2 * classifyseqscommand.cpp
5 * Created by westcott on 11/2/09.
6 * Copyright 2009 Schloss Lab. All rights reserved.
10 #include "classifyseqscommand.h"
14 //**********************************************************************************************************************
15 vector<string> ClassifySeqsCommand::setParameters(){
17 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptaxonomy);
18 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
19 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
20 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
21 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
22 CommandParameter psearch("search", "Multiple", "kmer-blast-suffix-distance", "kmer", "", "", "",false,false); parameters.push_back(psearch);
23 CommandParameter pksize("ksize", "Number", "", "8", "", "", "",false,false); parameters.push_back(pksize);
24 CommandParameter pmethod("method", "Multiple", "bayesian-knn", "bayesian", "", "", "",false,false); parameters.push_back(pmethod);
25 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
26 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pmatch);
27 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pmismatch);
28 CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "",false,false); parameters.push_back(pgapopen);
29 CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend);
30 //CommandParameter pflip("flip", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pflip);
31 CommandParameter pcutoff("cutoff", "Number", "", "0", "", "", "",false,true); parameters.push_back(pcutoff);
32 CommandParameter pprobs("probs", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pprobs);
33 CommandParameter piters("iters", "Number", "", "100", "", "", "",false,true); parameters.push_back(piters);
34 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
35 CommandParameter pnumwanted("numwanted", "Number", "", "10", "", "", "",false,true); parameters.push_back(pnumwanted);
36 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
37 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
39 vector<string> myArray;
40 for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
44 m->errorOut(e, "ClassifySeqsCommand", "setParameters");
48 //**********************************************************************************************************************
49 string ClassifySeqsCommand::getHelpString(){
51 string helpString = "";
52 helpString += "The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n";
53 helpString += "The classify.seqs command parameters are reference, fasta, name, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n";
54 helpString += "The reference, 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";
55 helpString += "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";
56 helpString += "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";
57 helpString += "The group parameter allows you add a group file so you can have the summary totals broken up by group.\n";
58 helpString += "The method parameter allows you to specify classification method to use. Your options are: bayesian and knn. The default is bayesian.\n";
59 helpString += "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n";
60 helpString += "The processors parameter allows you to specify the number of processors to use. The default is 1.\n";
62 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
64 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
65 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n";
66 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n";
67 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n";
68 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n";
69 helpString += "The numwanted parameter allows you to specify the number of sequence matches you want with the knn method. The default is 10.\n";
70 helpString += "The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy. The default is 0.\n";
71 helpString += "The probs parameter shuts off the bootstrapping results for the bayesian method. The default is true, meaning you want the bootstrapping to be shown.\n";
72 helpString += "The iters parameter allows you to specify how many iterations to do when calculating the bootstrap confidence score for your taxonomy with the bayesian method. The default is 100.\n";
73 //helpString += "The flip parameter allows you shut off mothur's The default is T.\n";
74 helpString += "The classify.seqs command should be in the following format: \n";
75 helpString += "classify.seqs(reference=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n";
76 helpString += "Example classify.seqs(fasta=amazon.fasta, reference=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n";
77 helpString += "The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n";
78 helpString += "The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n";
79 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
83 m->errorOut(e, "ClassifySeqsCommand", "getHelpString");
87 //**********************************************************************************************************************
88 ClassifySeqsCommand::ClassifySeqsCommand(){
90 abort = true; calledHelp = true;
92 vector<string> tempOutNames;
93 outputTypes["taxonomy"] = tempOutNames;
94 outputTypes["accnos"] = tempOutNames;
95 outputTypes["taxsummary"] = tempOutNames;
96 outputTypes["matchdist"] = tempOutNames;
99 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
103 //**********************************************************************************************************************
104 ClassifySeqsCommand::ClassifySeqsCommand(string option) {
106 abort = false; calledHelp = false;
107 rdb = ReferenceDB::getInstance();
109 //allow user to run help
110 if(option == "help") { help(); abort = true; calledHelp = true; }
111 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
114 vector<string> myArray = setParameters();
116 OptionParser parser(option);
117 map<string, string> parameters = parser.getParameters();
119 ValidParameters validParameter("classify.seqs");
120 map<string, string>::iterator it;
122 //check to make sure all parameters are valid for command
123 for (it = parameters.begin(); it != parameters.end(); it++) {
124 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
127 //initialize outputTypes
128 vector<string> tempOutNames;
129 outputTypes["taxonomy"] = tempOutNames;
130 outputTypes["taxsummary"] = tempOutNames;
131 outputTypes["matchdist"] = tempOutNames;
132 outputTypes["accnos"] = tempOutNames;
134 //if the user changes the output directory command factory will send this info to us in the output parameter
135 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
137 //if the user changes the input directory command factory will send this info to us in the output parameter
138 string inputDir = validParameter.validFile(parameters, "inputdir", false);
139 if (inputDir == "not found"){ inputDir = ""; }
142 it = parameters.find("reference");
143 //user has given a template file
144 if(it != parameters.end()){
145 path = m->hasPath(it->second);
146 //if the user has not given a path then, add inputdir. else leave path alone.
