classifyseqscommand.cpp

   1 /*
   2  *  classifyseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/2/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "classifyseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "bayesian.h"
  13 #include "phylotree.h"
  14 #include "phylosummary.h"
  15 #include "knn.h"
  16
  17 //**********************************************************************************************************************
  18
  19 ClassifySeqsCommand::ClassifySeqsCommand(string option)  {
  20         try {
  21                 abort = false;
  22
  23                 //allow user to run help
  24                 if(option == "help") { help(); abort = true; }
  25
  26                 else {
  27
  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)));
  31
  32                         OptionParser parser(option);
  33                         map<string, string> parameters = parser.getParameters();
  34
  35                         ValidParameters validParameter;
  36                         map<string, string>::iterator it;
  37
  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;  }
  41                         }
  42
  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 = "";         }
  45
  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 = "";          }
  49                         else {
  50                                 string path;
  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;         }
  57                                 }
  58
  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;         }
  65                                 }
  66
  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;            }
  73                                 }
  74                         }
  75
  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();
  81                                 abort = true;
  82                         }
  83                         else if (templateFileName == "not open") { abort = true; }
  84
  85                         groupfile = validParameter.validFile(parameters, "group", true);
  86                         if (groupfile == "not open") { abort = true; }
  87                         else if (groupfile == "not found") { groupfile = ""; }
  88
  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;  }
  91                         else {
  92                                 splitAtDash(fastaFileName, fastaFileNames);
  93
  94                                 //go through files and make sure they are good, if not, then disregard them
  95                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  96                                         if (inputDir != "") {
  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];               }
 100                                         }
 101
 102                                         int ableToOpen;
 103
 104                                         #ifdef USE_MPI
 105                                                 int pid;
 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
 108
 109                                                 if (pid == 0) {
 110                                         #endif
 111
 112                                         ifstream in;
 113                                         ableToOpen = openInputFile(fastaFileNames[i], in);
 114                                         in.close();
 115
 116                                         #ifdef USE_MPI
 117                                                         for (int j = 1; j < processors; j++) {
 118                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
 119                                                         }
 120                                                 }else{
 121                                                         MPI_Status status;
 122                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 123                                                 }
 124
 125                                         #endif
 126
 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);
 131                                                 i--;
 132                                         }
 133
 134                                 }
 135
 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; }
 138                         }
 139
 140
 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();
 145                                 abort = true;
 146                         }
 147                         else if (taxonomyFileName == "not open") { abort = true; }
 148
 149
 150                         namefile = validParameter.validFile(parameters, "name", false);
 151                         if (namefile == "not found") { namefile = "";  }
 152
 153                         else {
 154                                 splitAtDash(namefile, namefileNames);
 155
 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];         }
 162                                         }
 163                                         int ableToOpen;
 164
 165                                         #ifdef USE_MPI
 166                                                 int pid;
 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
 169
 170                                                 if (pid == 0) {
 171                                         #endif
 172
 173                                         ifstream in;
 174                                         ableToOpen = openInputFile(namefileNames[i], in);
 175                                         in.close();
 176
 177                                         #ifdef USE_MPI
 178                                                         for (int j = 1; j < processors; j++) {
 179                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
 180                                                         }
 181                                                 }else{
 182                                                         MPI_Status status;
 183                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 184                                                 }
 185
 186                                         #endif
 187                                         if (ableToOpen == 1) {  m->mothurOut("Unable to match name file with fasta file."); m->mothurOutEndLine(); abort = true;        }
 188
 189                                 }
 190                         }
 191
 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(); }
 194                         }
 195
 196                         //check for optional parameter and set defaults
 197                         // ...at some point should added some additional type checking...
