X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=classifyseqscommand.cpp;h=0538bff07d99ca1266d488a5844fcde49afaf3bc;hb=13193e6688c91b6a25e39d357caa7f4b4bf5de5f;hp=ba854e98b40403c8d963fb65c9c4bd6178950947;hpb=aa9238c0a9e6e7aa0ed8b8b606b08ad4fd7dcfe3;p=mothur.git diff --git a/classifyseqscommand.cpp b/classifyseqscommand.cpp index ba854e9..0538bff 100644 --- a/classifyseqscommand.cpp +++ b/classifyseqscommand.cpp @@ -11,6 +11,7 @@ #include "sequence.hpp" #include "bayesian.h" #include "phylotree.h" +#include "phylosummary.h" #include "knn.h" //********************************************************************************************************************** @@ -25,7 +26,7 @@ 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", "outputdir","inputdir"}; + 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); @@ -62,6 +63,14 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { //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 @@ -73,6 +82,10 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { } else if (templateFileName == "not open") { abort = true; } + groupfile = validParameter.validFile(parameters, "group", true); + if (groupfile == "not open") { abort = true; } + else if (groupfile == "not found") { groupfile = ""; } + 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 { @@ -87,15 +100,37 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { } 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--; } - in.close(); + } //make sure there is at least one valid file left @@ -125,12 +160,32 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option) { //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) { m->mothurOut("Unable to match name file with fasta file."); m->mothurOutEndLine(); abort = true; } 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; } + } } @@ -208,9 +263,13 @@ void ClassifySeqsCommand::help(){ 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"); @@ -253,21 +312,6 @@ int ClassifySeqsCommand::execute(){ 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(); if (outputDir == "") { outputDir += hasPath(fastaFileNames[s]); } @@ -282,7 +326,117 @@ int ClassifySeqsCommand::execute(){ 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(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; + 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_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; + 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); + +#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(); + } + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) if(processors == 1){ ifstream inFASTA; openInputFile(fastaFileNames[s], inFASTA); @@ -333,7 +487,7 @@ int ClassifySeqsCommand::execute(){ } } -#else + #else ifstream inFASTA; openInputFile(fastaFileNames[s], inFASTA); numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); @@ -342,51 +496,51 @@ int ClassifySeqsCommand::execute(){ lines.push_back(new linePair(0, numFastaSeqs)); driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]); -#endif - //make taxonomy tree from new taxonomy file - PhyloTree taxaBrowser; + #endif +#endif + + #ifdef USE_MPI + if (pid == 0) { //this part does not need to be paralellized + #endif + + m->mothurOutEndLine(); + m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine(); + start = time(NULL); + + PhyloSummary taxaSum(taxonomyFileName, groupfile); 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); - - //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 == "") { taxaSum.summarize(tempTaxonomyFile); } + else { + ifstream in; + openInputFile(tempTaxonomyFile, in); - if (namefile != "") { + //read in users taxonomy file and add sequences to tree + string name, taxon; + while(!in.eof()){ + in >> name >> taxon; gobble(in); + itNames = nameMap.find(name); 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 + taxaSum.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(); } - 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; } - //print summary file ofstream outTaxTree; openOutputFile(taxSummary, outTaxTree); - taxaBrowser.print(outTaxTree); + taxaSum.print(outTaxTree); outTaxTree.close(); //output taxonomy with the unclassified bins added @@ -398,9 +552,10 @@ 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 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; } @@ -416,14 +571,18 @@ int ClassifySeqsCommand::execute(){ remove(newTaxonomyFile.c_str()); rename(unclass.c_str(), newTaxonomyFile.c_str()); + 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(); - - m->mothurOutEndLine(); - m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine(); } delete classify; @@ -577,5 +736,121 @@ int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFNa 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); + 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 /**************************************************************************************************/