147 if (path == "") { parameters["reference"] = inputDir + it->second; }
150 it = parameters.find("taxonomy");
151 //user has given a template file
152 if(it != parameters.end()){
153 path = m->hasPath(it->second);
154 //if the user has not given a path then, add inputdir. else leave path alone.
155 if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
158 it = parameters.find("group");
159 //user has given a template file
160 if(it != parameters.end()){
161 path = m->hasPath(it->second);
162 //if the user has not given a path then, add inputdir. else leave path alone.
163 if (path == "") { parameters["group"] = inputDir + it->second; }
167 fastaFileName = validParameter.validFile(parameters, "fasta", false);
168 if (fastaFileName == "not found") {
169 //if there is a current fasta file, use it
170 string filename = m->getFastaFile();
171 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
172 else { m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
175 m->splitAtDash(fastaFileName, fastaFileNames);
177 //go through files and make sure they are good, if not, then disregard them
178 for (int i = 0; i < fastaFileNames.size(); i++) {
181 if (fastaFileNames[i] == "current") {
182 fastaFileNames[i] = m->getFastaFile();
183 if (fastaFileNames[i] != "") { m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
185 m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
186 //erase from file list
187 fastaFileNames.erase(fastaFileNames.begin()+i);
194 if (inputDir != "") {
195 string path = m->hasPath(fastaFileNames[i]);
196 //if the user has not given a path then, add inputdir. else leave path alone.
197 if (path == "") { fastaFileNames[i] = inputDir + fastaFileNames[i]; }
203 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
205 //if you can't open it, try default location
206 if (ableToOpen == 1) {
207 if (m->getDefaultPath() != "") { //default path is set
208 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
209 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
211 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
213 fastaFileNames[i] = tryPath;
217 if (ableToOpen == 1) {
218 if (m->getOutputDir() != "") { //default path is set
219 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
220 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
222 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
224 fastaFileNames[i] = tryPath;
230 if (ableToOpen == 1) {
231 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
232 //erase from file list
233 fastaFileNames.erase(fastaFileNames.begin()+i);
236 m->setFastaFile(fastaFileNames[i]);
242 //make sure there is at least one valid file left
243 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
246 namefile = validParameter.validFile(parameters, "name", false);
247 if (namefile == "not found") { namefile = ""; }
250 m->splitAtDash(namefile, namefileNames);
252 //go through files and make sure they are good, if not, then disregard them
253 for (int i = 0; i < namefileNames.size(); i++) {
255 if (namefileNames[i] == "current") {
256 namefileNames[i] = m->getNameFile();
257 if (namefileNames[i] != "") { m->mothurOut("Using " + namefileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
259 m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
260 //erase from file list
261 namefileNames.erase(namefileNames.begin()+i);
268 if (inputDir != "") {
269 string path = m->hasPath(namefileNames[i]);
270 //if the user has not given a path then, add inputdir. else leave path alone.