 198                         string temp;
 199                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
 200                         convert(temp, kmerSize);
 201
 202                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 203                         convert(temp, processors);
 204
 205                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 206
 207                         method = validParameter.validFile(parameters, "method", false);         if (method == "not found"){     method = "bayesian";    }
 208
 209                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 210                         convert(temp, match);
 211
 212                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 213                         convert(temp, misMatch);
 214
 215                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 216                         convert(temp, gapOpen);
 217
 218                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 219                         convert(temp, gapExtend);
 220
 221                         temp = validParameter.validFile(parameters, "numwanted", false);        if (temp == "not found"){       temp = "10";                    }
 222                         convert(temp, numWanted);
 223
 224                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found"){       temp = "0";                             }
 225                         convert(temp, cutoff);
 226
 227                         temp = validParameter.validFile(parameters, "probs", false);            if (temp == "not found"){       temp = "true";                  }
 228                         probs = isTrue(temp);
 229
 230                         temp = validParameter.validFile(parameters, "iters", false);            if (temp == "not found") { temp = "100";                        }
 231                         convert(temp, iters);
 232
 233
 234
 235                         if ((method == "bayesian") && (search != "kmer"))  {
 236                                 m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine();
 237                                 search = "kmer";
 238                         }
 239                 }
 240
 241         }
 242         catch(exception& e) {
 243                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 244                 exit(1);
 245         }
 246 }
 247
 248 //**********************************************************************************************************************
 249
 250 ClassifySeqsCommand::~ClassifySeqsCommand(){
 251
 252         if (abort == false) {
 253                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 254         }
 255 }
 256
 257 //**********************************************************************************************************************
 258
 259 void ClassifySeqsCommand::help(){
 260         try {
 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");
 270                 #ifdef USE_MPI
 271                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 272                 #endif
 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");
 287         }
 288         catch(exception& e) {
 289                 m->errorOut(e, "ClassifySeqsCommand", "help");
 290                 exit(1);
 291         }
 292 }
 293
 294
 295 //**********************************************************************************************************************
 296
 297 int ClassifySeqsCommand::execute(){
 298         try {
 299                 if (abort == true) {    return 0;       }
 300
 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);                               }
 303                 else {
 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);
 307                 }
 308
 309                 if (m->control_pressed) { delete classify; return 0; }
 310
 311                 vector<string> outputNames;
 312
 313                 for (int s = 0; s < fastaFileNames.size(); s++) {
 314
 315                         m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 316
 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";
 321
 322                         outputNames.push_back(newTaxonomyFile);
 323                         outputNames.push_back(taxSummary);
 324
 325                         int start = time(NULL);
 326                         int numFastaSeqs = 0;
 327                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 328
 329 #ifdef USE_MPI
 330                                 int pid, end, numSeqsPerProcessor;
 331                                 int tag = 2001;
 332                                 vector<long> MPIPos;
 333
 334                                 MPI_Status status;
 335                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 336                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 337
 338                                 MPI_File inMPI;
 339                                 MPI_File outMPINewTax;
 340                                 MPI_File outMPITempTax;
 341
 342                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 343                                 int inMode=MPI_MODE_RDONLY;
 344
 345                                 //char* outNewTax = new char[newTaxonomyFile.length()];
 346                                 //memcpy(outNewTax, newTaxonomyFile.c_str(), newTaxonomyFile.length());
 347
 348                                 char outNewTax[1024];
 349                                 strcpy(outNewTax, newTaxonomyFile.c_str());
 350
 351                                 //char* outTempTax = new char[tempTaxonomyFile.length()];
 352                                 //memcpy(outTempTax, tempTaxonomyFile.c_str(), tempTaxonomyFile.length());
 353
 354                                 char outTempTax[1024];
 355                                 strcpy(outTempTax, tempTaxonomyFile.c_str());
 356
 357                                 //char* inFileName = new char[fastaFileNames[s].length()];
 358                                 //memcpy(inFileName, fastaFileNames[s].c_str(), fastaFileNames[s].length());
 359
 360                                 char inFileName[1024];
 361                                 strcpy(inFileName, fastaFileNames[s].c_str());
 362
 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);
 366
 367                                 //delete outNewTax;
 368                                 //delete outTempTax;
 369                                 //delete inFileName;
 370
 371                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);   MPI_File_close(&outMPITempTax);  delete classify; return 0;  }
 372
 373                                 if(namefile != "") {  MPIReadNamesFile(namefileNames[s]);  }
 374
 375                                 if (pid == 0) { //you are the root process
 376
 377                                         MPIPos = setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 378
 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
 382
 383                                         //figure out how many sequences you have to align
 384                                         numSeqsPerProcessor = numFastaSeqs / processors;
 385                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 386                                         int startIndex =  pid * numSeqsPerProcessor;
 387
 388                                         //align your part
 389                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
 390
 391                                         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;  }
 392
 393                                         for (int i = 1; i < processors; i++) {
 394                                                 int done;
 395                                                 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 396                                         }
 397                                 }else{ //you are a child process
 398                                         MPI_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
 399                                         MPIPos.resize(numFastaSeqs+1);
 400                                         MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
 401
 402                                         //figure out how many sequences you have to align
 403                                         numSeqsPerProcessor = numFastaSeqs / processors;
 404                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 405                                         int startIndex =  pid * numSeqsPerProcessor;
 406
 407                                         //align your part
 408                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
 409
 410                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);   MPI_File_close(&outMPITempTax);  delete classify; return 0;  }
 411
 412                                         int done = 0;
 413                                         MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 414                                 }
 415
 416                                 //close files
 417                                 MPI_File_close(&inMPI);
 418                                 MPI_File_close(&outMPINewTax);
 419                                 MPI_File_close(&outMPITempTax);
 420
 421 #else
 422                         //read namefile
 423                         if(namefile != "") {
 424                                 nameMap.clear(); //remove old names
 425
 426                                 ifstream inNames;
 427                                 openInputFile(namefileNames[s], inNames);
 428
 429                                 string firstCol, secondCol;
 430                                 while(!inNames.eof()) {
 431                                         inNames >> firstCol >> secondCol; gobble(inNames);
 432                                         nameMap[firstCol] = getNumNames(secondCol);  //ex. seq1 seq1,seq3,seq5 -> seq1 = 3.
