true; } + if (fastaFileName == "not found") { m->mothurOut("fasta is a required parameter for the classify.seqs command."); m->mothurOutEndLine(); abort = true; } else { splitAtDash(fastaFileName, 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 = 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; + + #ifdef USE_MPI + int pid; + 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 + + if (pid == 0) { + #endif + ifstream in; - ableToOpen = openInputFile(fastaFileNames[i], in); + ableToOpen = 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() + getSimpleName(fastaFileNames[i]); + m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine(); + ableToOpen = openInputFile(tryPath, in, "noerror"); + fastaFileNames[i] = tryPath; + } + } + in.close(); + + #ifdef USE_MPI + for (int j = 1; j < processors; j++) { + MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD); + } + }else{ + MPI_Status status; + MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status); + } + + #endif + if (ableToOpen == 1) { - mothurOut(fastaFileNames[i] + " will be disregarded."); mothurOutEndLine(); + m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); //erase from file list fastaFileNames.erase(fastaFileNames.begin()+i); i--; } - in.close(); + } //make sure there is at least one valid file left - if (fastaFileNames.size() == 0) { mothurOut("no valid files."); mothurOutEndLine(); abort = true; } + if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; } } taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true); if (taxonomyFileName == "not found") { - mothurOut("taxonomy is a required parameter for the classify.seqs command."); - mothurOutEndLine(); + m->mothurOut("taxonomy is a required parameter for the classify.seqs command."); + m->mothurOutEndLine(); abort = true; } else if (taxonomyFileName == "not open") { abort = true; } @@ -82,21 +156,126 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ namefile = validParameter.validFile(parameters, "name", false); if (namefile == "not found") { namefile = ""; } + else { splitAtDash(namefile, namefileNames); //go through files and make sure they are good, if not, then disregard them for (int i = 0; i < namefileNames.size(); i++) { + if (inputDir != "") { + string path = hasPath(namefileNames[i]); + //if the user has not given a path then, add inputdir. else leave path alone. + if (path == "") { namefileNames[i] = inputDir + namefileNames[i]; } + } int ableToOpen; + + #ifdef USE_MPI + int pid; + 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 + + if (pid == 0) { + #endif + ifstream in; - ableToOpen = openInputFile(namefileNames[i], in); - if (ableToOpen == 1) { mothurOut("Unable to match name file with fasta file."); mothurOutEndLine(); abort = true; } + ableToOpen = openInputFile(namefileNames[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() + getSimpleName(namefileNames[i]); + m->mothurOut("Unable to open " + namefileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine(); + ableToOpen = openInputFile(tryPath, in, "noerror"); + namefileNames[i] = tryPath; + } + } in.close(); + + #ifdef USE_MPI + for (int j = 1; j < processors; j++) { + MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD); + } + }else{ + MPI_Status status; + MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status); + } + + #endif + + if (ableToOpen == 1) { + m->mothurOut("Unable to open " + namefileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); abort = true; + //erase from file list + namefileNames.erase(namefileNames.begin()+i); + i--; + } + } } if (namefile != "") { - if (namefileNames.size() != fastaFileNames.size()) { abort = true; mothurOut("If you provide a name file, you must have one for each fasta file."); mothurOutEndLine(); } + if (namefileNames.size() != fastaFileNames.size()) { abort = true; m->mothurOut("If you provide a name file, you must have one for each fasta file."); m->mothurOutEndLine(); } + } + + groupfile = validParameter.validFile(parameters, "group", false); + if (groupfile == "not found") { groupfile = ""; } + else { + splitAtDash(groupfile, groupfileNames); + + //go through files and make sure they are good, if not, then disregard them + for (int i = 0; i < groupfileNames.size(); i++) { + if (inputDir != "") { + string path = hasPath(groupfileNames[i]); + //if the user has not given a path then, add inputdir. else leave path alone. + if (path == "") { groupfileNames[i] = inputDir + groupfileNames[i]; } + } + int ableToOpen; + + #ifdef USE_MPI + int pid; + 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 + + if (pid == 0) { + #endif + + ifstream in; + ableToOpen = openInputFile(groupfileNames[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() + getSimpleName(groupfileNames[i]); + m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine(); + ableToOpen = openInputFile(tryPath, in, "noerror"); + groupfileNames[i] = tryPath; + } + } + in.close(); + + #ifdef USE_MPI + for (int j = 1; j < processors; j++) { + MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD); + } + }else{ + MPI_Status status; + MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status); + } + + #endif + + if (ableToOpen == 1) { + m->mothurOut("Unable to open " + groupfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); groupfileNames[i] = ""; + //erase from file list + groupfileNames.erase(groupfileNames.begin()+i); + i--; + } + } + } + + if (groupfile != "") { + if (groupfileNames.size() != fastaFileNames.size()) { abort = true; m->mothurOut("If you provide a group file, you must have one for each fasta file."); m->mothurOutEndLine(); } + }else { + for (int i = 0; i < fastaFileNames.size(); i++) { groupfileNames.push_back(""); } } //check for optional parameter and set defaults @@ -139,14 +318,14 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ if ((method == "bayesian") && (search != "kmer")) { - mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); mothurOutEndLine(); + m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine(); search = "kmer"; } } } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand"); + m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand"); exit(1); } } @@ -164,30 +343,35 @@ ClassifySeqsCommand::~ClassifySeqsCommand(){ void ClassifySeqsCommand::help(){ try { - mothurOut("The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n"); - mothurOut("The classify.seqs command parameters are template, fasta, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n"); - mothurOut("The template, fasta and taxonomy parameters are required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n"); - mothurOut("The search parameter allows you to specify the method to find most similar template. Your options are: suffix, kmer and blast. The default is kmer.\n"); - mothurOut("The method parameter allows you to specify classification method to use. Your options are: bayesian and knn. The default is bayesian.\n"); - mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n"); - mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n"); - mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n"); - mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n"); - mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n"); - mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n"); - mothurOut("The numwanted parameter allows you to specify the number of sequence matches you want with the knn method. The default is 10.\n"); - mothurOut("The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy. The default is 0.\n"); - mothurOut("The probs parameter shut off the bootstrapping results for the bayesian method. The default is true, meaning you want the bootstrapping to be run.\n"); - mothurOut("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"); - mothurOut("The classify.seqs command should be in the following format: \n"); - mothurOut("classify.seqs(template=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n"); - mothurOut("Example classify.seqs(fasta=amazon.fasta, template=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n"); - mothurOut("The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n"); - mothurOut("The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n"); - mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n"); + m->mothurOut("The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n"); + m->mothurOut("The classify.seqs command parameters are template, fasta, name, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n"); + m->mothurOut("The template, fasta and taxonomy parameters are required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n"); + m->mothurOut("The search parameter allows you to specify the method to find most similar template. Your options are: suffix, kmer, blast and distance. The default is kmer.\n"); + m->mothurOut("The name parameter allows you add a names file with your fasta file, if you enter multiple fasta files, you must enter matching names files for them.\n"); + m->mothurOut("The group parameter allows you add a group file so you can have the summary totals broken up by group.\n"); + m->mothurOut("The method parameter allows you to specify classification method to use. Your options are: bayesian and knn. The default is bayesian.\n"); + m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n"); + m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n"); + #ifdef USE_MPI + m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n"); + #endif + m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n"); + m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n"); + m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n"); + m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n"); + m->mothurOut("The numwanted parameter allows you to specify the number of sequence matches you want with the knn method. The default is 10.\n"); + m->mothurOut("The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy. The default is 0.