271 if (path == "") { namefileNames[i] = inputDir + namefileNames[i]; }
276 ableToOpen = m->openInputFile(namefileNames[i], in, "noerror");
278 //if you can't open it, try default location
279 if (ableToOpen == 1) {
280 if (m->getDefaultPath() != "") { //default path is set
281 string tryPath = m->getDefaultPath() + m->getSimpleName(namefileNames[i]);
282 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
284 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
286 namefileNames[i] = tryPath;
290 if (ableToOpen == 1) {
291 if (m->getOutputDir() != "") { //default path is set
292 string tryPath = m->getOutputDir() + m->getSimpleName(namefileNames[i]);
293 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
295 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
297 namefileNames[i] = tryPath;
302 if (ableToOpen == 1) {
303 m->mothurOut("Unable to open " + namefileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); abort = true;
304 //erase from file list
305 namefileNames.erase(namefileNames.begin()+i);
308 m->setNameFile(namefileNames[i]);
314 if (namefile != "") {
315 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(); }
318 groupfile = validParameter.validFile(parameters, "group", false);
319 if (groupfile == "not found") { groupfile = ""; }
321 m->splitAtDash(groupfile, groupfileNames);
323 //go through files and make sure they are good, if not, then disregard them
324 for (int i = 0; i < groupfileNames.size(); i++) {
325 if (inputDir != "") {
326 string path = m->hasPath(groupfileNames[i]);
327 //if the user has not given a path then, add inputdir. else leave path alone.
328 if (path == "") { groupfileNames[i] = inputDir + groupfileNames[i]; }
333 ableToOpen = m->openInputFile(groupfileNames[i], in, "noerror");
335 //if you can't open it, try default location
336 if (ableToOpen == 1) {
337 if (m->getDefaultPath() != "") { //default path is set
338 string tryPath = m->getDefaultPath() + m->getSimpleName(groupfileNames[i]);
339 m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
341 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
343 groupfileNames[i] = tryPath;
347 if (ableToOpen == 1) {
348 if (m->getOutputDir() != "") { //default path is set
349 string tryPath = m->getOutputDir() + m->getSimpleName(groupfileNames[i]);
350 m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
352 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
354 groupfileNames[i] = tryPath;
360 if (ableToOpen == 1) {
361 m->mothurOut("Unable to open " + groupfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); groupfileNames[i] = "";
362 //erase from file list
363 groupfileNames.erase(groupfileNames.begin()+i);
366 m->setGroupFile(groupfileNames[i]);
371 if (groupfile != "") {
372 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(); }
374 for (int i = 0; i < fastaFileNames.size(); i++) { groupfileNames.push_back(""); }
377 //check for optional parameter and set defaults
378 // ...at some point should added some additional type checking...
380 temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; }
381 m->mothurConvert(temp, kmerSize);
383 temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
384 m->setProcessors(temp);
385 m->mothurConvert(temp, processors);
387 temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
388 save = m->isTrue(temp);
390 if (save) { //clear out old references
394 //this has to go after save so that if the user sets save=t and provides no reference we abort
395 templateFileName = validParameter.validFile(parameters, "reference", true);
396 if (templateFileName == "not found") {
397 //check for saved reference sequences
398 if (rdb->referenceSeqs.size() != 0) {
399 templateFileName = "saved";
401 m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the classify.seqs command.");
402 m->mothurOutEndLine();
405 }else if (templateFileName == "not open") { abort = true; }
406 else { if (save) { rdb->setSavedReference(templateFileName); } }
408 //this has to go after save so that if the user sets save=t and provides no reference we abort
409 taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
410 if (taxonomyFileName == "not found") {
411 //check for saved reference sequences
412 if (rdb->wordGenusProb.size() != 0) {
413 taxonomyFileName = "saved";
415 m->mothurOut("[ERROR]: You don't have any saved taxonomy information and the taxonomy parameter is a required for the classify.seqs command.");
416 m->mothurOutEndLine();
419 }else if (taxonomyFileName == "not open") { abort = true; }
420 else { if (save) { rdb->setSavedTaxonomy(taxonomyFileName); } }
422 search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }
424 method = validParameter.validFile(parameters, "method", false); if (method == "not found"){ method = "bayesian"; }
426 temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }
427 m->mothurConvert(temp, match);
429 temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }
430 m->mothurConvert(temp, misMatch);
432 temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; }
433 m->mothurConvert(temp, gapOpen);
435 temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; }
436 m->mothurConvert(temp, gapExtend);
438 temp = validParameter.validFile(parameters, "numwanted", false); if (temp == "not found"){ temp = "10"; }
439 m->mothurConvert(temp, numWanted);
441 temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found"){ temp = "0"; }
442 m->mothurConvert(temp, cutoff);
444 temp = validParameter.validFile(parameters, "probs", false); if (temp == "not found"){ temp = "true"; }
445 probs = m->isTrue(temp);
447 //temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "T"; }
448 //flip = m->isTrue(temp);
451 temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "100"; }
452 m->mothurConvert(temp, iters);
455 if ((method == "bayesian") && (search != "kmer")) {
456 m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine();
461 if (namefileNames.size() == 0){
462 if (fastaFileNames.size() != 0) {
463 vector<string> files; files.push_back(fastaFileNames[fastaFileNames.size()-1]);
464 parser.getNameFile(files);
471 catch(exception& e) {
472 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
477 //**********************************************************************************************************************
478 ClassifySeqsCommand::~ClassifySeqsCommand(){
479 if (abort == false) {
480 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
483 //**********************************************************************************************************************
485 int ClassifySeqsCommand::execute(){
487 if (abort == true) { if (calledHelp) { return 0; } return 2; }
489 if(method == "bayesian"){ classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip); }
490 else if(method == "knn"){ classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted, rand()); }
492 m->mothurOut(search + " is not a valid method option. I will run the command using bayesian.");
493 m->mothurOutEndLine();
494 classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip);
497 if (m->control_pressed) { delete classify; return 0; }
499 for (int s = 0; s < fastaFileNames.size(); s++) {
501 m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
503 string baseTName = taxonomyFileName;
504 if (taxonomyFileName == "saved") {baseTName = rdb->getSavedTaxonomy(); }
506 string RippedTaxName = m->getRootName(m->getSimpleName(baseTName));
507 RippedTaxName = m->getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1));
508 if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); }
509 if (RippedTaxName != "") { RippedTaxName += "."; }
511 if (outputDir == "") { outputDir += m->hasPath(fastaFileNames[s]); }
512 string newTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "taxonomy";
513 string newaccnosFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "flip.accnos";
514 string tempTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "taxonomy.temp";
515 string taxSummary = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "tax.summary";
517 if ((method == "knn") && (search == "distance")) {
518 string DistName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "match.dist";
519 classify->setDistName(DistName); outputNames.push_back(DistName); outputTypes["matchdist"].push_back(DistName);
522 outputNames.push_back(newTaxonomyFile); outputTypes["taxonomy"].push_back(newTaxonomyFile);
523 outputNames.push_back(newaccnosFile); outputTypes["accnos"].push_back(newaccnosFile);
524 outputNames.push_back(taxSummary); outputTypes["taxsummary"].push_back(taxSummary);
526 int start = time(NULL);
527 int numFastaSeqs = 0;
528 for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
531 int pid, numSeqsPerProcessor;
533 vector<unsigned long long> MPIPos;
536 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
537 MPI_Comm_size(MPI_COMM_WORLD, &processors);
540 MPI_File outMPINewTax;
541 MPI_File outMPITempTax;
544 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
545 int inMode=MPI_MODE_RDONLY;
547 char outNewTax[1024];
548 strcpy(outNewTax, newTaxonomyFile.c_str());
550 char outTempTax[1024];
551 strcpy(outTempTax, tempTaxonomyFile.c_str());
554 strcpy(outAcc, newaccnosFile.c_str());
556 char inFileName[1024];
557 strcpy(inFileName, fastaFileNames[s].c_str());
559 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
560 MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax);
561 MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax);
562 MPI_File_open(MPI_COMM_WORLD, outAcc, outMode, MPI_INFO_NULL, &outMPIAcc);
564 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); delete classify; return 0; }
