2 * classifyseqscommand.cpp
5 * Created by westcott on 11/2/09.
6 * Copyright 2009 Schloss Lab. All rights reserved.
10 #include "classifyseqscommand.h"
11 #include "sequence.hpp"
13 #include "phylotree.h"
14 #include "phylosummary.h"
17 //**********************************************************************************************************************
19 ClassifySeqsCommand::ClassifySeqsCommand(string option) {
23 //allow user to run help
24 if(option == "help") { help(); abort = true; }
28 //valid paramters for this command
29 string AlignArray[] = {"template","fasta","name","group","search","ksize","method","processors","taxonomy","match","mismatch","gapopen","gapextend","numwanted","cutoff","probs","iters", "outputdir","inputdir"};
30 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
32 OptionParser parser(option);
33 map<string, string> parameters = parser.getParameters();
35 ValidParameters validParameter;
36 map<string, string>::iterator it;
38 //check to make sure all parameters are valid for command
39 for (it = parameters.begin(); it != parameters.end(); it++) {
40 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
43 //if the user changes the output directory command factory will send this info to us in the output parameter
44 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
46 //if the user changes the input directory command factory will send this info to us in the output parameter
47 string inputDir = validParameter.validFile(parameters, "inputdir", false);
48 if (inputDir == "not found"){ inputDir = ""; }
51 it = parameters.find("template");
52 //user has given a template file
53 if(it != parameters.end()){
54 path = hasPath(it->second);
55 //if the user has not given a path then, add inputdir. else leave path alone.
56 if (path == "") { parameters["template"] = inputDir + it->second; }
59 it = parameters.find("taxonomy");
60 //user has given a template file
61 if(it != parameters.end()){
62 path = hasPath(it->second);
63 //if the user has not given a path then, add inputdir. else leave path alone.
64 if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
67 it = parameters.find("group");
68 //user has given a template file
69 if(it != parameters.end()){
70 path = hasPath(it->second);
71 //if the user has not given a path then, add inputdir. else leave path alone.
72 if (path == "") { parameters["group"] = inputDir + it->second; }
76 //check for required parameters
77 templateFileName = validParameter.validFile(parameters, "template", true);
78 if (templateFileName == "not found") {
79 m->mothurOut("template is a required parameter for the classify.seqs command.");
80 m->mothurOutEndLine();
83 else if (templateFileName == "not open") { abort = true; }
85 groupfile = validParameter.validFile(parameters, "group", true);
86 if (groupfile == "not open") { abort = true; }
87 else if (groupfile == "not found") { groupfile = ""; }
89 fastaFileName = validParameter.validFile(parameters, "fasta", false);
90 if (fastaFileName == "not found") { m->mothurOut("fasta is a required parameter for the classify.seqs command."); m->mothurOutEndLine(); abort = true; }
92 splitAtDash(fastaFileName, fastaFileNames);
94 //go through files and make sure they are good, if not, then disregard them
95 for (int i = 0; i < fastaFileNames.size(); i++) {
97 string path = hasPath(fastaFileNames[i]);
98 //if the user has not given a path then, add inputdir. else leave path alone.
99 if (path == "") { fastaFileNames[i] = inputDir + fastaFileNames[i]; }
106 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
107 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
113 ableToOpen = openInputFile(fastaFileNames[i], in);
117 for (int j = 1; j < processors; j++) {
118 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
122 MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
127 if (ableToOpen == 1) {
128 m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
129 //erase from file list
130 fastaFileNames.erase(fastaFileNames.begin()+i);
136 //make sure there is at least one valid file left
137 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
141 taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
142 if (taxonomyFileName == "not found") {
143 m->mothurOut("taxonomy is a required parameter for the classify.seqs command.");
144 m->mothurOutEndLine();
147 else if (taxonomyFileName == "not open") { abort = true; }
150 namefile = validParameter.validFile(parameters, "name", false);
151 if (namefile == "not found") { namefile = ""; }
154 splitAtDash(namefile, namefileNames);
156 //go through files and make sure they are good, if not, then disregard them
157 for (int i = 0; i < namefileNames.size(); i++) {
158 if (inputDir != "") {
159 string path = hasPath(namefileNames[i]);
160 //if the user has not given a path then, add inputdir. else leave path alone.
161 if (path == "") { namefileNames[i] = inputDir + namefileNames[i]; }
167 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
168 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
174 ableToOpen = openInputFile(namefileNames[i], in);
178 for (int j = 1; j < processors; j++) {
179 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
183 MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
187 if (ableToOpen == 1) { m->mothurOut("Unable to match name file with fasta file."); m->mothurOutEndLine(); abort = true; }
192 if (namefile != "") {
193 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(); }
196 //check for optional parameter and set defaults
197 // ...at some point should added some additional type checking...
