chimeracheckcommand.cpp

   1 /*
   2  *  chimeracheckcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/31/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeracheckcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> ChimeraCheckCommand::getValidParameters(){
  14         try {
  15                 string AlignArray[] =  {"fasta","processors","increment","template","ksize","svg", "name","outputdir","inputdir" };
  16                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "ChimeraCheckCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 vector<string> ChimeraCheckCommand::getRequiredParameters(){
  26         try {
  27                 string AlignArray[] =  {"template","fasta"};
  28                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  29                 return myArray;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "ChimeraCheckCommand", "getRequiredParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 vector<string> ChimeraCheckCommand::getRequiredFiles(){
  38         try {
  39                 vector<string> myArray;
  40                 return myArray;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "ChimeraCheckCommand", "getRequiredFiles");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 ChimeraCheckCommand::ChimeraCheckCommand(){
  49         try {
  50                 vector<string> tempOutNames;
  51                 outputTypes["chimera"] = tempOutNames;
  52         }
  53         catch(exception& e) {
  54                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  55                 exit(1);
  56         }
  57 }
  58 //***************************************************************************************************************
  59 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  60         try {
  61                 abort = false;
  62
  63                 //allow user to run help
  64                 if(option == "help") { help(); abort = true; }
  65
  66                 else {
  67                         //valid paramters for this command
  68                         string Array[] =  {"fasta","processors","increment","template","ksize","svg", "name","outputdir","inputdir" };
  69                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  70
  71                         OptionParser parser(option);
  72                         map<string,string> parameters = parser.getParameters();
  73
  74                         ValidParameters validParameter("chimera.check");
  75                         map<string,string>::iterator it;
  76
  77                         //check to make sure all parameters are valid for command
  78                         for (it = parameters.begin(); it != parameters.end(); it++) {
  79                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  80                         }
  81
  82                         vector<string> tempOutNames;
  83                         outputTypes["chimera"] = tempOutNames;
  84
  85                         //if the user changes the input directory command factory will send this info to us in the output parameter
  86                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  87                         if (inputDir == "not found"){   inputDir = "";          }
  88                         else {
  89                                 it = parameters.find("template");
  90                                 //user has given a template file
  91                                 if(it != parameters.end()){
  92                                         string path = m->hasPath(it->second);
  93                                         //if the user has not given a path then, add inputdir. else leave path alone.
  94                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  95                                 }
  96                         }
  97
  98                         //check for required parameters
  99                         fastafile = validParameter.validFile(parameters, "fasta", false);
 100                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 101                         else {
 102                                 m->splitAtDash(fastafile, fastaFileNames);
 103
 104                                 //go through files and make sure they are good, if not, then disregard them
 105                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 106                                         if (inputDir != "") {
 107                                                 string path = m->hasPath(fastaFileNames[i]);
 108                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 109                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 110                                         }
 111
 112                                         int ableToOpen;
 113                                         ifstream in;
 114
 115                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 116
 117                                         //if you can't open it, try default location
 118                                         if (ableToOpen == 1) {
 119                                                 if (m->getDefaultPath() != "") { //default path is set
 120                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 121                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 122                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 123                                                         fastaFileNames[i] = tryPath;
 124                                                 }
 125                                         }
 126
 127                                         //if you can't open it, try default location
 128                                         if (ableToOpen == 1) {
 129                                                 if (m->getOutputDir() != "") { //default path is set
 130                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 131                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 132                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 133                                                         fastaFileNames[i] = tryPath;
 134                                                 }
 135                                         }
 136
 137                                         in.close();
 138
 139                                         if (ableToOpen == 1) {
 140                                                 m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 141                                                 //erase from file list
 142                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 143                                                 i--;
 144                                         }
 145                                 }
 146
 147                                 //make sure there is at least one valid file left
 148                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 149                         }
 150
 151                         //if the user changes the output directory command factory will send this info to us in the output parameter
 152                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 153
 154                         templatefile = validParameter.validFile(parameters, "template", true);
 155                         if (templatefile == "not open") { abort = true; }
 156                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 157
 158                         namefile = validParameter.validFile(parameters, "name", false);
 159                         if (namefile == "not found") { namefile = ""; }
 160                         else {
 161                                 m->splitAtDash(namefile, nameFileNames);
 162
 163                                 //go through files and make sure they are good, if not, then disregard them
 164                                 for (int i = 0; i < nameFileNames.size(); i++) {
 165                                         if (inputDir != "") {
 166                                                 string path = m->hasPath(nameFileNames[i]);
 167                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 168                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 169                                         }
 170
 171                                         int ableToOpen;
 172                                         ifstream in;
 173
 174                                         ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 175
 176                                         //if you can't open it, try default location
 177                                         if (ableToOpen == 1) {
 178                                                 if (m->getDefaultPath() != "") { //default path is set
 179                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 180                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 181                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 182                                                         nameFileNames[i] = tryPath;
 183                                                 }
 184                                         }
 185
 186                                         //if you can't open it, try default location
 187                                         if (ableToOpen == 1) {