 433                                 }
 434                                 inNames.close();
 435                         }
 436
 437         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 438                         if(processors == 1){
 439                                 ifstream inFASTA;
 440                                 openInputFile(fastaFileNames[s], inFASTA);
 441                                 numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 442                                 inFASTA.close();
 443
 444                                 lines.push_back(new linePair(0, numFastaSeqs));
 445
 446                                 driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 447                         }
 448                         else{
 449                                 vector<int> positions;
 450                                 processIDS.resize(0);
 451
 452                                 ifstream inFASTA;
 453                                 openInputFile(fastaFileNames[s], inFASTA);
 454
 455                                 string input;
 456                                 while(!inFASTA.eof()){
 457                                         input = getline(inFASTA);
 458                                         if (input.length() != 0) {
 459                                                 if(input[0] == '>'){    int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);       }
 460                                         }
 461                                 }
 462                                 inFASTA.close();
 463
 464                                 numFastaSeqs = positions.size();
 465
 466                                 int numSeqsPerProcessor = numFastaSeqs / processors;
 467
 468                                 for (int i = 0; i < processors; i++) {
 469                                         int startPos = positions[ i * numSeqsPerProcessor ];
 470                                         if(i == processors - 1){
 471                                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 472                                         }
 473                                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 474                                 }
 475                                 createProcesses(newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 476
 477                                 rename((newTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), newTaxonomyFile.c_str());
 478                                 rename((tempTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), tempTaxonomyFile.c_str());
 479
 480                                 for(int i=1;i<processors;i++){
 481                                         appendTaxFiles((newTaxonomyFile + toString(processIDS[i]) + ".temp"), newTaxonomyFile);
 482                                         appendTaxFiles((tempTaxonomyFile + toString(processIDS[i]) + ".temp"), tempTaxonomyFile);
 483                                         remove((newTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
 484                                         remove((tempTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
 485                                 }
 486
 487                         }
 488         #else
 489                         ifstream inFASTA;
 490                         openInputFile(fastaFileNames[s], inFASTA);
 491                         numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 492                         inFASTA.close();
 493
 494                         lines.push_back(new linePair(0, numFastaSeqs));
 495
 496                         driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 497         #endif
 498 #endif
 499
 500                 #ifdef USE_MPI
 501                         if (pid == 0) {  //this part does not need to be paralellized
 502                 #endif
 503
 504                         m->mothurOutEndLine();
 505                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 506                         start = time(NULL);
 507
 508                         PhyloSummary taxaSum(taxonomyFileName, groupfile);
 509
 510                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete classify; return 0; }
 511
 512                         if (namefile == "") {  taxaSum.summarize(tempTaxonomyFile);  }
 513                         else {
 514                                 ifstream in;
 515                                 openInputFile(tempTaxonomyFile, in);
 516
 517                                 //read in users taxonomy file and add sequences to tree
 518                                 string name, taxon;
 519                                 while(!in.eof()){
 520                                         in >> name >> taxon; gobble(in);
 521
 522                                         itNames = nameMap.find(name);
 523
 524                                         if (itNames == nameMap.end()) {
 525                                                 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
 526                                         }else{
 527                                                 for (int i = 0; i < itNames->second; i++) {
 528                                                         taxaSum.addSeqToTree(name+toString(i), taxon);  //add it as many times as there are identical seqs
 529                                                 }
 530                                         }
 531                                 }
 532                                 in.close();
 533                         }
 534                         remove(tempTaxonomyFile.c_str());
 535
 536                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete classify; return 0; }
 537
 538                         //print summary file
 539                         ofstream outTaxTree;
 540                         openOutputFile(taxSummary, outTaxTree);
 541                         taxaSum.print(outTaxTree);
 542                         outTaxTree.close();
 543
 544                         //output taxonomy with the unclassified bins added
 545                         ifstream inTax;
 546                         openInputFile(newTaxonomyFile, inTax);
 547
 548                         ofstream outTax;
 549                         string unclass = newTaxonomyFile + ".unclass.temp";
 550                         openOutputFile(unclass, outTax);
 551
 552                         //get maxLevel from phylotree so you know how many 'unclassified's to add
 553                         int maxLevel = taxaSum.getMaxLevel();
 554
 555                         //read taxfile - this reading and rewriting is done to preserve the confidence scores.