199 temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; }
200 convert(temp, kmerSize);
202 temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
203 convert(temp, processors);
205 search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }
207 method = validParameter.validFile(parameters, "method", false); if (method == "not found"){ method = "bayesian"; }
209 temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }
210 convert(temp, match);
212 temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }
213 convert(temp, misMatch);
215 temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; }
216 convert(temp, gapOpen);
218 temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; }
219 convert(temp, gapExtend);
221 temp = validParameter.validFile(parameters, "numwanted", false); if (temp == "not found"){ temp = "10"; }
222 convert(temp, numWanted);
224 temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found"){ temp = "0"; }
225 convert(temp, cutoff);
227 temp = validParameter.validFile(parameters, "probs", false); if (temp == "not found"){ temp = "true"; }
228 probs = isTrue(temp);
230 temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "100"; }
231 convert(temp, iters);
235 if ((method == "bayesian") && (search != "kmer")) {
236 m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine();
242 catch(exception& e) {
243 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
248 //**********************************************************************************************************************
250 ClassifySeqsCommand::~ClassifySeqsCommand(){
252 if (abort == false) {
253 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
257 //**********************************************************************************************************************
259 void ClassifySeqsCommand::help(){
261 m->mothurOut("The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n");
262 m->mothurOut("The classify.seqs command parameters are template, fasta, name, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n");
263 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");
264 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");
265 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");
266 m->mothurOut("The group parameter allows you add a group file so you can have the summary totals broken up by group.\n");
267 m->mothurOut("The method parameter allows you to specify classification method to use. Your options are: bayesian and knn. The default is bayesian.\n");
268 m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n");
269 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
271 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
273 m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");
274 m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n");
275 m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n");
276 m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n");
277 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");
278 m->mothurOut("The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy. The default is 0.\n");
279 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");
280 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");
281 m->mothurOut("The classify.seqs command should be in the following format: \n");
282 m->mothurOut("classify.seqs(template=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n");
283 m->mothurOut("Example classify.seqs(fasta=amazon.fasta, template=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n");
284 m->mothurOut("The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n");
285 m->mothurOut("The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n");
286 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
288 catch(exception& e) {
289 m->errorOut(e, "ClassifySeqsCommand", "help");
295 //**********************************************************************************************************************
297 int ClassifySeqsCommand::execute(){
299 if (abort == true) { return 0; }
301 if(method == "bayesian"){ classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters); }
302 else if(method == "knn"){ classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted); }
304 m->mothurOut(search + " is not a valid method option. I will run the command using bayesian.");
305 m->mothurOutEndLine();
306 classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters);
309 if (m->control_pressed) { delete classify; return 0; }
311 vector<string> outputNames;
313 for (int s = 0; s < fastaFileNames.size(); s++) {
315 m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
317 if (outputDir == "") { outputDir += hasPath(fastaFileNames[s]); }
318 string newTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "taxonomy";
319 string tempTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "taxonomy.temp";
320 string taxSummary = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "tax.summary";
322 outputNames.push_back(newTaxonomyFile);
323 outputNames.push_back(taxSummary);
325 int start = time(NULL);
326 int numFastaSeqs = 0;
327 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
330 int pid, end, numSeqsPerProcessor;
335 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
336 MPI_Comm_size(MPI_COMM_WORLD, &processors);
339 MPI_File outMPINewTax;
340 MPI_File outMPITempTax;
342 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
343 int inMode=MPI_MODE_RDONLY;
345 //char* outNewTax = new char[newTaxonomyFile.length()];
346 //memcpy(outNewTax, newTaxonomyFile.c_str(), newTaxonomyFile.length());
348 char outNewTax[1024];
