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                 abort = true; calledHelp = true;
  51                 vector<string> tempOutNames;
  52                 outputTypes["chimera"] = tempOutNames;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  56                 exit(1);
  57         }
  58 }
  59 //***************************************************************************************************************
  60 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  61         try {
  62                 abort = false; calledHelp = false;
  63
  64                 //allow user to run help
  65                 if(option == "help") { help(); abort = true; calledHelp = true; }
  66
  67                 else {
  68                         //valid paramters for this command
  69                         string Array[] =  {"fasta","processors","increment","template","ksize","svg", "name","outputdir","inputdir" };
  70                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  71
  72                         OptionParser parser(option);
  73                         map<string,string> parameters = parser.getParameters();
  74
  75                         ValidParameters validParameter("chimera.check");
  76                         map<string,string>::iterator it;
  77
  78                         //check to make sure all parameters are valid for command
  79                         for (it = parameters.begin(); it != parameters.end(); it++) {
  80                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  81                         }
  82
  83                         vector<string> tempOutNames;
  84                         outputTypes["chimera"] = tempOutNames;
  85
  86                         //if the user changes the input directory command factory will send this info to us in the output parameter
  87                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  88                         if (inputDir == "not found"){   inputDir = "";          }
  89                         else {
  90                                 it = parameters.find("template");
  91                                 //user has given a template file
  92                                 if(it != parameters.end()){
  93                                         string path = m->hasPath(it->second);
  94                                         //if the user has not given a path then, add inputdir. else leave path alone.
  95                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  96                                 }
  97                         }
  98
  99                         //check for required parameters
 100                         fastafile = validParameter.validFile(parameters, "fasta", false);
 101                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 102                         else {
 103                                 m->splitAtDash(fastafile, fastaFileNames);
 104
 105                                 //go through files and make sure they are good, if not, then disregard them
 106                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 107                                         if (inputDir != "") {
 108                                                 string path = m->hasPath(fastaFileNames[i]);
 109                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 110                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 111                                         }
 112
 113                                         int ableToOpen;
 114                                         ifstream in;
 115
 116                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 117
 118                                         //if you can't open it, try default location
 119                                         if (ableToOpen == 1) {
 120                                                 if (m->getDefaultPath() != "") { //default path is set
 121                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 122                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 123                                                         ifstream in2;
 124                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 125                                                         in2.close();
 126                                                         fastaFileNames[i] = tryPath;
 127                                                 }
 128                                         }
 129
 130                                         //if you can't open it, try default location
 131                                         if (ableToOpen == 1) {
 132                                                 if (m->getOutputDir() != "") { //default path is set
 133                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 134                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 135                                                         ifstream in2;
 136                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 137                                                         in2.close();
 138                                                         fastaFileNames[i] = tryPath;
 139                                                 }
 140                                         }
 141
 142                                         in.close();
 143
 144                                         if (ableToOpen == 1) {
 145                                                 m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 146                                                 //erase from file list
 147                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 148                                                 i--;
 149                                         }
 150                                 }
 151
 152                                 //make sure there is at least one valid file left
 153                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 154                         }
 155
 156                         //if the user changes the output directory command factory will send this info to us in the output parameter
 157                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 158
 159                         templatefile = validParameter.validFile(parameters, "template", true);
 160                         if (templatefile == "not open") { abort = true; }
 161                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 162
 163                         namefile = validParameter.validFile(parameters, "name", false);
 164                         if (namefile == "not found") { namefile = ""; }
 165                         else {
 166                                 m->splitAtDash(namefile, nameFileNames);
 167
 168                                 //go through files and make sure they are good, if not, then disregard them
 169                                 for (int i = 0; i < nameFileNames.size(); i++) {
 170                                         if (inputDir != "") {
 171                                                 string path = m->hasPath(nameFileNames[i]);
 172                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 173                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 174                                         }
 175
 176                                         int ableToOpen;
 177                                         ifstream in;
 178
 179                                         ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 180
 181                                         //if you can't open it, try default location
 182                                         if (ableToOpen == 1) {
 183                                                 if (m->getDefaultPath() != "") { //default path is set
 184                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 185                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 186                                                         ifstream in2;
 187                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 188                                                         in2.close();
 189                                                         nameFileNames[i] = tryPath;
 190                                                 }
 191                                         }
 192
 193                                         //if you can't open it, try default location
 194                                         if (ableToOpen == 1) {
 195                                                 if (m->getOutputDir() != "") { //default path is set
 196                                                         string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 197                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 198                                                         ifstream in2;
