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
  86                         fastaFileName = validParameter.validFile(parameters, "fasta", false);
  87                         if (fastaFileName == "not found") { m->mothurOut("fasta is a required parameter for the classify.seqs command."); m->mothurOutEndLine(); abort = true;  }
  88                         else {
  89                                 splitAtDash(fastaFileName, fastaFileNames);
  90
  91                                 //go through files and make sure they are good, if not, then disregard them
  92                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  93                                         if (inputDir != "") {
  94                                                 string path = hasPath(fastaFileNames[i]);
  95                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  96                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
  97                                         }
  98
  99                                         int ableToOpen;
 100
 101                                         #ifdef USE_MPI
 102                                                 int pid;
 103                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 104                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 105
 106                                                 if (pid == 0) {
 107                                         #endif
 108
 109                                         ifstream in;
 110                                         ableToOpen = openInputFile(fastaFileNames[i], in);
 111                                         in.close();
 112
 113                                         #ifdef USE_MPI
 114                                                         for (int j = 1; j < processors; j++) {
 115                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
 116                                                         }
 117                                                 }else{
 118                                                         MPI_Status status;
 119                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 120                                                 }
 121
 122                                         #endif
 123
 124                                         if (ableToOpen == 1) {
 125                                                 m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
 126                                                 //erase from file list
 127                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 128                                                 i--;
 129                                         }
 130
 131                                 }
 132
 133                                 //make sure there is at least one valid file left
 134                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 135                         }
 136
 137
 138                         taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
 139                         if (taxonomyFileName == "not found") {
 140                                 m->mothurOut("taxonomy is a required parameter for the classify.seqs command.");
 141                                 m->mothurOutEndLine();
 142                                 abort = true;
 143                         }
 144                         else if (taxonomyFileName == "not open") { abort = true; }
 145
 146
 147                         namefile = validParameter.validFile(parameters, "name", false);
 148                         if (namefile == "not found") { namefile = "";  }
 149
 150                         else {
 151                                 splitAtDash(namefile, namefileNames);
 152
 153                                 //go through files and make sure they are good, if not, then disregard them
 154                                 for (int i = 0; i < namefileNames.size(); i++) {
 155                                         if (inputDir != "") {
 156                                                 string path = hasPath(namefileNames[i]);
 157                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 158                                                 if (path == "") {       namefileNames[i] = inputDir + namefileNames[i];         }
 159                                         }
 160                                         int ableToOpen;
 161
 162                                         #ifdef USE_MPI
 163                                                 int pid;
 164                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 165                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 166
 167                                                 if (pid == 0) {
 168                                         #endif
 169
 170                                         ifstream in;
 171                                         ableToOpen = openInputFile(namefileNames[i], in);
 172                                         in.close();
 173
 174                                         #ifdef USE_MPI
 175                                                         for (int j = 1; j < processors; j++) {
 176                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
 177                                                         }
 178                                                 }else{
 179                                                         MPI_Status status;
 180                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 181                                                 }
 182
 183                                         #endif
 184                                         if (ableToOpen == 1) {  m->mothurOut("Unable to match name file with fasta file."); m->mothurOutEndLine(); abort = true;        }
 185
 186                                 }
 187                         }
 188
 189                         if (namefile != "") {
 190                                 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(); }
 191                         }
 192
 193                         groupfile = validParameter.validFile(parameters, "group", false);
 194                         if (groupfile == "not found") { groupfile = "";  }
 195                         else {
 196                                 splitAtDash(groupfile, groupfileNames);
 197
 198                                 //go through files and make sure they are good, if not, then disregard them
 199                                 for (int i = 0; i < groupfileNames.size(); i++) {
 200                                         if (inputDir != "") {
 201                                                 string path = hasPath(groupfileNames[i]);
 202                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 203                                                 if (path == "") {       groupfileNames[i] = inputDir + groupfileNames[i];               }
 204                                         }
 205                                         int ableToOpen;
 206
 207                                         #ifdef USE_MPI
 208                                                 int pid;
 209                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 210                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 211
 212                                                 if (pid == 0) {
 213                                         #endif
 214
 215                                         ifstream in;
 216                                         ableToOpen = openInputFile(groupfileNames[i], in);
 217                                         in.close();
 218
 219                                         #ifdef USE_MPI
 220                                                         for (int j = 1; j < processors; j++) {
 221                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
 222                                                         }
 223                                                 }else{
 224                                                         MPI_Status status;
 225                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 226                                                 }
 227
 228                                         #endif
 229                                         if (ableToOpen == 1) {  m->mothurOut("Unable to match group file with fasta file, not using " + groupfileNames[i] + "."); m->mothurOutEndLine(); groupfileNames[i] = "";        }
 230
 231                                 }
 232                         }
 233
 234                         if (groupfile != "") {
 235                                 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(); }
 236                         }else {
 237                                 for (int i = 0; i < fastaFileNames.size(); i++) {  groupfileNames.push_back("");  }
 238                         }
 239
 240                         //check for optional parameter and set defaults
 241                         // ...at some point should added some additional type checking...