349 strcpy(outNewTax, newTaxonomyFile.c_str());
351 //char* outTempTax = new char[tempTaxonomyFile.length()];
352 //memcpy(outTempTax, tempTaxonomyFile.c_str(), tempTaxonomyFile.length());
354 char outTempTax[1024];
355 strcpy(outTempTax, tempTaxonomyFile.c_str());
357 //char* inFileName = new char[fastaFileNames[s].length()];
358 //memcpy(inFileName, fastaFileNames[s].c_str(), fastaFileNames[s].length());
360 char inFileName[1024];
361 strcpy(inFileName, fastaFileNames[s].c_str());
363 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
364 MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax);
365 MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax);
371 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; }
373 if(namefile != "") { MPIReadNamesFile(namefileNames[s]); }
375 if (pid == 0) { //you are the root process
377 MPIPos = setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
379 //send file positions to all processes
380 MPI_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
381 MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
383 //figure out how many sequences you have to align
384 numSeqsPerProcessor = numFastaSeqs / processors;
385 int startIndex = pid * numSeqsPerProcessor;
386 if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
390 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
392 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; }
394 for (int i = 1; i < processors; i++) {
396 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
398 }else{ //you are a child process
399 MPI_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
400 MPIPos.resize(numFastaSeqs+1);
401 MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
403 //figure out how many sequences you have to align
404 numSeqsPerProcessor = numFastaSeqs / processors;
405 int startIndex = pid * numSeqsPerProcessor;
406 if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
410 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
412 if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPITempTax); delete classify; return 0; }
415 MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
419 MPI_File_close(&inMPI);
420 MPI_File_close(&outMPINewTax);
421 MPI_File_close(&outMPITempTax);
426 nameMap.clear(); //remove old names
429 openInputFile(namefileNames[s], inNames);
431 string firstCol, secondCol;
432 while(!inNames.eof()) {
433 inNames >> firstCol >> secondCol; gobble(inNames);
434 nameMap[firstCol] = getNumNames(secondCol); //ex. seq1 seq1,seq3,seq5 -> seq1 = 3.
439 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
442 openInputFile(fastaFileNames[s], inFASTA);
443 numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
446 lines.push_back(new linePair(0, numFastaSeqs));
448 driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
451 vector<int> positions;
452 processIDS.resize(0);
455 openInputFile(fastaFileNames[s], inFASTA);
458 while(!inFASTA.eof()){
459 input = getline(inFASTA);
460 if (input.length() != 0) {
461 if(input[0] == '>'){ int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
466 numFastaSeqs = positions.size();
468 int numSeqsPerProcessor = numFastaSeqs / processors;
470 for (int i = 0; i < processors; i++) {
471 int startPos = positions[ i * numSeqsPerProcessor ];
472 if(i == processors - 1){
473 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
475 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
477 createProcesses(newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
479 rename((newTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), newTaxonomyFile.c_str());
480 rename((tempTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), tempTaxonomyFile.c_str());
482 for(int i=1;i<processors;i++){
483 appendTaxFiles((newTaxonomyFile + toString(processIDS[i]) + ".temp"), newTaxonomyFile);
484 appendTaxFiles((tempTaxonomyFile + toString(processIDS[i]) + ".temp"), tempTaxonomyFile);
485 remove((newTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
486 remove((tempTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
492 openInputFile(fastaFileNames[s], inFASTA);
493 numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
496 lines.push_back(new linePair(0, numFastaSeqs));
498 driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
503 if (pid == 0) { //this part does not need to be paralellized
506 m->mothurOutEndLine();
507 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
510 PhyloSummary taxaSum(taxonomyFileName, groupfile);
512 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; }
514 if (namefile == "") { taxaSum.summarize(tempTaxonomyFile); }
517 openInputFile(tempTaxonomyFile, in);
519 //read in users taxonomy file and add sequences to tree
522 in >> name >> taxon; gobble(in);
524 itNames = nameMap.find(name);
526 if (itNames == nameMap.end()) {
527 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
529 for (int i = 0; i < itNames->second; i++) {
530 taxaSum.addSeqToTree(name+toString(i), taxon); //add it as many times as there are identical seqs
536 remove(tempTaxonomyFile.c_str());
538 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete classify; return 0; }
542 openOutputFile(taxSummary, outTaxTree);
543 taxaSum.print(outTaxTree);
546 //output taxonomy with the unclassified bins added
548 openInputFile(newTaxonomyFile, inTax);
551 string unclass = newTaxonomyFile + ".unclass.temp";
552 openOutputFile(unclass, outTax);
554 //get maxLevel from phylotree so you know how many 'unclassified's to add
555 int maxLevel = taxaSum.getMaxLevel();
557 //read taxfile - this reading and rewriting is done to preserve the confidence scores.