 556                         string name, taxon;
 557                         while (!inTax.eof()) {
 558                                 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; }
 559
 560                                 inTax >> name >> taxon; gobble(inTax);
 561
 562                                 string newTax = addUnclassifieds(taxon, maxLevel);
 563
 564                                 outTax << name << '\t' << newTax << endl;
 565                         }
 566                         inTax.close();
 567                         outTax.close();
 568
 569                         remove(newTaxonomyFile.c_str());
 570                         rename(unclass.c_str(), newTaxonomyFile.c_str());
 571
 572                         m->mothurOutEndLine();
 573                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for  " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 574
 575                         #ifdef USE_MPI
 576                                 }
 577                         #endif
 578
 579                         m->mothurOutEndLine();
 580                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 581                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 582                         m->mothurOutEndLine();
 583
 584                 }
 585
 586                 delete classify;
 587                 return 0;
 588         }
 589         catch(exception& e) {
 590                 m->errorOut(e, "ClassifySeqsCommand", "execute");
 591                 exit(1);
 592         }
 593 }
 594
 595 /**************************************************************************************************/
 596 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
 597         try{
 598                 string newTax, taxon;
 599                 int level = 0;
 600
 601                 //keep what you have counting the levels
 602                 while (tax.find_first_of(';') != -1) {
 603                         //get taxon
 604                         taxon = tax.substr(0,tax.find_first_of(';'))+';';
 605                         tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 606                         newTax += taxon;
 607                         level++;
 608                 }
 609
 610                 //add "unclassified" until you reach maxLevel
 611                 while (level < maxlevel) {
 612                         newTax += "unclassified;";
 613                         level++;
 614                 }
 615
 616                 return newTax;
 617         }
 618         catch(exception& e) {
 619                 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
 620                 exit(1);
 621         }
 622 }
 623
 624 /**************************************************************************************************/
 625
 626 void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string filename) {
 627         try {
 628 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 629                 int process = 0;
 630                 //              processIDS.resize(0);
 631
 632                 //loop through and create all the processes you want
 633                 while (process != processors) {
 634                         int pid = fork();
 635
 636                         if (pid > 0) {
 637                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 638                                 process++;
 639                         }else if (pid == 0){
 640                                 driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", filename);
 641                                 exit(0);
 642                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 643                 }
 644
 645                 //force parent to wait until all the processes are done
 646                 for (int i=0;i<processors;i++) {
 647                         int temp = processIDS[i];
 648                         wait(&temp);
 649                 }
 650 #endif
 651         }
 652         catch(exception& e) {
 653                 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
 654                 exit(1);
 655         }
 656 }
 657 /**************************************************************************************************/
 658
 659 void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) {
 660         try{
 661
 662                 ofstream output;
 663                 ifstream input;
 664                 openOutputFileAppend(filename, output);
 665                 openInputFile(temp, input);
 666
 667                 while(char c = input.get()){
 668                         if(input.eof())         {       break;                  }
 669                         else                            {       output << c;    }
 670                 }
 671
 672                 input.close();
 673                 output.close();
 674         }
 675         catch(exception& e) {
 676                 m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles");
 677                 exit(1);
 678         }
 679 }
 680
 681 //**********************************************************************************************************************
 682
 683 int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFName, string filename){
 684         try {
 685                 ofstream outTax;
 686                 openOutputFile(taxFName, outTax);
 687
 688                 ofstream outTaxSimple;
 689                 openOutputFile(tempTFName, outTaxSimple);
 690
 691                 ifstream inFASTA;
 692                 openInputFile(filename, inFASTA);
 693
 694                 inFASTA.seekg(line->start);
 695
 696                 string taxonomy;
 697
 698                 for(int i=0;i<line->numSeqs;i++){
 699                         if (m->control_pressed) { return 0; }
 700
 701                         Sequence* candidateSeq = new Sequence(inFASTA);
 702
 703                         if (candidateSeq->getName() != "") {
 704                                 taxonomy = classify->getTaxonomy(candidateSeq);
 705
 706                                 if (m->control_pressed) { delete candidateSeq; return 0; }
 707
 708                                 if (taxonomy != "bad seq") {
 709                                         //output confidence scores or not
 710                                         if (probs) {
