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::setParameters(){
  14         try {
  15                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  18                 CommandParameter psvg("svg", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psvg);
  19                 CommandParameter pincrement("increment", "Number", "", "10", "", "", "",false,false); parameters.push_back(pincrement);
  20                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  21                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ChimeraCheckCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraCheckCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.check command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  39                 helpString += "This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n";
  40                 helpString += "The chimera.check command parameters are fasta, reference, processors, ksize, increment, svg and name.\n";
  41                 helpString += "The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required unless you have a valid current fasta file. \n";
  42                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  43                 helpString += "The reference parameter allows you to enter a reference file containing known non-chimeric sequences, and is required. \n";
  44                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  45 #ifdef USE_MPI
  46                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  47 #endif
  48                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n";
  49                 helpString += "The ksize parameter allows you to input kmersize, default is 7. \n";
  50                 helpString += "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";
  51                 helpString += "The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n";
  52                 helpString += "You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n";
  53                 helpString += "The chimera.check command should be in the following format: \n";
  54                 helpString += "chimera.check(fasta=yourFastaFile, reference=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n";
  55                 helpString += "Example: chimera.check(fasta=AD.fasta, reference=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  57                 return helpString;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "ChimeraCheckCommand", "getHelpString");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 ChimeraCheckCommand::ChimeraCheckCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["chimera"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  74                 exit(1);
  75         }
  76 }
  77 //***************************************************************************************************************
  78 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84
  85                 else {
  86                         vector<string> myArray = setParameters();
  87
  88                         OptionParser parser(option);
  89                         map<string,string> parameters = parser.getParameters();
  90
  91                         ValidParameters validParameter("chimera.check");
  92                         map<string,string>::iterator it;
  93
  94                         //check to make sure all parameters are valid for command
  95                         for (it = parameters.begin(); it != parameters.end(); it++) {
  96                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  97                         }
  98
  99                         vector<string> tempOutNames;
 100                         outputTypes["chimera"] = tempOutNames;
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 it = parameters.find("reference");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         string path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 112                                 }
 113                         }
 114
 115                         //check for required parameters
 116                         fastafile = validParameter.validFile(parameters, "fasta", false);
 117                         if (fastafile == "not found") {
 118                                 //if there is a current fasta file, use it
 119                                 string filename = m->getFastaFile();
 120                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 121                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 122                         }else {
 123                                 m->splitAtDash(fastafile, fastaFileNames);
 124
 125                                 //go through files and make sure they are good, if not, then disregard them
 126                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 127                                         if (inputDir != "") {
 128                                                 string path = m->hasPath(fastaFileNames[i]);
 129                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 130                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 131                                         }
 132
 133                                         int ableToOpen;
 134                                         ifstream in;
 135
 136                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 137
 138                                         //if you can't open it, try default location
 139                                         if (ableToOpen == 1) {
 140                                                 if (m->getDefaultPath() != "") { //default path is set
 141                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 142                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 143                                                         ifstream in2;
 144                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 145                                                         in2.close();
 146                                                         fastaFileNames[i] = tryPath;
 147                                                 }
 148                                         }
 149
 150                                         //if you can't open it, try default location
 151                                         if (ableToOpen == 1) {
 152                                                 if (m->getOutputDir() != "") { //default path is set
 153                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 154                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 155                                                         ifstream in2;
 156                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 157                                                         in2.close();
 158                                                         fastaFileNames[i] = tryPath;
 159                                                 }
 160                                         }
 161
 162                                         in.close();
 163
 164                                         if (ableToOpen == 1) {
 165                                                 m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 166                                                 //erase from file list
 167                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 168                                                 i--;
 169                                         }
 170                                 }
 171
 172                                 //make sure there is at least one valid file left
 173                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 174                         }
 175
 176                         //if the user changes the output directory command factory will send this info to us in the output parameter
 177                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 178
 179                         templatefile = validParameter.validFile(parameters, "reference", true);
 180                         if (templatefile == "not open") { abort = true; }
 181                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("reference is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 182
 183                         namefile = validParameter.validFile(parameters, "name", false);
 184                         if (namefile == "not found") { namefile = ""; }
 185                         else {
 186                                 m->splitAtDash(namefile, nameFileNames);
 187
 188                                 //go through files and make sure they are good, if not, then disregard them
 189                                 for (int i = 0; i < nameFileNames.size(); i++) {
 190                                         if (inputDir != "") {
 191                                                 string path = m->hasPath(nameFileNames[i]);
 192                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 193                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 194                                         }
 195
 196                                         int ableToOpen;
 197                                         ifstream in;
 198
 199                                         ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 200
 201                                         //if you can't open it, try default location
 202                                         if (ableToOpen == 1) {
 203                                                 if (m->getDefaultPath() != "") { //default path is set
 204                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 205                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 206                                                         ifstream in2;
 207                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 208                                                         in2.close();
