X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=classifyseqscommand.cpp;h=26df57ffbb56d6ddd0cc3532013c2be65fe5e3f8;hb=438aa88dbc092d9c1c80ec3fa20c8e47f97101c4;hp=69ef7127182001dc7d2b292650b3be38e0445a1d;hpb=63e089e0b3aad1741bab60119ed7ccc784dce347;p=mothur.git diff --git a/classifyseqscommand.cpp b/classifyseqscommand.cpp index 69ef712..26df57f 100644 --- a/classifyseqscommand.cpp +++ b/classifyseqscommand.cpp @@ -9,16 +9,14 @@ #include "classifyseqscommand.h" #include "sequence.hpp" -#include "phylotype.h" #include "bayesian.h" -#include "doTaxonomy.h" +#include "phylotree.h" #include "knn.h" //********************************************************************************************************************** -ClassifySeqsCommand::ClassifySeqsCommand(string option){ +ClassifySeqsCommand::ClassifySeqsCommand(string option) { try { - // globaldata = GlobalData::getInstance(); abort = false; //allow user to run help @@ -27,44 +25,160 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ else { //valid paramters for this command - string AlignArray[] = {"template","fasta","search","ksize","method","processors","taxonomy","match","mismatch","gapopen","gapextend","numwanted"}; + string AlignArray[] = {"template","fasta","name","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; + 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; } + } + } + //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", true); - if (fastaFileName == "not found") { - mothurOut("fasta is a required parameter for the classify.seqs command."); - mothurOutEndLine(); - abort = true; + fastaFileName = validParameter.validFile(parameters, "fasta", false); + 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); + 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(fastaFileNames[i] + " 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; } } - else if (fastaFileName == "not open") { 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; } + + + 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); + 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 match name file with fasta file."); m->mothurOutEndLine(); abort = true; } + + } + } + + if (namefile != "") { + 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(); } + } //check for optional parameter and set defaults // ...at some point should added some additional type checking... @@ -77,7 +191,7 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; } - method = validParameter.validFile(parameters, "method", false); if (method == "not found"){ method = "phylotype"; } + 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); @@ -93,12 +207,27 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option){ temp = validParameter.validFile(parameters, "numwanted", false); if (temp == "not found"){ temp = "10"; } convert(temp, numWanted); + + temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found"){ temp = "0"; } + convert(temp, cutoff); + + temp = validParameter.validFile(parameters, "probs", false); if (temp == "not found"){ temp = "true"; } + probs = isTrue(temp); + + temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "100"; } + convert(temp, iters); + + + if ((method == "bayesian") && (search != "kmer")) { + 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); } } @@ -116,27 +245,34 @@ 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 and numwanted.\n"); - mothurOut("The template, fasta and taxonomy parameters are required.\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: phylotype, bayesian and knn. The default is phylotype.\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 -1.0.\n"); - mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -2.0.\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 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=phylotype, 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 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 shut off the bootstrapping results for the bayesian method. The default is true, meaning you want the bootstrapping to be run.\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); } } @@ -148,123 +284,328 @@ int ClassifySeqsCommand::execute(){ try { if (abort == true) { return 0; } - int start = time(NULL); - - if(method == "phylotype"){ classify = new PhyloType(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch); } - //else if(method == "bayesian") { classify = new Bayesian(taxonomyFileName, templateFileName, search); } - 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 phylotype."); - mothurOutEndLine(); - classify = new PhyloType(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch); + 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); } - - int numFastaSeqs = 0; - string newTaxonomyFile = getRootName(fastaFileName) + "taxonomy"; - string taxSummary = getRootName(fastaFileName) + "tax.summary"; + if (m->control_pressed) { delete classify; return 0; } -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - if(processors == 1){ - ifstream inFASTA; - openInputFile(fastaFileName, inFASTA); - numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); - inFASTA.close(); - - lines.push_back(new linePair(0, numFastaSeqs)); + vector outputNames; + + for (int s = 0; s < fastaFileNames.size(); s++) { + + m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine(); - driver(lines[0], newTaxonomyFile); + if (outputDir == "") { outputDir += hasPath(fastaFileNames[s]); } + string newTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "taxonomy"; + string tempTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "taxonomy.temp"; + string taxSummary = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "tax.summary"; - } - else{ - vector positions; - processIDS.resize(0); + outputNames.push_back(newTaxonomyFile); + outputNames.push_back(taxSummary); - ifstream inFASTA; - openInputFile(fastaFileName, inFASTA); + int start = time(NULL); + int numFastaSeqs = 0; + for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); - string input; - while(!inFASTA.eof()){ - input = getline(inFASTA); - if (input.length() != 0) { - if(input[0] == '>'){ int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); } +#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[newTaxonomyFile.length()]; + strcpy(outNewTax, newTaxonomyFile.c_str()); + + char outTempTax[tempTaxonomyFile.length()]; + strcpy(outTempTax, tempTaxonomyFile.c_str()); + + char inFileName[fastaFileNames[s].length()]; + strcpy(inFileName, fastaFileNames[s].c_str()); + + MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer + MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax); + MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax); + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; } + + if(namefile != "") { MPIReadNamesFile(namefileNames[s]); } + + if (pid == 0) { //you are the root process + + MPIPos = setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs + + //send file positions to all processes + MPI_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs + MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + int startIndex = 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_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs + MPIPos.resize(numFastaSeqs+1); + MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + int startIndex = 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); + +#else + + //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(); } - inFASTA.close(); - - numFastaSeqs = positions.size(); - - int numSeqsPerProcessor = numFastaSeqs / processors; - - for (int i = 0; i < processors; i++) { - int startPos = positions[ i * numSeqsPerProcessor ]; - if(i == processors - 1){ - numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + if(processors == 1){ + ifstream inFASTA; + openInputFile(fastaFileNames[s], inFASTA); + numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); + inFASTA.close(); + + lines.push_back(new linePair(0, numFastaSeqs)); + + driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); + } + else{ + vector positions; + processIDS.resize(0); + + ifstream inFASTA; + openInputFile(fastaFileNames[s], inFASTA); + + string input; + while(!inFASTA.eof()){ + input = getline(inFASTA); + if (input.length() != 0) { + if(input[0] == '>'){ int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); } + } + } + inFASTA.close(); + + numFastaSeqs = positions.size(); + + int numSeqsPerProcessor = numFastaSeqs / processors; + + for (int i = 0; i < processors; i++) { + int startPos = positions[ i * numSeqsPerProcessor ]; + if(i == processors - 1){ + numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; + } + lines.push_back(new linePair(startPos, numSeqsPerProcessor)); + } + createProcesses(newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); + + rename((newTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), newTaxonomyFile.c_str()); + rename((tempTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), tempTaxonomyFile.c_str()); + + for(int i=1;i(inFASTA),istreambuf_iterator(), '>'); + inFASTA.close(); - for(int i=1;i(inFASTA),istreambuf_iterator(), '>'); - inFASTA.close(); - - lines.push_back(new linePair(0, numFastaSeqs)); - - driver(lines[0], newTaxonomyFile); + driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); + #endif #endif delete classify; - //make taxonomy tree from new taxonomy file - ifstream inTaxonomy; - openInputFile(newTaxonomyFile, inTaxonomy); + #ifdef USE_MPI + if (pid == 0) { //this part does not need to be paralellized + #endif + + //make taxonomy tree from new taxonomy file + PhyloTree taxaBrowser; + + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; } + + ifstream in; + openInputFile(tempTaxonomyFile, in); - string accession, taxaList; - PhyloTree taxaBrowser; + //read in users taxonomy file and add sequences to tree + string name, taxon; + + while(!in.eof()){ + in >> name >> taxon; gobble(in); + + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } remove(tempTaxonomyFile.c_str()); delete classify; return 0; } + + if (namefile != "") { + itNames = nameMap.find(name); - //read in users taxonomy file and add sequences to tree - while(!inTaxonomy.eof()){ - inTaxonomy >> accession >> taxaList; + if (itNames == nameMap.end()) { + 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 + } + } + }else { taxaBrowser.addSeqToTree(name, taxon); } //add it once + } + in.close(); + + taxaBrowser.assignHeirarchyIDs(0); + + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } remove(tempTaxonomyFile.c_str()); delete classify; return 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; } + - taxaBrowser.addSeqToTree(accession, taxaList); + //print summary file + ofstream outTaxTree; + openOutputFile(taxSummary, outTaxTree); + taxaBrowser.print(outTaxTree); + outTaxTree.close(); - gobble(inTaxonomy); + //output taxonomy with the unclassified bins added + ifstream inTax; + openInputFile(newTaxonomyFile, inTax); + + ofstream outTax; + string unclass = newTaxonomyFile + ".unclass.temp"; + openOutputFile(unclass, outTax); + + //get maxLevel from phylotree so you know how many 'unclassified's to add + int maxLevel = taxaBrowser.getMaxLevel(); + + //read taxfile - this reading and rewriting is