 188                                                 if (m->getOutputDir() != "") { //default path is set
 189                                                         string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 190                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 191                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 192                                                         nameFileNames[i] = tryPath;
 193                                                 }
 194                                         }
 195
 196                                         in.close();
 197
 198                                         if (ableToOpen == 1) {
 199                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 200                                                 //erase from file list
 201                                                 nameFileNames.erase(nameFileNames.begin()+i);
 202                                                 i--;
 203                                         }
 204
 205                                 }
 206
 207                                 //make sure there is at least one valid file left
 208                                 if (nameFileNames.size() != 0) {
 209                                         if (nameFileNames.size() != fastaFileNames.size()) {
 210                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 211                                                  abort = true;
 212                                         }
 213                                 }
 214                         }
 215
 216                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 217                         convert(temp, processors);
 218
 219                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 220                         convert(temp, ksize);
 221
 222                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 223                         svg = m->isTrue(temp);
 224                         if (nameFileNames.size() != 0) { svg = true; }
 225
 226                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 227                         convert(temp, increment);
 228                 }
 229         }
 230         catch(exception& e) {
 231                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 232                 exit(1);
 233         }
 234 }
 235 //**********************************************************************************************************************
 236
 237 void ChimeraCheckCommand::help(){
 238         try {
 239
 240                 m->mothurOut("The chimera.check command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 241                 m->mothurOut("This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n");
 242                 m->mothurOut("The chimera.check command parameters are fasta, template, processors, ksize, increment, svg and name.\n");
 243                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 244                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 245                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 246                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 247                 #ifdef USE_MPI
 248                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 249                 #endif
 250                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n");
 251                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7. \n");
 252                 m->mothurOut("The svg parameter allows you to specify whether or not you would like a svg file outputted for each query sequence, default is False.\n");
 253                 m->mothurOut("The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n");
 254                 m->mothurOut("You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n");
 255                 m->mothurOut("The chimera.check command should be in the following format: \n");
 256                 m->mothurOut("chimera.check(fasta=yourFastaFile, template=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n");
 257                 m->mothurOut("Example: chimera.check(fasta=AD.fasta, template=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n");
 258                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "ChimeraCheckCommand", "help");
 262                 exit(1);
 263         }
 264 }
 265
 266 //***************************************************************************************************************
 267
 268 ChimeraCheckCommand::~ChimeraCheckCommand(){    /*      do nothing      */      }
 269
 270 //***************************************************************************************************************
 271
 272 int ChimeraCheckCommand::execute(){
 273         try{
 274
 275                 if (abort == true) { return 0; }
 276
 277                 for (int i = 0; i < fastaFileNames.size(); i++) {
 278
 279                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 280
 281                         int start = time(NULL);
 282
 283                         string thisNameFile = "";
 284                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 285
 286                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 287
 288                         if (m->control_pressed) { delete chimera;       return 0;       }
 289
 290                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 291                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 292                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 293
 294                 #ifdef USE_MPI
 295
 296                                 int pid, end, numSeqsPerProcessor;
 297                                 int tag = 2001;
 298                                 vector<unsigned long int> MPIPos;
 299
 300                                 MPI_Status status;
 301                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 302                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 303
 304                                 MPI_File inMPI;
 305                                 MPI_File outMPI;
 306
 307                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 308                                 int inMode=MPI_MODE_RDONLY;
 309
 310                                 char outFilename[1024];
 311                                 strcpy(outFilename, outputFileName.c_str());
 312
 313                                 char inFileName[1024];
 314                                 strcpy(inFileName, fastaFileNames[i].c_str());
 315
 316                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 317                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 318
 319                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } outputTypes.clear(); delete chimera; return 0;  }
 320
 321                                 if (pid == 0) { //you are the root process
 322                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 323
 324                                         //send file positions to all processes
 325                                         for(int j = 1; j < processors; j++) {
 326                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 327                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 328                                         }
 329
 330                                         //figure out how many sequences you have to align
 331                                         numSeqsPerProcessor = numSeqs / processors;
 332                                         int startIndex =  pid * numSeqsPerProcessor;
 333                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 334
 335
 336                                         //align your part
 337                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 338
 339                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); }   outputTypes.clear(); delete chimera; return 0;  }
 340
 341                                         //wait on chidren
 342                                         for(int j = 1; j < processors; j++) {
 343                                                 char buf[5];
 344                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 345                                         }
 346                                 }else{ //you are a child process
 347                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 348                                         MPIPos.resize(numSeqs+1);
 349                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 350
 351                                         //figure out how many sequences you have to align
 352                                         numSeqsPerProcessor = numSeqs / processors;
 353                                         int startIndex =  pid * numSeqsPerProcessor;
 354                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 355
 356                                         //align your part
 357                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 358
 359                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  outputTypes.clear(); delete chimera; return 0;  }
 360
 361                                         //tell parent you are done.