\n"); + m->mothurOut("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"); + m->mothurOut("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"); + m->mothurOut("The classify.seqs command should be in the following format: \n"); + m->mothurOut("classify.seqs(template=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n"); + m->mothurOut("Example classify.seqs(fasta=amazon.fasta, template=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n"); + m->mothurOut("The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n"); + m->mothurOut("The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n"); + m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n"); } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "help"); + m->errorOut(e, "ClassifySeqsCommand", "help"); exit(1); } } @@ -199,46 +383,140 @@ int ClassifySeqsCommand::execute(){ try { if (abort == true) { return 0; } - 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); } + else if(method == "knn"){ classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted); } else { - mothurOut(search + " is not a valid method option. I will run the command using bayesian."); - mothurOutEndLine(); + 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); } - + + if (m->control_pressed) { delete classify; return 0; } + + vector outputNames; for (int s = 0; s < fastaFileNames.size(); s++) { - //read namefile - if(namefile != "") { - nameMap.clear(); //remove old names - - ifstream inNames; - openInputFile(namefileNames[s], inNames); - - string firstCol, secondCol; - while(!inNames.eof()) { - inNames >> firstCol >> secondCol; gobble(inNames); - nameMap[firstCol] = getNumNames(secondCol); //ex. seq1 seq1,seq3,seq5 -> seq1 = 3. - } - inNames.close(); - } + m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine(); + + string RippedTaxName = getRootName(getSimpleName(taxonomyFileName)); + RippedTaxName = getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1)); + if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); } + RippedTaxName += "."; - mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); mothurOutEndLine(); - string newTaxonomyFile = getRootName(fastaFileNames[s]) + getRootName(taxonomyFileName) + "taxonomy"; - string tempTaxonomyFile = getRootName(fastaFileNames[s]) + "taxonomy.temp"; - string taxSummary = getRootName(fastaFileNames[s]) + getRootName(taxonomyFileName) + "tax.summary"; + if (outputDir == "") { outputDir += hasPath(fastaFileNames[s]); } + string newTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + RippedTaxName + "taxonomy"; + string tempTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "taxonomy.temp"; + string taxSummary = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + RippedTaxName + "tax.summary"; + + outputNames.push_back(newTaxonomyFile); + outputNames.push_back(taxSummary); int start = time(NULL); int numFastaSeqs = 0; for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) +#ifdef USE_MPI + int pid, end, numSeqsPerProcessor; + int tag = 2001; + vector MPIPos; + + 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 outMPINewTax; + MPI_File outMPITempTax; + + int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY; + int inMode=MPI_MODE_RDONLY; + + //char* outNewTax = new char[newTaxonomyFile.length()]; + //memcpy(outNewTax, newTaxonomyFile.c_str(), newTaxonomyFile.length()); + + char outNewTax[1024]; + strcpy(outNewTax, newTaxonomyFile.c_str()); + + //char* outTempTax = new char[tempTaxonomyFile.length()]; + //memcpy(outTempTax, tempTaxonomyFile.c_str(), tempTaxonomyFile.length()); + + char outTempTax[1024]; + strcpy(outTempTax, tempTaxonomyFile.c_str()); + + //char* inFileName = new char[fastaFileNames[s].length()]; + //memcpy(inFileName, fastaFileNames[s].c_str(), fastaFileNames[s].length()); + + 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, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax); + MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax); + + //delete outNewTax; + //delete outTempTax; + //delete inFileName; + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; } + + if (pid == 0) { //you are the root process + + MPIPos = setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs + + //send file positions to all processes + for(int i = 1; i < processors; i++) { + MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD); + MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD); + } + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + int startIndex = pid * numSeqsPerProcessor; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + + + //align your part + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos); + + if (m->control_pressed) { 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; } + + for (int i = 1; i < processors; i++) { + int done; + MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status); + } + }else{ //you are a child process + MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status); + MPIPos.resize(numFastaSeqs+1); + MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status); + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + int startIndex = pid * numSeqsPerProcessor; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + + + //align your part + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos); + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; } + + int done = 0; + MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD); + } + + //close files + MPI_File_close(&inMPI); + MPI_File_close(&outMPINewTax); + MPI_File_close(&outMPITempTax); + MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case + +#else + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) if(processors == 