 242                         string temp;
 243                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
 244                         convert(temp, kmerSize);
 245
 246                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 247                         convert(temp, processors);
 248
 249                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 250
 251                         method = validParameter.validFile(parameters, "method", false);         if (method == "not found"){     method = "bayesian";    }
 252
 253                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 254                         convert(temp, match);
 255
 256                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 257                         convert(temp, misMatch);
 258
 259                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 260                         convert(temp, gapOpen);
 261
 262                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 263                         convert(temp, gapExtend);
 264
 265                         temp = validParameter.validFile(parameters, "numwanted", false);        if (temp == "not found"){       temp = "10";                    }
 266                         convert(temp, numWanted);
 267
 268                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found"){       temp = "0";                             }
 269                         convert(temp, cutoff);
 270
 271                         temp = validParameter.validFile(parameters, "probs", false);            if (temp == "not found"){       temp = "true";                  }
 272                         probs = isTrue(temp);
 273
 274                         temp = validParameter.validFile(parameters, "iters", false);            if (temp == "not found") { temp = "100";                        }
 275                         convert(temp, iters);
 276
 277
 278
 279                         if ((method == "bayesian") && (search != "kmer"))  {
 280                                 m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine();
 281                                 search = "kmer";
 282                         }
 283                 }
 284
 285         }
 286         catch(exception& e) {
 287                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 288                 exit(1);
 289         }
 290 }
 291
 292 //**********************************************************************************************************************
 293
 294 ClassifySeqsCommand::~ClassifySeqsCommand(){
 295
 296         if (abort == false) {
 297                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 298         }
 299 }
 300
 301 //**********************************************************************************************************************
 302
 303 void ClassifySeqsCommand::help(){
 304         try {
 305                 m->mothurOut("The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n");
 306                 m->mothurOut("The classify.seqs command parameters are template, fasta, name, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n");
 307                 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");
 308                 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");
 309                 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");
 310                 m->mothurOut("The group parameter allows you add a group file so you can have the summary totals broken up by group.\n");
 311                 m->mothurOut("The method parameter allows you to specify classification method to use.  Your options are: bayesian and knn. The default is bayesian.\n");
 312                 m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 8.\n");
 313                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 314                 #ifdef USE_MPI
 315                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 316                 #endif
 317                 m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");
 318                 m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n");
 319                 m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n");
 320                 m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n");
 321                 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");
 322                 m->mothurOut("The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy.  The default is 0.\n");
 323                 m->mothurOut("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");
 324                 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");
 325                 m->mothurOut("The classify.seqs command should be in the following format: \n");
 326                 m->mothurOut("classify.seqs(template=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n");
 327                 m->mothurOut("Example classify.seqs(fasta=amazon.fasta, template=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n");
 328                 m->mothurOut("The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n");
 329                 m->mothurOut("The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n");
 330                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "ClassifySeqsCommand", "help");
 334                 exit(1);
 335         }
 336 }
 337
 338
 339 //**********************************************************************************************************************
 340
 341 int ClassifySeqsCommand::execute(){
 342         try {
 343                 if (abort == true) {    return 0;       }
 344
 345                 if(method == "bayesian"){       classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters);           }
 346                 else if(method == "knn"){       classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted);                               }
 347                 else {
 348                         m->mothurOut(search + " is not a valid method option. I will run the command using bayesian.");
 349                         m->mothurOutEndLine();
 350                         classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters);
 351                 }
 352
 353                 if (m->control_pressed) { delete classify; return 0; }
 354
 355                 vector<string> outputNames;
 356
 357                 for (int s = 0; s < fastaFileNames.size(); s++) {
 358
 359                         m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 360
 361                         if (outputDir == "") { outputDir += hasPath(fastaFileNames[s]); }
 362                         string newTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "taxonomy";
 363                         string tempTaxonomyFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "taxonomy.temp";
 364                         string taxSummary = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + getRootName(getSimpleName(taxonomyFileName)) + "tax.summary";