 362                                         char buf[5];
 363                                         strcpy(buf, "done");
 364                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 365                                 }
 366
 367                                 //close files
 368                                 MPI_File_close(&inMPI);
 369                                 MPI_File_close(&outMPI);
 370                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 371                 #else
 372
 373                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 374
 375                         for (int s = 0; s < (positions.size()-1); s++) {
 376                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 377                         }
 378
 379                         //break up file
 380                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 381                                 if(processors == 1){
 382                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 383
 384                                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); } for (int j = 0; j < lines.size(); j++) {  delete lines[j];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 385
 386                                 }else{
 387                                         processIDS.resize(0);
 388
 389                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 390
 391                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 392
 393                                         //append output files
 394                                         for(int j=1;j<processors;j++){
 395                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 396                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 397                                         }
 398
 399                                         if (m->control_pressed) {
 400                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 401                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 402                                                 delete chimera;
 403                                                 return 0;
 404                                         }
 405                                 }
 406
 407                         #else
 408                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 409
 410                                 if (m->control_pressed) { for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear(); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear(); delete chimera; return 0; }
 411                         #endif
 412                 #endif
 413                         delete chimera;
 414                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 415
 416                         m->mothurOutEndLine(); m->mothurOut("This method does not determine if a sequence is chimeric, but allows you to make that determination based on the IS values."); m->mothurOutEndLine();
 417                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 418
 419                 }
 420
 421                 m->mothurOutEndLine();
 422                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 423                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 424                 m->mothurOutEndLine();
 425
 426                 return 0;
 427
 428         }
 429         catch(exception& e) {
 430                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 431                 exit(1);
 432         }
 433 }
 434 //**********************************************************************************************************************
 435
 436 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 437         try {
 438                 ofstream out;
 439                 m->openOutputFile(outputFName, out);
 440
 441                 ofstream out2;
 442
 443                 ifstream inFASTA;
 444                 m->openInputFile(filename, inFASTA);
 445
 446                 inFASTA.seekg(filePos->start);
 447
 448                 bool done = false;
 449                 int count = 0;
 450
 451                 while (!done) {
 452
 453                         if (m->control_pressed) {       return 1;       }
 454
 455                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 456
 457                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 458                                 //find chimeras
 459                                 chimera->getChimeras(candidateSeq);
 460
 461                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 462
 463                                 //print results
 464                                 chimera->print(out, out2);
 465                         }
 466                         delete candidateSeq;
 467
 468                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 469                                 unsigned long int pos = inFASTA.tellg();
 470                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 471                         #else
 472                                 if (inFASTA.eof()) { break; }
 473                         #endif
 474
 475                         //report progress
 476                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 477                 }
 478                 //report progress
 479                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 480
 481                 out.close();
 482                 inFASTA.close();
 483
 484                 return count;
 485         }
 486         catch(exception& e) {
 487                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 488                 exit(1);
 489         }
 490 }
 491 //**********************************************************************************************************************
 492 #ifdef USE_MPI
 493 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 494         try {
 495                 MPI_File outAccMPI;
 496                 MPI_Status status;
 497                 int pid;
 498                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 499
 500                 for(int i=0;i<num;i++){
 501
 502                         if (m->control_pressed) { return 0; }
 503
 504                         //read next sequence
 505                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 506
 507                         char* buf4 = new char[length];
 508                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 509
 510                         string tempBuf = buf4;
 511                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 512                         istringstream iss (tempBuf,istringstream::in);
 513                         delete buf4;
 514
 515                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 516
 517                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 518                                 //find chimeras
 519                                 chimera->getChimeras(candidateSeq);
 520
 521                                 //print results
 522                                 chimera->print(outMPI, outAccMPI);
 523                         }
 524                         delete candidateSeq;
 525
 526                         //report progress
 527                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 528                 }
 529                 //report progress
 530                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 531
 532                 return 0;
 533         }
 534         catch(exception& e) {
 535                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 536                 exit(1);
 537         }
 538 }
 539 #endif
 540
 541 /**************************************************************************************************/
 542
 543 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 544         try {
 545 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 546                 int process = 0;
 547                 int num = 0;
 548
 549                 //loop through and create all the processes you want
 550                 while (process != processors) {
 551                         int pid = fork();
 552
 553                         if (pid > 0) {
 554                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 555                                 process++;
 556                         }else if (pid == 0){
 557                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 558
 559                                 //pass numSeqs to parent
 560                                 ofstream out;
 561                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 562                                 m->openOutputFile(tempFile, out);
 563                                 out << num << endl;
 564                                 out.close();
 565
 566                                 exit(0);
 567                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 568                 }
 569
 570                 //force parent to wait until all the processes are done
 571                 for (int i=0;i<processors;i++) {
 572                         int temp = processIDS[i];
 573                         wait(&temp);
 574                 }
 575
 576                 for (int i = 0; i < processIDS.size(); i++) {
 577                         ifstream in;
 578                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 579                         m->openInputFile(tempFile, in);
 580                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 581                         in.close(); remove(tempFile.c_str());
 582                 }
 583
 584                 return num;
 585 #endif
 586         }
 587         catch(exception& e) {
 588                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 589                 exit(1);
 590         }
 591 }
 592 /**************************************************************************************************/
 593
 594