 711                                                 outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
 712                                         }else{
 713                                                 outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
 714                                         }
 715
 716                                         outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
 717                                 }
 718                         }
 719                         delete candidateSeq;
 720
 721                         if((i+1) % 100 == 0){
 722                                 m->mothurOut("Classifying sequence " + toString(i+1)); m->mothurOutEndLine();
 723                         }
 724                 }
 725
 726                 inFASTA.close();
 727                 outTax.close();
 728                 outTaxSimple.close();
 729
 730                 return 1;
 731         }
 732         catch(exception& e) {
 733                 m->errorOut(e, "ClassifySeqsCommand", "driver");
 734                 exit(1);
 735         }
 736 }
 737 //**********************************************************************************************************************
 738 #ifdef USE_MPI
 739 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, vector<long>& MPIPos){
 740         try {
 741                 MPI_Status statusNew;
 742                 MPI_Status statusTemp;
 743                 MPI_Status status;
 744
 745                 int pid;
 746                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 747
 748                 string taxonomy;
 749                 string outputString;
 750
 751                 for(int i=0;i<num;i++){
 752
 753                         if (m->control_pressed) { return 0; }
 754
 755                         //read next sequence
 756                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 757                         char* buf4 = new char[length];
 758                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 759
 760                         string tempBuf = buf4;
 761                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 762                         istringstream iss (tempBuf,istringstream::in);
 763                         delete buf4;
 764
 765                         Sequence* candidateSeq = new Sequence(iss);
 766
 767                         if (candidateSeq->getName() != "") {
 768                                 taxonomy = classify->getTaxonomy(candidateSeq);
 769
 770                                 if (taxonomy != "bad seq") {
 771                                         //output confidence scores or not
 772                                         if (probs) {
 773                                                 outputString =  candidateSeq->getName() + "\t" + taxonomy + "\n";
 774                                         }else{
 775                                                 outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
 776                                         }
 777
 778                                         int length = outputString.length();
 779                                         char* buf2 = new char[length];
 780                                         memcpy(buf2, outputString.c_str(), length);
 781
 782                                         MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
 783                                         delete buf2;
 784
 785                                         outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
 786                                         length = outputString.length();
 787                                         char* buf = new char[length];
 788                                         memcpy(buf, outputString.c_str(), length);
 789
 790                                         MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
 791                                         delete buf;
 792                                 }
 793                         }
 794                         delete candidateSeq;
 795
 796                         if((i+1) % 100 == 0){   cout << "Classifying sequence " << (i+1) << endl;       }
 797                 }
 798
 799                 if(num % 100 != 0){     cout << "Classifying sequence " << (num) << endl;       }
 800
 801
 802                 return 1;
 803         }
 804         catch(exception& e) {
 805                 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
 806                 exit(1);
 807         }
 808 }
 809
 810 //**********************************************************************************************************************
 811 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
 812         try {
 813
 814                 nameMap.clear(); //remove old names
 815
 816                 MPI_File inMPI;
 817                 MPI_Offset size;
 818                 MPI_Status status;
 819
 820                 //char* inFileName = new char[nameFilename.length()];
 821                 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
 822
 823                 char inFileName[1024];
 824                 strcpy(inFileName, nameFilename.c_str());
 825
 826                 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 827                 MPI_File_get_size(inMPI, &size);
 828                 //delete inFileName;
 829
 830                 char* buffer = new char[size];
 831                 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
 832
 833                 string tempBuf = buffer;
 834                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 835                 istringstream iss (tempBuf,istringstream::in);
 836                 delete buffer;
 837
 838                 string firstCol, secondCol;
 839                 while(!iss.eof()) {
 840                         iss >> firstCol >> secondCol; gobble(iss);
 841                         nameMap[firstCol] = getNumNames(secondCol);  //ex. seq1 seq1,seq3,seq5 -> seq1 = 3.
 842                 }
 843
 844                 MPI_File_close(&inMPI);
 845
 846                 return 1;
 847         }
 848         catch(exception& e) {
 849                 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
 850                 exit(1);
 851         }
 852 }
 853 #endif
 854 /**************************************************************************************************/