 365
 366                         outputNames.push_back(newTaxonomyFile);
 367                         outputNames.push_back(taxSummary);
 368
 369                         int start = time(NULL);
 370                         int numFastaSeqs = 0;
 371                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 372
 373 #ifdef USE_MPI
 374                                 int pid, end, numSeqsPerProcessor;
 375                                 int tag = 2001;
 376                                 vector<long> MPIPos;
 377
 378                                 MPI_Status status;
 379                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 380                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 381
 382                                 MPI_File inMPI;
 383                                 MPI_File outMPINewTax;
 384                                 MPI_File outMPITempTax;
 385
 386                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 387                                 int inMode=MPI_MODE_RDONLY;
 388
 389                                 //char* outNewTax = new char[newTaxonomyFile.length()];
 390                                 //memcpy(outNewTax, newTaxonomyFile.c_str(), newTaxonomyFile.length());
 391
 392                                 char outNewTax[1024];
 393                                 strcpy(outNewTax, newTaxonomyFile.c_str());
 394
 395                                 //char* outTempTax = new char[tempTaxonomyFile.length()];
 396                                 //memcpy(outTempTax, tempTaxonomyFile.c_str(), tempTaxonomyFile.length());
 397
 398                                 char outTempTax[1024];
 399                                 strcpy(outTempTax, tempTaxonomyFile.c_str());
 400
 401                                 //char* inFileName = new char[fastaFileNames[s].length()];
 402                                 //memcpy(inFileName, fastaFileNames[s].c_str(), fastaFileNames[s].length());
 403
 404                                 char inFileName[1024];
 405                                 strcpy(inFileName, fastaFileNames[s].c_str());
 406
 407                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 408                                 MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax);
 409                                 MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax);
 410
 411                                 //delete outNewTax;
 412                                 //delete outTempTax;
 413                                 //delete inFileName;
 414
 415                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);   MPI_File_close(&outMPITempTax);  delete classify; return 0;  }
 416
 417                                 if (pid == 0) { //you are the root process
 418
 419                                         MPIPos = setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 420
 421                                         //send file positions to all processes
 422                                         for(int i = 1; i < processors; i++) {
 423                                                 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 424                                                 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 425                                         }
 426
 427                                         //figure out how many sequences you have to align
 428                                         numSeqsPerProcessor = numFastaSeqs / processors;
 429                                         int startIndex =  pid * numSeqsPerProcessor;
 430                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 431
 432
 433                                         //align your part
 434                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
 435
 436                                         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;  }
 437
 438                                         for (int i = 1; i < processors; i++) {
 439                                                 int done;
 440                                                 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 441                                         }
 442                                 }else{ //you are a child process
 443                                         MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 444                                         MPIPos.resize(numFastaSeqs+1);
 445                                         MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 446
 447                                         //figure out how many sequences you have to align
 448                                         numSeqsPerProcessor = numFastaSeqs / processors;
 449                                         int startIndex =  pid * numSeqsPerProcessor;
 450                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 451
 452
 453                                         //align your part
 454                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, MPIPos);
 455
 456                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);   MPI_File_close(&outMPITempTax);  delete classify; return 0;  }
 457
 458                                         int done = 0;
 459                                         MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 460                                 }
 461
 462                                 //close files
 463                                 MPI_File_close(&inMPI);
 464                                 MPI_File_close(&outMPINewTax);
 465                                 MPI_File_close(&outMPITempTax);
 466                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 467
 468 #else
 469                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 470                         if(processors == 1){
 471                                 ifstream inFASTA;
 472                                 openInputFile(fastaFileNames[s], inFASTA);
 473                                 numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 474                                 inFASTA.close();
 475
 476                                 lines.push_back(new linePair(0, numFastaSeqs));
 477
 478                                 driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 479                         }
 480                         else{
 481                                 vector<int> positions;
 482                                 processIDS.resize(0);
 483
 484                                 ifstream inFASTA;
 485                                 openInputFile(fastaFileNames[s], inFASTA);
 486
 487                                 string input;
 488                                 while(!inFASTA.eof()){
 489                                         input = getline(inFASTA);
 490                                         if (input.length() != 0) {