1){ ifstream inFASTA; openInputFile(fastaFileNames[s], inFASTA); - numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); + getNumSeqs(inFASTA, numFastaSeqs); inFASTA.close(); lines.push_back(new linePair(0, numFastaSeqs)); @@ -246,7 +524,7 @@ int ClassifySeqsCommand::execute(){ driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); } else{ - vector positions; + vector positions; processIDS.resize(0); ifstream inFASTA; @@ -256,7 +534,7 @@ int ClassifySeqsCommand::execute(){ while(!inFASTA.eof()){ input = getline(inFASTA); if (input.length() != 0) { - if(input[0] == '>'){ int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); } + if(input[0] == '>'){ unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); } } } inFASTA.close(); @@ -264,9 +542,9 @@ int ClassifySeqsCommand::execute(){ numFastaSeqs = positions.size(); int numSeqsPerProcessor = numFastaSeqs / processors; - + for (int i = 0; i < processors; i++) { - int startPos = positions[ i * numSeqsPerProcessor ]; + unsigned long int startPos = positions[ i * numSeqsPerProcessor ]; if(i == processors - 1){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; } @@ -285,51 +563,93 @@ int ClassifySeqsCommand::execute(){ } } -#else + #else ifstream inFASTA; openInputFile(fastaFileNames[s], inFASTA); - numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); + getNumSeqs(inFASTA, numFastaSeqs); inFASTA.close(); lines.push_back(new linePair(0, numFastaSeqs)); driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); -#endif - //make taxonomy tree from new taxonomy file - PhyloTree taxaBrowser; - - ifstream in; - openInputFile(tempTaxonomyFile, in); + #endif +#endif + + 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 + + if(namefile != "") { m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush(); MPIReadNamesFile(namefileNames[s]); m->mothurOut(" Done."); m->mothurOutEndLine(); } + #else + //read namefile + if(namefile != "") { + + m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush(); + + nameMap.clear(); //remove old names + + ifstream inNames; + openInputFile(namefileNames[s], inNames); + + string firstCol, secondCol; + while(!inNames.eof()) { + inNames >> firstCol >> secondCol; gobble(inNames); + + vector temp; + splitAtComma(secondCol, temp); + + nameMap[firstCol] = temp; + } + inNames.close(); + + m->mothurOut(" Done."); m->mothurOutEndLine(); + } + #endif + + string group = ""; + if (groupfile != "") { group = groupfileNames[s]; } + + PhyloSummary taxaSum(taxonomyFileName, group); + + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } - //read in users taxonomy file and add sequences to tree - string name, taxon; - while(!in.eof()){ - in >> name >> taxon; gobble(in); + if (namefile == "") { taxaSum.summarize(tempTaxonomyFile); } + else { + ifstream in; + openInputFile(tempTaxonomyFile, in); + + //read in users taxonomy file and add sequences to tree + string name, taxon; - if (namefile != "") { + while(!in.eof()){ + in >> name >> taxon; gobble(in); + itNames = nameMap.find(name); if (itNames == nameMap.end()) { - mothurOut(name + " is not in your name file please correct."); mothurOutEndLine(); exit(1); + m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1); }else{ - for (int i = 0; i < itNames->second; i++) { - taxaBrowser.addSeqToTree(name+toString(i), taxon); //add it as many times as there are identical seqs + for (int i = 0; i < itNames->second.size(); i++) { + taxaSum.addSeqToTree(itNames->second[i], taxon); //add it as many times as there are identical seqs } + itNames->second.clear(); + nameMap.erase(itNames->first); } - }else { taxaBrowser.addSeqToTree(name, taxon); } //add it once + } + in.close(); } - in.close(); - - taxaBrowser.assignHeirarchyIDs(0); - - taxaBrowser.binUnclassified(); - remove(tempTaxonomyFile.c_str()); + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } + //print summary file ofstream outTaxTree; openOutputFile(taxSummary, outTaxTree); - taxaBrowser.print(outTaxTree); + taxaSum.print(outTaxTree); outTaxTree.close(); //output taxonomy with the unclassified bins added @@ -341,10 +661,13 @@ int ClassifySeqsCommand::execute(){ openOutputFile(unclass, outTax); //get maxLevel from phylotree so you know how many 'unclassified's to add - int maxLevel = taxaBrowser.getMaxLevel(); + int maxLevel = taxaSum.getMaxLevel(); - //read taxfile - this reading and rewriting is done to preserve the confidence sscores. + //read taxfile - this reading and rewriting is done to preserve the confidence scores. + string name, taxon; while (!inTax.eof()) { + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } remove(unclass.c_str()); delete classify; return 0; } + inTax >> name >> taxon; gobble(inTax); string newTax = addUnclassifieds(taxon, maxLevel); @@ -357,15 +680,24 @@ int ClassifySeqsCommand::execute(){ remove(newTaxonomyFile.c_str()); rename(unclass.c_str(), newTaxonomyFile.c_str()); - mothurOutEndLine(); - mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); mothurOutEndLine(); mothurOutEndLine(); + m->mothurOutEndLine(); + m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine(); + + #ifdef USE_MPI + } + #endif + + m->mothurOutEndLine(); + m->mothurOut("Output File Names: "); m->mothurOutEndLine(); + for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); } + m->mothurOutEndLine(); } delete classify; return 0; } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "execute"); + m->errorOut(e, "ClassifySeqsCommand", "execute"); exit(1); } } @@ -379,7 +711,7 @@ string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) { //keep what you have counting the levels while (tax.find_first_of(';') != -1) { //get taxon - taxon = tax.substr(0,tax.find_first_of(';')); + taxon = tax.substr(0,tax.find_first_of(';'))+';'; tax = tax.substr(tax.find_first_of(';')+1, tax.length()); newTax += taxon; level++; @@ -394,7 +726,7 @@ string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) { return newTax; } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "addUnclassifieds"); + m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds"); exit(1); } } @@ -417,7 +749,7 @@ void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile }else if (pid == 0){ driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", filename); exit(0); - }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); 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 @@ -428,7 +760,7 @@ void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile #endif } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "createProcesses"); + m->errorOut(e, "ClassifySeqsCommand", "createProcesses"); exit(1); } } @@ -451,7 +783,7 @@ void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) { output.close(); } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "appendTaxFiles"); + m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles"); exit(1); } } @@ -474,12 +806,15 @@ int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFNa string taxonomy; for(int i=0;inumSeqs;i++){ - - Sequence* candidateSeq = new Sequence(inFASTA); - + if (m->control_pressed) { return 0; } + + Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA); + if (candidateSeq->getName() != "") { taxonomy = classify->getTaxonomy(candidateSeq); + if (m->control_pressed) { delete candidateSeq; return 0; } + if (taxonomy != "bad seq") { //output confidence scores or not if (probs) { @@ -494,7 +829,7 @@ int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFNa delete candidateSeq; if((i+1) % 100 == 0){ - mothurOut("Classifying sequence " + toString(i+1)); mothurOutEndLine(); + m->mothurOut("Classifying sequence " + toString(i+1)); m->mothurOutEndLine(); } } @@ -505,9 +840,129 @@ int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFNa return 1; } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "driver"); + m->errorOut(e, "ClassifySeqsCommand", "driver"); exit(1); } } +//********************************************************************************************************************** +#ifdef USE_MPI +int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, vector& MPIPos){ + try { + MPI_Status statusNew; + MPI_Status statusTemp; + MPI_Status status; + + int pid; + MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are + + string taxonomy; + string outputString; + + for(int i=0;icontrol_pressed) { return 0; } + + //read next sequence + int length = MPIPos[start+i+1] - MPIPos[start+i]; + char* buf4 = new char[length]; + MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status); + + string tempBuf = buf4; + if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); } + istringstream iss (tempBuf,istringstream::in); + delete buf4; + Sequence* candidateSeq = new Sequence(iss); + + 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); + + 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; + } + } + delete candidateSeq; + + if((i+1) % 100 == 0){ cout << "Classifying sequence " << (i+1) << endl; } + } + + if(num % 100 != 0){ cout << "Classifying sequence " << (num) << endl; } + + + return 1; + } + catch(exception& e) { + m->errorOut(e, "ClassifySeqsCommand", "driverMPI"); + exit(1); + } +} + +//********************************************************************************************************************** +int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){ + try { + + nameMap.clear(); //remove old names + + MPI_File inMPI; + MPI_Offset size; + MPI_Status status; + + //char* inFileName = new char[nameFilename.length()]; + //memcpy(inFileName, nameFilename.c_str(), nameFilename.length()); + + char inFileName[1024]; + strcpy(inFileName, nameFilename.c_str()); + + MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); + MPI_File_get_size(inMPI, &size); + //delete inFileName; + + char* buffer = new char[size]; + MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status); + + string tempBuf = buffer; + if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size); } + istringstream iss (tempBuf,istringstream::in); + delete buffer; + + string firstCol, secondCol; + while(!iss.eof()) { + iss >> firstCol >> secondCol; gobble(iss); + + vector temp; + splitAtComma(secondCol, temp); + + nameMap[firstCol] = temp; + } + + MPI_File_close(&inMPI); + + return 1; + } + catch(exception& e) { + m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile"); + exit(1); + } +} +#endif /**************************************************************************************************/