566 if (pid == 0) { //you are the root process
568 MPIPos = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
570 //send file positions to all processes
571 for(int i = 1; i < processors; i++) {
572 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
573 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
576 //figure out how many sequences you have to align
577 numSeqsPerProcessor = numFastaSeqs / processors;
578 int startIndex = pid * numSeqsPerProcessor;
579 if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
583 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
585 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; }
587 for (int i = 1; i < processors; i++) {
589 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
591 }else{ //you are a child process
592 MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
593 MPIPos.resize(numFastaSeqs+1);
594 MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
596 //figure out how many sequences you have to align
597 numSeqsPerProcessor = numFastaSeqs / processors;
598 int startIndex = pid * numSeqsPerProcessor;
599 if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
603 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
605 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc); MPI_File_close(&outMPITempTax); delete classify; return 0; }
608 MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
612 MPI_File_close(&inMPI);
613 MPI_File_close(&outMPINewTax);
614 MPI_File_close(&outMPITempTax);
615 MPI_File_close(&outMPIAcc);
616 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
620 vector<unsigned long long> positions;
621 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
622 positions = m->divideFile(fastaFileNames[s], processors);
623 for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(new linePair(positions[i], positions[(i+1)])); }
625 if (processors == 1) {
626 lines.push_back(new linePair(0, 1000));
628 positions = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs);
629 if (positions.size() < processors) { processors = positions.size(); }
631 //figure out how many sequences you have to process
632 int numSeqsPerProcessor = numFastaSeqs / processors;
633 for (int i = 0; i < processors; i++) {
634 int startIndex = i * numSeqsPerProcessor;
635 if(i == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; }
636 lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
641 numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
643 numFastaSeqs = createProcesses(newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
647 if (!m->isBlank(newaccnosFile)) { m->mothurOutEndLine(); m->mothurOut("[WARNING]: mothur suspects some of your sequences may be reversed, please check " + newaccnosFile + " for the list of the sequences."); m->mothurOutEndLine(); }
649 m->mothurOutEndLine();
650 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
656 if (pid == 0) { //this part does not need to be paralellized
658 if(namefile != "") { m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush(); MPIReadNamesFile(namefileNames[s]); m->mothurOut(" Done."); m->mothurOutEndLine(); }
663 m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();
665 nameMap.clear(); //remove old names
668 m->openInputFile(namefileNames[s], inNames);
670 string firstCol, secondCol;
671 while(!inNames.eof()) {
672 inNames >> firstCol >> secondCol; m->gobble(inNames);
675 m->splitAtComma(secondCol, temp);
677 nameMap[firstCol] = temp;
681 m->mothurOut(" Done."); m->mothurOutEndLine();
686 if (groupfile != "") { group = groupfileNames[s]; }
688 PhyloSummary taxaSum(baseTName, group);
690 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; }
692 if (namefile == "") { taxaSum.summarize(tempTaxonomyFile); }
695 m->openInputFile(tempTaxonomyFile, in);
697 //read in users taxonomy file and add sequences to tree
701 in >> name >> taxon; m->gobble(in);
703 itNames = nameMap.find(name);
705 if (itNames == nameMap.end()) {
706 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
708 for (int i = 0; i < itNames->second.size(); i++) {
709 taxaSum.addSeqToTree(itNames->second[i], taxon); //add it as many times as there are identical seqs
711 itNames->second.clear();
712 nameMap.erase(itNames->first);
717 m->mothurRemove(tempTaxonomyFile);
719 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete classify; return 0; }
723 m->openOutputFile(taxSummary, outTaxTree);
724 taxaSum.print(outTaxTree);
727 //output taxonomy with the unclassified bins added
729 m->openInputFile(newTaxonomyFile, inTax);
732 string unclass = newTaxonomyFile + ".unclass.temp";
733 m->openOutputFile(unclass, outTax);
735 //get maxLevel from phylotree so you know how many 'unclassified's to add
736 int maxLevel = taxaSum.getMaxLevel();
738 //read taxfile - this reading and rewriting is done to preserve the confidence scores.