 491                                                 if(input[0] == '>'){    int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);       }
 492                                         }
 493                                 }
 494                                 inFASTA.close();
 495
 496                                 numFastaSeqs = positions.size();
 497
 498                                 int numSeqsPerProcessor = numFastaSeqs / processors;
 499
 500                                 for (int i = 0; i < processors; i++) {
 501                                         int startPos = positions[ i * numSeqsPerProcessor ];
 502                                         if(i == processors - 1){
 503                                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 504                                         }
 505                                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 506                                 }
 507                                 createProcesses(newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 508
 509                                 rename((newTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), newTaxonomyFile.c_str());
 510                                 rename((tempTaxonomyFile + toString(processIDS[0]) + ".temp").c_str(), tempTaxonomyFile.c_str());
 511
 512                                 for(int i=1;i<processors;i++){
 513                                         appendTaxFiles((newTaxonomyFile + toString(processIDS[i]) + ".temp"), newTaxonomyFile);
 514                                         appendTaxFiles((tempTaxonomyFile + toString(processIDS[i]) + ".temp"), tempTaxonomyFile);
 515                                         remove((newTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
 516                                         remove((tempTaxonomyFile + toString(processIDS[i]) + ".temp").c_str());
 517                                 }
 518
 519                         }
 520         #else
 521                         ifstream inFASTA;
 522                         openInputFile(fastaFileNames[s], inFASTA);
 523                         numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 524                         inFASTA.close();
 525
 526                         lines.push_back(new linePair(0, numFastaSeqs));
 527
 528                         driver(lines[0], newTaxonomyFile, tempTaxonomyFile, fastaFileNames[s]);
 529         #endif
 530 #endif
 531
 532                 m->mothurOutEndLine();
 533                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 534                 start = time(NULL);
 535
 536
 537                 #ifdef USE_MPI
 538                         if (pid == 0) {  //this part does not need to be paralellized
 539
 540                                 if(namefile != "") { m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();  MPIReadNamesFile(namefileNames[s]);  m->mothurOut("  Done."); m->mothurOutEndLine(); }
 541                 #else
 542                         //read namefile
 543                         if(namefile != "") {
 544
 545                             m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();
 546
 547                                 nameMap.clear(); //remove old names
 548
 549                                 ifstream inNames;
 550                                 openInputFile(namefileNames[s], inNames);
 551
 552                                 string firstCol, secondCol;
 553                                 while(!inNames.eof()) {
 554                                         inNames >> firstCol >> secondCol; gobble(inNames);
 555
 556                                         vector<string> temp;
 557                                         splitAtComma(secondCol, temp);
 558
 559                                         nameMap[firstCol] = temp;
 560                                 }
 561                                 inNames.close();
 562
 563                                 m->mothurOut("  Done."); m->mothurOutEndLine();
 564                         }
 565                 #endif
 566
 567                         string group = "";
 568                         if (groupfile != "") {  group = groupfileNames[s]; }
 569
 570                         PhyloSummary taxaSum(taxonomyFileName, group);
 571
 572                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete classify; return 0; }
 573
 574                         if (namefile == "") {  taxaSum.summarize(tempTaxonomyFile);  }
 575                         else {
 576                                 ifstream in;
 577                                 openInputFile(tempTaxonomyFile, in);
 578
 579                                 //read in users taxonomy file and add sequences to tree
 580                                 string name, taxon;
 581
 582                                 while(!in.eof()){
 583                                         in >> name >> taxon; gobble(in);
 584
 585                                         itNames = nameMap.find(name);
 586
 587                                         if (itNames == nameMap.end()) {
 588                                                 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
 589                                         }else{
 590                                                 for (int i = 0; i < itNames->second.size(); i++) {
 591                                                         taxaSum.addSeqToTree(itNames->second[i], taxon);  //add it as many times as there are identical seqs
 592                                                 }
 593                                                 itNames->second.clear();
 594                                                 nameMap.erase(itNames->first);
 595                                         }
 596                                 }
 597                                 in.close();
 598                         }
 599                         remove(tempTaxonomyFile.c_str());
 600
 601                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete classify; return 0; }
 602
 603                         //print summary file
 604                         ofstream outTaxTree;
 605                         openOutputFile(taxSummary, outTaxTree);
 606                         taxaSum.print(outTaxTree);
 607                         outTaxTree.close();
 608
 609                         //output taxonomy with the unclassified bins added
 610                         ifstream inTax;
 611                         openInputFile(newTaxonomyFile, inTax);
 612
 613                         ofstream outTax;
 614                         string unclass = newTaxonomyFile + ".unclass.temp";
 615                         openOutputFile(unclass, outTax);
 616
 617                         //get maxLevel from phylotree so you know how many 'unclassified's to add
 618                         int maxLevel = taxaSum.getMaxLevel();
 619
 620                         //read taxfile - this reading and rewriting is done to preserve the confidence scores.