done to preserve the confidence scores. + 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); + + outTax << name << '\t' << newTax << endl; + } + inTax.close(); + outTax.close(); + + remove(newTaxonomyFile.c_str()); + rename(unclass.c_str(), newTaxonomyFile.c_str()); + + #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(); + + + m->mothurOutEndLine(); + m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine(); } - inTaxonomy.close(); - - taxaBrowser.assignHeirarchyIDs(0); - ofstream outTaxTree; - openOutputFile(taxSummary, outTaxTree); + return 0; + } + catch(exception& e) { + m->errorOut(e, "ClassifySeqsCommand", "execute"); + exit(1); + } +} + +/**************************************************************************************************/ +string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) { + try{ + string newTax, taxon; + int level = 0; - taxaBrowser.print(outTaxTree); + //keep what you have counting the levels + while (tax.find_first_of(';') != -1) { + //get taxon + taxon = tax.substr(0,tax.find_first_of(';'))+';'; + tax = tax.substr(tax.find_first_of(';')+1, tax.length()); + newTax += taxon; + level++; + } - mothurOutEndLine(); - mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); - mothurOutEndLine(); - mothurOutEndLine(); + //add "unclassified" until you reach maxLevel + while (level < maxlevel) { + newTax += "unclassified;"; + level++; + } - return 0; + return newTax; } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "execute"); + m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds"); exit(1); } } + /**************************************************************************************************/ -void ClassifySeqsCommand::createProcesses(string taxFileName) { +void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string filename) { try { #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) int process = 0; @@ -278,9 +619,9 @@ void ClassifySeqsCommand::createProcesses(string taxFileName) { 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], taxFileName + toString(getpid()) + ".temp"); + 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 @@ -291,7 +632,7 @@ void ClassifySeqsCommand::createProcesses(string taxFileName) { #endif } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "createProcesses"); + m->errorOut(e, "ClassifySeqsCommand", "createProcesses"); exit(1); } } @@ -314,47 +655,174 @@ void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) { output.close(); } catch(exception& e) { - errorOut(e, "ClassifySeqsCommand", "appendTaxFiles"); + m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles"); exit(1); } } //********************************************************************************************************************** -int ClassifySeqsCommand::driver(linePair* line, string taxFName){ +int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFName, string filename){ try { ofstream outTax; openOutputFile(taxFName, outTax); + + ofstream outTaxSimple; + openOutputFile(tempTFName, outTaxSimple); ifstream inFASTA; - openInputFile(fastaFileName, inFASTA); + openInputFile(filename, inFASTA); inFASTA.seekg(line->start); string taxonomy; for(int i=0;inumSeqs;i++){ + if (m->control_pressed) { return 0; } Sequence* candidateSeq = new Sequence(inFASTA); + + if (candidateSeq->getName() != "") { + taxonomy = classify->getTaxonomy(candidateSeq); + + if (m->control_pressed) { delete candidateSeq; return 0; } - taxonomy = classify->getTaxonomy(candidateSeq); + if ((taxonomy != "bad seq") && (taxonomy != "")) { + //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; + }else{ m->mothurOut("Sequence: " + candidateSeq->getName() + " is bad."); m->mothurOutEndLine(); } + } + delete candidateSeq; - if (taxonomy != "bad seq") { - outTax << candidateSeq->getName() << '\t' << taxonomy << endl; + if((i+1) % 100 == 0){ + m->mothurOut("Classifying sequence " + toString(i+1)); m->mothurOutEndLine(); } - - delete candidateSeq; } inFASTA.close(); outTax.close(); + outTaxSimple.close(); + + return 1; + } + catch(exception& e) { + 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[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); + + Sequence* candidateSeq = new Sequence(iss); + + if (candidateSeq->getName() != "") { + taxonomy = classify->getTaxonomy(candidateSeq); + + if ((taxonomy != "bad seq") && (taxonomy != "")) { + //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[length]; + strcpy(buf2, outputString.c_str()); + + MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew); + + outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n"; + length = outputString.length(); + char buf[length]; + strcpy(buf, outputString.c_str()); + + MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp); + }else{ cout << "Sequence: " << candidateSeq->getName() << " is bad." << endl; } + } + 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) { - errorOut(e, "ClassifySeqsCommand", "driver"); + 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[nameFilename.length()]; + 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); + + char buffer[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); + + string firstCol, secondCol; + while(!iss.eof()) { + iss >> firstCol >> secondCol; gobble(iss); + nameMap[firstCol] = getNumNames(secondCol); //ex. seq1 seq1,seq3,seq5 -> seq1 = 3. + } + + MPI_File_close(&inMPI); + + return 1; + } + catch(exception& e) { + m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile"); + exit(1); + } +} +#endif /**************************************************************************************************/