 199                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 200                                                         in2.close();
 201                                                         nameFileNames[i] = tryPath;
 202                                                 }
 203                                         }
 204
 205                                         in.close();
 206
 207                                         if (ableToOpen == 1) {
 208                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 209                                                 //erase from file list
 210                                                 nameFileNames.erase(nameFileNames.begin()+i);
 211                                                 i--;
 212                                         }
 213
 214                                 }
 215
 216                                 //make sure there is at least one valid file left
 217                                 if (nameFileNames.size() != 0) {
 218                                         if (nameFileNames.size() != fastaFileNames.size()) {
 219                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 220                                                  abort = true;
 221                                         }
 222                                 }
 223                         }
 224
 225                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 226                         convert(temp, processors);
 227
 228                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 229                         convert(temp, ksize);
 230
 231                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 232                         svg = m->isTrue(temp);
 233                         if (nameFileNames.size() != 0) { svg = true; }
 234
 235                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 236                         convert(temp, increment);
 237                 }
 238         }
 239         catch(exception& e) {
 240                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 241                 exit(1);
 242         }
 243 }
 244 //**********************************************************************************************************************
 245
 246 void ChimeraCheckCommand::help(){
 247         try {
 248
 249                 m->mothurOut("The chimera.check command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 250                 m->mothurOut("This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n");
 251                 m->mothurOut("The chimera.check command parameters are fasta, template, processors, ksize, increment, svg and name.\n");
 252                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 253                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 254                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 255                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 256                 #ifdef USE_MPI
 257                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 258                 #endif
 259                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n");
 260                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7. \n");
 261                 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");
 262                 m->mothurOut("The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n");
 263                 m->mothurOut("You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n");
 264                 m->mothurOut("The chimera.check command should be in the following format: \n");
 265                 m->mothurOut("chimera.check(fasta=yourFastaFile, template=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n");
 266                 m->mothurOut("Example: chimera.check(fasta=AD.fasta, template=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n");
 267                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 268         }
 269         catch(exception& e) {
 270                 m->errorOut(e, "ChimeraCheckCommand", "help");
 271                 exit(1);
 272         }
 273 }
 274
 275 //***************************************************************************************************************
 276
 277 ChimeraCheckCommand::~ChimeraCheckCommand(){    /*      do nothing      */      }
 278
 279 //***************************************************************************************************************
 280
 281 int ChimeraCheckCommand::execute(){
 282         try{
 283
 284                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 285
 286                 for (int i = 0; i < fastaFileNames.size(); i++) {
 287
 288                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 289
 290                         int start = time(NULL);
 291
 292                         string thisNameFile = "";
 293                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 294
 295                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 296
 297                         if (m->control_pressed) { delete chimera;       return 0;       }
 298
 299                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 300                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 301                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 302
 303                 #ifdef USE_MPI
 304
 305                                 int pid, numSeqsPerProcessor;
 306                                 int tag = 2001;
 307                                 vector<unsigned long int> MPIPos;
 308
 309                                 MPI_Status status;
 310                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 311                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 312
 313                                 MPI_File inMPI;
 314                                 MPI_File outMPI;
 315
 316                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 317                                 int inMode=MPI_MODE_RDONLY;
 318
 319                                 char outFilename[1024];
 320                                 strcpy(outFilename, outputFileName.c_str());
 321
 322                                 char inFileName[1024];
 323                                 strcpy(inFileName, fastaFileNames[i].c_str());
 324
 325                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 326                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 327
 328                                 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;  }
 329
 330                                 if (pid == 0) { //you are the root process
 331                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 332
 333                                         //send file positions to all processes
 334                                         for(int j = 1; j < processors; j++) {
 335                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 336                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 337                                         }
 338
 339                                         //figure out how many sequences you have to align
 340                                         numSeqsPerProcessor = numSeqs / processors;
 341                                         int startIndex =  pid * numSeqsPerProcessor;
 342                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 343
 344
 345                                         //align your part
 346                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 347
 348                                         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;  }
 349
 350                                         //wait on chidren
 351                                         for(int j = 1; j < processors; j++) {
 352                                                 char buf[5];
 353                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 354                                         }
 355                                 }else{ //you are a child process
 356                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 357                                         MPIPos.resize(numSeqs+1);
 358                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 359
 360                                         //figure out how many sequences you have to align
 361                                         numSeqsPerProcessor = numSeqs / processors;
 362                                         int startIndex =  pid * numSeqsPerProcessor;
 363                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 364
 365                                         //align your part
 366                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 367
 368                                         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;  }
 369
 370                                         //tell parent you are done.