 621                         string name, taxon;
 622                         while (!inTax.eof()) {
 623                                 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; }
 624
 625                                 inTax >> name >> taxon; gobble(inTax);
 626
 627                                 string newTax = addUnclassifieds(taxon, maxLevel);
 628
 629                                 outTax << name << '\t' << newTax << endl;
 630                         }
 631                         inTax.close();
 632                         outTax.close();
 633
 634                         remove(newTaxonomyFile.c_str());
 635                         rename(unclass.c_str(), newTaxonomyFile.c_str());
 636
 637                         m->mothurOutEndLine();
 638                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 639
 640                         #ifdef USE_MPI
 641                                 }
 642                         #endif
 643
 644                         m->mothurOutEndLine();
 645                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 646                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 647                         m->mothurOutEndLine();
 648                 }
 649
 650                 delete classify;
 651                 return 0;
 652         }
 653         catch(exception& e) {
 654                 m->errorOut(e, "ClassifySeqsCommand", "execute");
 655                 exit(1);
 656         }
 657 }
 658
 659 /**************************************************************************************************/
 660 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
 661         try{
 662                 string newTax, taxon;
 663                 int level = 0;
 664
 665                 //keep what you have counting the levels
 666                 while (tax.find_first_of(';') != -1) {
 667                         //get taxon
 668                         taxon = tax.substr(0,tax.find_first_of(';'))+';';
 669                         tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 670                         newTax += taxon;
 671                         level++;
 672                 }
 673
 674                 //add "unclassified" until you reach maxLevel
 675                 while (level < maxlevel) {
 676                         newTax += "unclassified;";
 677                         level++;
 678                 }
 679
 680                 return newTax;
 681         }
 682         catch(exception& e) {
 683                 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
 684                 exit(1);
 685         }
 686 }
 687
 688 /**************************************************************************************************/
 689
 690 void ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string filename) {
 691         try {
 692 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 693                 int process = 0;
 694                 //              processIDS.resize(0);
 695
 696                 //loop through and create all the processes you want
 697                 while (process != processors) {
 698                         int pid = fork();
 699
 700                         if (pid > 0) {
 701                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 702                                 process++;
 703                         }else if (pid == 0){
 704                                 driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", filename);
 705                                 exit(0);
 706                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 707                 }
 708
 709                 //force parent to wait until all the processes are done
 710                 for (int i=0;i<processors;i++) {
 711                         int temp = processIDS[i];
 712                         wait(&temp);
 713                 }
 714 #endif
 715         }
 716         catch(exception& e) {
 717                 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
 718                 exit(1);
 719         }
 720 }
 721 /**************************************************************************************************/
 722
 723 void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) {
 724         try{
 725
 726                 ofstream output;
 727                 ifstream input;
 728                 openOutputFileAppend(filename, output);
 729                 openInputFile(temp, input);
 730
 731                 while(char c = input.get()){
 732                         if(input.eof())         {       break;                  }
 733                         else                            {       output << c;    }
 734                 }
 735
 736                 input.close();
 737                 output.close();
 738         }
 739         catch(exception& e) {
 740                 m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles");
 741                 exit(1);
 742         }
 743 }
 744
 745 //**********************************************************************************************************************
 746
 747 int ClassifySeqsCommand::driver(linePair* line, string taxFName, string tempTFName, string filename){
 748         try {
 749                 ofstream outTax;
 750                 openOutputFile(taxFName, outTax);
 751
 752                 ofstream outTaxSimple;
 753                 openOutputFile(tempTFName, outTaxSimple);
 754
 755                 ifstream inFASTA;
 756                 openInputFile(filename, inFASTA);
 757
 758                 inFASTA.seekg(line->start);
 759
 760                 string taxonomy;
 761
 762                 for(int i=0;i<line->numSeqs;i++){
 763                         if (m->control_pressed) { return 0; }
 764
 765                         Sequence* candidateSeq = new Sequence(inFASTA);
 766
 767                         if (candidateSeq->getName() != "") {
 768                                 taxonomy = classify->getTaxonomy(candidateSeq);
 769
 770                                 if (m->control_pressed) { delete candidateSeq; return 0; }
 771
 772                                 if (taxonomy != "bad seq") {
 773                                         //output confidence scores or not
 774                                         if (probs) {
 775                                                 outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