740 while (!inTax.eof()) {
741 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->mothurRemove(unclass); delete classify; return 0; }
743 inTax >> name >> taxon; m->gobble(inTax);
745 string newTax = addUnclassifieds(taxon, maxLevel);
747 outTax << name << '\t' << newTax << endl;
752 m->mothurRemove(newTaxonomyFile);
753 rename(unclass.c_str(), newTaxonomyFile.c_str());
755 m->mothurOutEndLine();
756 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
762 m->mothurOutEndLine();
763 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
764 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
765 m->mothurOutEndLine();
768 //set taxonomy file as new current taxonomyfile
770 itTypes = outputTypes.find("taxonomy");
771 if (itTypes != outputTypes.end()) {
772 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
776 itTypes = outputTypes.find("accnos");
777 if (itTypes != outputTypes.end()) {
778 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
785 catch(exception& e) {
786 m->errorOut(e, "ClassifySeqsCommand", "execute");
791 /**************************************************************************************************/
792 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
794 string newTax, taxon;
797 //keep what you have counting the levels
798 while (tax.find_first_of(';') != -1) {
800 taxon = tax.substr(0,tax.find_first_of(';'))+';';
801 tax = tax.substr(tax.find_first_of(';')+1, tax.length());
806 //add "unclassified" until you reach maxLevel
807 while (level < maxlevel) {
808 newTax += "unclassified;";
814 catch(exception& e) {
815 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
820 /**************************************************************************************************/
822 int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string accnos, string filename) {
828 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
831 //loop through and create all the processes you want
832 while (process != processors) {
836 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
839 num = driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", filename);
841 //pass numSeqs to parent
843 string tempFile = filename + toString(getpid()) + ".num.temp";
844 m->openOutputFile(tempFile, out);
850 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
851 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
856 //parent does its part
857 num = driver(lines[0], taxFileName, tempTaxFile, accnos, filename);
859 //force parent to wait until all the processes are done
860 for (int i=0;i<processIDS.size();i++) {
861 int temp = processIDS[i];
865 for (int i = 0; i < processIDS.size(); i++) {
867 string tempFile = filename + toString(processIDS[i]) + ".num.temp";
868 m->openInputFile(tempFile, in);
869 if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
870 in.close(); m->mothurRemove(m->getFullPathName(tempFile));
873 //////////////////////////////////////////////////////////////////////////////////////////////////////
874 //Windows version shared memory, so be careful when passing variables through the alignData struct.
875 //Above fork() will clone, so memory is separate, but that's not the case with windows,
876 //////////////////////////////////////////////////////////////////////////////////////////////////////
878 vector<classifyData*> pDataArray;
879 DWORD dwThreadIdArray[processors-1];
880 HANDLE hThreadArray[processors-1];
882 //Create processor worker threads.
883 for( int i=0; i<processors-1; i++ ){
884 // Allocate memory for thread data.
885 string extension = "";
886 if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
888 classifyData* tempclass = new classifyData((accnos + extension), probs, method, templateFileName, taxonomyFileName, (taxFileName + extension), (tempTaxFile + extension), filename, search, kmerSize, iters, numWanted, m, lines[i]->start, lines[i]->end, match, misMatch, gapOpen, gapExtend, cutoff, i, flip);
889 pDataArray.push_back(tempclass);
891 //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
892 //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
893 hThreadArray[i] = CreateThread(NULL, 0, MyClassThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
897 //parent does its part
898 num = driver(lines[processors-1], taxFileName + toString(processors-1) + ".temp", tempTaxFile + toString(processors-1) + ".temp", accnos + toString(processors-1) + ".temp", filename);
899 processIDS.push_back((processors-1));
901 //Wait until all threads have terminated.
902 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
904 //Close all thread handles and free memory allocations.
905 for(int i=0; i < pDataArray.size(); i++){
906 num += pDataArray[i]->count;
907 CloseHandle(hThreadArray[i]);
908 delete pDataArray[i];
912 vector<string> nonBlankAccnosFiles;
913 if (!(m->isBlank(accnos))) { nonBlankAccnosFiles.push_back(accnos); }
914 else { m->mothurRemove(accnos); } //remove so other files can be renamed to it
916 for(int i=0;i<processIDS.size();i++){
917 appendTaxFiles((taxFileName + toString(processIDS[i]) + ".temp"), taxFileName);
918 appendTaxFiles((tempTaxFile + toString(processIDS[i]) + ".temp"), tempTaxFile);
919 if (!(m->isBlank(accnos + toString(processIDS[i]) + ".temp"))) {
920 nonBlankAccnosFiles.push_back(accnos + toString(processIDS[i]) + ".temp");
921 }else { m->mothurRemove((accnos + toString(processIDS[i]) + ".temp")); }
923 m->mothurRemove((m->getFullPathName(taxFileName) + toString(processIDS[i]) + ".temp"));
924 m->mothurRemove((m->getFullPathName(tempTaxFile) + toString(processIDS[i]) + ".temp"));
927 //append accnos files
928 if (nonBlankAccnosFiles.size() != 0) {
929 rename(nonBlankAccnosFiles[0].c_str(), accnos.c_str());