 371                                         char buf[5];
 372                                         strcpy(buf, "done");
 373                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 374                                 }
 375
 376                                 //close files
 377                                 MPI_File_close(&inMPI);
 378                                 MPI_File_close(&outMPI);
 379                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 380                 #else
 381
 382                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 383
 384                         for (int s = 0; s < (positions.size()-1); s++) {
 385                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 386                         }
 387
 388                         //break up file
 389                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 390                                 if(processors == 1){
 391                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 392
 393                                         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; }
 394
 395                                 }else{
 396                                         processIDS.resize(0);
 397
 398                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 399
 400                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 401
 402                                         //append output files
 403                                         for(int j=1;j<processors;j++){
 404                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 405                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 406                                         }
 407
 408                                         if (m->control_pressed) {
 409                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 410                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 411                                                 delete chimera;
 412                                                 return 0;
 413                                         }
 414                                 }
 415
 416                         #else
 417                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 418
 419                                 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; }
 420                         #endif
 421                 #endif
 422                         delete chimera;
 423                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 424
 425                         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();
 426                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 427
 428                 }
 429
 430                 m->mothurOutEndLine();
 431                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 432                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 433                 m->mothurOutEndLine();
 434
 435                 return 0;
 436
 437         }
 438         catch(exception& e) {
 439                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 440                 exit(1);
 441         }
 442 }
 443 //**********************************************************************************************************************
 444
 445 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 446         try {
 447                 ofstream out;
 448                 m->openOutputFile(outputFName, out);
 449
 450                 ofstream out2;
 451
 452                 ifstream inFASTA;
 453                 m->openInputFile(filename, inFASTA);
 454
 455                 inFASTA.seekg(filePos->start);
 456
 457                 bool done = false;
 458                 int count = 0;
 459
 460                 while (!done) {
 461
 462                         if (m->control_pressed) {       return 1;       }
 463
 464                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 465
 466                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 467                                 //find chimeras
 468                                 chimera->getChimeras(candidateSeq);
 469
 470                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 471
 472                                 //print results
 473                                 chimera->print(out, out2);
 474                         }
 475                         delete candidateSeq;
 476
 477                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 478                                 unsigned long int pos = inFASTA.tellg();
 479                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 480                         #else
 481                                 if (inFASTA.eof()) { break; }
 482                         #endif
 483
 484                         //report progress
 485                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 486                 }
 487                 //report progress
 488                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 489
 490                 out.close();
 491                 inFASTA.close();
 492
 493                 return count;
 494         }
 495         catch(exception& e) {
 496                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 497                 exit(1);
 498         }
 499 }
 500 //**********************************************************************************************************************
 501 #ifdef USE_MPI
 502 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 503         try {
 504                 MPI_File outAccMPI;
 505                 MPI_Status status;
 506                 int pid;
 507                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 508
 509                 for(int i=0;i<num;i++){
 510
 511                         if (m->control_pressed) { return 0; }
 512
 513                         //read next sequence
 514                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 515
 516                         char* buf4 = new char[length];
 517                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 518
 519                         string tempBuf = buf4;
 520                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 521                         istringstream iss (tempBuf,istringstream::in);
 522                         delete buf4;
 523
 524                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 525
 526                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 527                                 //find chimeras
 528                                 chimera->getChimeras(candidateSeq);
 529
 530                                 //print results
 531                                 chimera->print(outMPI, outAccMPI);
 532                         }
 533                         delete candidateSeq;
 534
 535                         //report progress
 536                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 537                 }
 538                 //report progress
 539                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 540
 541                 return 0;
 542         }
 543         catch(exception& e) {
 544                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 545                 exit(1);
 546         }
 547 }
 548 #endif
 549
 550 /**************************************************************************************************/
 551
 552 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 553         try {
 554 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 555                 int process = 0;
 556                 int num = 0;
 557
 558                 //loop through and create all the processes you want
 559                 while (process != processors) {
 560                         int pid = fork();
 561
 562                         if (pid > 0) {
 563                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 564                                 process++;
 565                         }else if (pid == 0){
 566                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 567
 568                                 //pass numSeqs to parent
 569                                 ofstream out;
 570                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 571                                 m->openOutputFile(tempFile, out);
 572                                 out << num << endl;
 573                                 out.close();
 574
 575                                 exit(0);
 576                         }else {
 577                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 578                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 579                                 exit(0);
 580                         }
 581                 }
 582
 583                 //force parent to wait until all the processes are done
 584                 for (int i=0;i<processors;i++) {
 585                         int temp = processIDS[i];
 586                         wait(&temp);
 587                 }
 588
 589                 for (int i = 0; i < processIDS.size(); i++) {
 590                         ifstream in;
 591                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 592                         m->openInputFile(tempFile, in);
 593                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 594                         in.close(); remove(tempFile.c_str());
 595                 }
 596
 597                 return num;
 598 #endif
 599         }
 600         catch(exception& e) {
 601                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 602                 exit(1);
 603         }
 604 }
 605 /**************************************************************************************************/
 606
 607