 776                                         }else{
 777                                                 outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
 778                                         }
 779
 780                                         outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
 781                                 }
 782                         }
 783                         delete candidateSeq;
 784
 785                         if((i+1) % 100 == 0){
 786                                 m->mothurOut("Classifying sequence " + toString(i+1)); m->mothurOutEndLine();
 787                         }
 788                 }
 789
 790                 inFASTA.close();
 791                 outTax.close();
 792                 outTaxSimple.close();
 793
 794                 return 1;
 795         }
 796         catch(exception& e) {
 797                 m->errorOut(e, "ClassifySeqsCommand", "driver");
 798                 exit(1);
 799         }
 800 }
 801 //**********************************************************************************************************************
 802 #ifdef USE_MPI
 803 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, vector<long>& MPIPos){
 804         try {
 805                 MPI_Status statusNew;
 806                 MPI_Status statusTemp;
 807                 MPI_Status status;
 808
 809                 int pid;
 810                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 811
 812                 string taxonomy;
 813                 string outputString;
 814
 815                 for(int i=0;i<num;i++){
 816
 817                         if (m->control_pressed) { return 0; }
 818
 819                         //read next sequence
 820                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 821                         char* buf4 = new char[length];
 822                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 823
 824                         string tempBuf = buf4;
 825                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 826                         istringstream iss (tempBuf,istringstream::in);
 827                         delete buf4;
 828
 829                         Sequence* candidateSeq = new Sequence(iss);
 830
 831                         if (candidateSeq->getName() != "") {
 832                                 taxonomy = classify->getTaxonomy(candidateSeq);
 833
 834                                 if (taxonomy != "bad seq") {
 835                                         //output confidence scores or not
 836                                         if (probs) {
 837                                                 outputString =  candidateSeq->getName() + "\t" + taxonomy + "\n";
 838                                         }else{
 839                                                 outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
 840                                         }
 841
 842                                         int length = outputString.length();
 843                                         char* buf2 = new char[length];
 844                                         memcpy(buf2, outputString.c_str(), length);
 845
 846                                         MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
 847                                         delete buf2;
 848
 849                                         outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
 850                                         length = outputString.length();
 851                                         char* buf = new char[length];
 852                                         memcpy(buf, outputString.c_str(), length);
 853
 854                                         MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
 855                                         delete buf;
 856                                 }
 857                         }
 858                         delete candidateSeq;
 859
 860                         if((i+1) % 100 == 0){   cout << "Classifying sequence " << (i+1) << endl;       }
 861                 }
 862
 863                 if(num % 100 != 0){     cout << "Classifying sequence " << (num) << endl;       }
 864
 865
 866                 return 1;
 867         }
 868         catch(exception& e) {
 869                 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
 870                 exit(1);
 871         }
 872 }
 873
 874 //**********************************************************************************************************************
 875 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
 876         try {
 877
 878                 nameMap.clear(); //remove old names
 879
 880                 MPI_File inMPI;
 881                 MPI_Offset size;
 882                 MPI_Status status;
 883
 884                 //char* inFileName = new char[nameFilename.length()];
 885                 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
 886
 887                 char inFileName[1024];
 888                 strcpy(inFileName, nameFilename.c_str());
 889
 890                 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 891                 MPI_File_get_size(inMPI, &size);
 892                 //delete inFileName;
 893
 894                 char* buffer = new char[size];
 895                 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
 896
 897                 string tempBuf = buffer;
 898                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 899                 istringstream iss (tempBuf,istringstream::in);
 900                 delete buffer;
 901
 902                 string firstCol, secondCol;
 903                 while(!iss.eof()) {
 904                         iss >> firstCol >> secondCol; gobble(iss);
 905
 906                         vector<string> temp;
 907                         splitAtComma(secondCol, temp);
 908
 909                         nameMap[firstCol] = temp;
 910                 }
 911
 912                 MPI_File_close(&inMPI);
 913
 914                 return 1;
 915         }
 916         catch(exception& e) {
 917                 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
 918                 exit(1);
 919         }
 920 }
 921 #endif
 922 /**************************************************************************************************/