931 for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
932 appendTaxFiles(nonBlankAccnosFiles[h], accnos);
933 m->mothurRemove(nonBlankAccnosFiles[h]);
935 }else { //recreate the accnosfile if needed
937 m->openOutputFile(accnos, out);
944 catch(exception& e) {
945 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
949 /**************************************************************************************************/
951 void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) {
956 m->openOutputFileAppend(filename, output);
957 m->openInputFile(temp, input);
959 while(char c = input.get()){
960 if(input.eof()) { break; }
961 else { output << c; }
967 catch(exception& e) {
968 m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles");
973 //**********************************************************************************************************************
975 int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempTFName, string accnos, string filename){
978 m->openOutputFile(taxFName, outTax);
980 ofstream outTaxSimple;
981 m->openOutputFile(tempTFName, outTaxSimple);
984 m->openOutputFile(accnos, outAcc);
987 m->openInputFile(filename, inFASTA);
991 inFASTA.seekg(filePos->start);
997 if (m->control_pressed) {
1000 outTaxSimple.close();
1001 outAcc.close(); return 0; }
1003 Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
1005 if (candidateSeq->getName() != "") {
1007 taxonomy = classify->getTaxonomy(candidateSeq);
1009 if (m->control_pressed) { delete candidateSeq; return 0; }
1011 if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); }
1013 //output confidence scores or not
1015 outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
1017 outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1020 if (classify->getFlipped()) { outAcc << candidateSeq->getName() << endl; }
1022 outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1026 delete candidateSeq;
1028 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1029 unsigned long long pos = inFASTA.tellg();
1030 if ((pos == -1) || (pos >= filePos->end)) { break; }
1032 if (inFASTA.eof()) { break; }
1036 if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
1040 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
1044 outTaxSimple.close();
1049 catch(exception& e) {
1050 m->errorOut(e, "ClassifySeqsCommand", "driver");
1054 //**********************************************************************************************************************
1056 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, MPI_File& accFile, vector<unsigned long long>& MPIPos){
1058 MPI_Status statusNew;
1059 MPI_Status statusTemp;
1060 MPI_Status statusAcc;
1064 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1067 string outputString;
1069 for(int i=0;i<num;i++){
1071 if (m->control_pressed) { return 0; }
1073 //read next sequence
1074 int length = MPIPos[start+i+1] - MPIPos[start+i];
1075 char* buf4 = new char[length];
1076 MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1078 string tempBuf = buf4;
1079 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
1080 istringstream iss (tempBuf,istringstream::in);
1083 Sequence* candidateSeq = new Sequence(iss);
1085 if (candidateSeq->getName() != "") {
1086 taxonomy = classify->getTaxonomy(candidateSeq);
1088 if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); }
1090 //output confidence scores or not
1092 outputString = candidateSeq->getName() + "\t" + taxonomy + "\n";
1094 outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1097 int length = outputString.length();
1098 char* buf2 = new char[length];
1099 memcpy(buf2, outputString.c_str(), length);
1101 MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
1104 outputString = candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1105 length = outputString.length();
1106 char* buf = new char[length];
1107 memcpy(buf, outputString.c_str(), length);
1109 MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
1112 if (classify->getFlipped()) {
1113 outputString = candidateSeq->getName() + "\n";
1114 length = outputString.length();
1115 char* buf3 = new char[length];
1116 memcpy(buf3, outputString.c_str(), length);
1118 MPI_File_write_shared(accFile, buf3, length, MPI_CHAR, &statusAcc);
1123 delete candidateSeq;
1125 if((i+1) % 100 == 0){ cout << "Classifying sequence " << (i+1) << endl; }
1128 if(num % 100 != 0){ cout << "Classifying sequence " << (num) << endl; }
1133 catch(exception& e) {
1134 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
1139 //**********************************************************************************************************************
1140 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
1143 nameMap.clear(); //remove old names
1149 //char* inFileName = new char[nameFilename.length()];
1150 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
1152 char inFileName[1024];
1153 strcpy(inFileName, nameFilename.c_str());
1155 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
1156 MPI_File_get_size(inMPI, &size);
1157 //delete inFileName;
1159 char* buffer = new char[size];
1160 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
1162 string tempBuf = buffer;
1163 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size); }
1164 istringstream iss (tempBuf,istringstream::in);
1167 string firstCol, secondCol;
1169 iss >> firstCol >> secondCol; m->gobble(iss);
1171 vector<string> temp;
1172 m->splitAtComma(secondCol, temp);
1174 nameMap[firstCol] = temp;
1177 MPI_File_close(&inMPI);
1181 catch(exception& e) {
1182 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
1187 /**************************************************************************************************/