559 while (!inTax.eof()) {
560 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; }
562 inTax >> name >> taxon; gobble(inTax);
564 string newTax = addUnclassifieds(taxon, maxLevel);
566 outTax << name << '\t' << newTax << endl;
571 remove(newTaxonomyFile.c_str());
572 rename(unclass.c_str(), newTaxonomyFile.c_str());
574 m->mothurOutEndLine();
575 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
581 m->mothurOutEndLine();
582 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
583 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
584 m->mothurOutEndLine();
591 catch(exception& e) {
592 m->errorOut(e, "ClassifySeqsCommand", "execute");
597 /**************************************************************************************************/
598 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
600 string newTax, taxon;
603 //keep what you have counting the levels
604 while (tax.find_first_of(';') != -1) {
606 taxon = tax.substr(0,tax.find_first_of(';'))+';';
607 tax = tax.substr(tax.find_first_of(';')+1, tax.length());
612 //add "unclassified" until you reach maxLevel
613 while (level < maxlevel) {
614 newTax += "unclassified;";
620 catch(exception& e) {
621 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
626 /**************************************************************************************************/
628 void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string filename) {
630 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
632 // processIDS.resize(0);
634 //loop through and create all the processes you want
635 while (process != processors) {
639 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
642 driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", filename);
644 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
647 //force parent to wait until all the processes are done
648 for (int i=0;i<processors;i++) {
649 int temp = processIDS[i];
654 catch(exception& e) {
655 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
659 /**************************************************************************************************/
661 void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) {
666 openOutputFileAppend(filename, output);
667 openInputFile(temp, input);
669 while(char c = input.get()){
670 if(input.eof()) { break; }
671 else { output << c; }
677 catch(exception& e) {
678 m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles");
683 //**********************************************************************************************************************
685 int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFName, string filename){
688 openOutputFile(taxFName, outTax);
690 ofstream outTaxSimple;
691 openOutputFile(tempTFName, outTaxSimple);
694 openInputFile(filename, inFASTA);
696 inFASTA.seekg(line->start);
700 for(int i=0;i<line->numSeqs;i++){
701 if (m->control_pressed) { return 0; }
703 Sequence* candidateSeq = new Sequence(inFASTA);
705 if (candidateSeq->getName() != "") {
706 taxonomy = classify->getTaxonomy(candidateSeq);
708 if (m->control_pressed) { delete candidateSeq; return 0; }
710 if (taxonomy != "bad seq") {
711 //output confidence scores or not
713 outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
715 outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
718 outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
723 if((i+1) % 100 == 0){
724 m->mothurOut("Classifying sequence " + toString(i+1)); m->mothurOutEndLine();
730 outTaxSimple.close();
734 catch(exception& e) {
735 m->errorOut(e, "ClassifySeqsCommand", "driver");
739 //**********************************************************************************************************************
741 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, vector<long>& MPIPos){
743 MPI_Status statusNew;
744 MPI_Status statusTemp;
748 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
753 for(int i=0;i<num;i++){
755 if (m->control_pressed) { return 0; }
758 int length = MPIPos[start+i+1] - MPIPos[start+i];
759 char* buf4 = new char[length];
760 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
762 string tempBuf = buf4;
763 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
764 istringstream iss (tempBuf,istringstream::in);
767 Sequence* candidateSeq = new Sequence(iss);
769 if (candidateSeq->getName() != "") {
770 taxonomy = classify->getTaxonomy(candidateSeq);
772 if (taxonomy != "bad seq") {
773 //output confidence scores or not
775 outputString = candidateSeq->getName() + "\t" + taxonomy + "\n";
777 outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
780 int length = outputString.length();
781 char* buf2 = new char[length];
782 memcpy(buf2, outputString.c_str(), length);
784 MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
787 outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
788 length = outputString.length();
789 char* buf = new char[length];
790 memcpy(buf, outputString.c_str(), length);
792 MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
798 if((i+1) % 100 == 0){ cout << "Classifying sequence " << (i+1) << endl; }
801 if(num % 100 != 0){ cout << "Classifying sequence " << (num) << endl; }
806 catch(exception& e) {
807 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
812 //**********************************************************************************************************************
813 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
816 nameMap.clear(); //remove old names
822 //char* inFileName = new char[nameFilename.length()];
823 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
825 char inFileName[1024];
826 strcpy(inFileName, nameFilename.c_str());
828 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
829 MPI_File_get_size(inMPI, &size);
832 char* buffer = new char[size];
833 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
835 string tempBuf = buffer;
836 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size); }
837 istringstream iss (tempBuf,istringstream::in);
840 string firstCol, secondCol;
842 iss >> firstCol >> secondCol; gobble(iss);
843 nameMap[firstCol] = getNumNames(secondCol); //ex. seq1 seq1,seq3,seq5 -> seq1 = 3.
846 MPI_File_close(&inMPI);
850 catch(exception& e) {
851 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
856 /**************************************************************************************************/