 209                                                         nameFileNames[i] = tryPath;
 210                                                 }
 211                                         }
 212
 213                                         //if you can't open it, try default location
 214                                         if (ableToOpen == 1) {
 215                                                 if (m->getOutputDir() != "") { //default path is set
 216                                                         string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 217                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 218                                                         ifstream in2;
 219                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 220                                                         in2.close();
 221                                                         nameFileNames[i] = tryPath;
 222                                                 }
 223                                         }
 224
 225                                         in.close();
 226
 227                                         if (ableToOpen == 1) {
 228                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 229                                                 //erase from file list
 230                                                 nameFileNames.erase(nameFileNames.begin()+i);
 231                                                 i--;
 232                                         }
 233
 234                                 }
 235
 236                                 //make sure there is at least one valid file left
 237                                 if (nameFileNames.size() != 0) {
 238                                         if (nameFileNames.size() != fastaFileNames.size()) {
 239                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 240                                                  abort = true;
 241                                         }
 242                                 }
 243                         }
 244
 245                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 246                         m->setProcessors(temp);
 247                         convert(temp, processors);
 248
 249                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 250                         convert(temp, ksize);
 251
 252                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 253                         svg = m->isTrue(temp);
 254                         if (nameFileNames.size() != 0) { svg = true; }
 255
 256                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 257                         convert(temp, increment);
 258                 }
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 262                 exit(1);
 263         }
 264 }
 265 //***************************************************************************************************************
 266
 267 int ChimeraCheckCommand::execute(){
 268         try{
 269
 270                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 271
 272                 for (int i = 0; i < fastaFileNames.size(); i++) {
 273
 274                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 275
 276                         int start = time(NULL);
 277
 278                         string thisNameFile = "";
 279                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 280
 281                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 282
 283                         if (m->control_pressed) { delete chimera;       return 0;       }
 284
 285                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 286                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 287                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 288
 289                 #ifdef USE_MPI
 290
 291                                 int pid, numSeqsPerProcessor;
 292                                 int tag = 2001;
 293                                 vector<unsigned long int> MPIPos;
 294
 295                                 MPI_Status status;
 296                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 297                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 298
 299                                 MPI_File inMPI;
 300                                 MPI_File outMPI;
 301
 302                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 303                                 int inMode=MPI_MODE_RDONLY;
 304
 305                                 char outFilename[1024];
 306                                 strcpy(outFilename, outputFileName.c_str());
 307
 308                                 char inFileName[1024];
 309                                 strcpy(inFileName, fastaFileNames[i].c_str());
 310
 311                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 312                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 313
 314                                 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;  }
 315
 316                                 if (pid == 0) { //you are the root process
 317                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 318
 319                                         //send file positions to all processes
 320                                         for(int j = 1; j < processors; j++) {
 321                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 322                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 323                                         }
 324
 325                                         //figure out how many sequences you have to align
 326                                         numSeqsPerProcessor = numSeqs / processors;
 327                                         int startIndex =  pid * numSeqsPerProcessor;
 328                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 329
 330
 331                                         //align your part
 332                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 333
 334                                         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;  }
 335
 336                                         //wait on chidren
 337                                         for(int j = 1; j < processors; j++) {
 338                                                 char buf[5];
 339                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 340                                         }
 341                                 }else{ //you are a child process
 342                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 343                                         MPIPos.resize(numSeqs+1);
 344                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 345
 346                                         //figure out how many sequences you have to align
 347                                         numSeqsPerProcessor = numSeqs / processors;
 348                                         int startIndex =  pid * numSeqsPerProcessor;
 349                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 350
 351                                         //align your part
 352                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 353
 354                                         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;  }
 355
 356                                         //tell parent you are done.
 357                                         char buf[5];
 358                                         strcpy(buf, "done");
 359                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 360                                 }
 361
 362                                 //close files
 363                                 MPI_File_close(&inMPI);
 364                                 MPI_File_close(&outMPI);
 365                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 366                 #else
 367
 368                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 369
 370                         for (int s = 0; s < (positions.size()-1); s++) {
 371                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 372                         }
 373
 374                         //break up file
 375                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 376                                 if(processors == 1){
 377                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 378
 379                                         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; }
 380
 381                                 }else{
 382                                         processIDS.resize(0);
 383
 384                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 385
 386                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 387
 388                                         //append output files
 389                                         for(int j=1;j<processors;j++){
 390                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 391                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 392                                         }
 393
 394                                         if (m->control_pressed) {
 395                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 396                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 397                                                 delete chimera;
 398                                                 return 0;
 399                                         }
 400                                 }
 401
 402                         #else
 403                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 404
 405                                 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; }
 406                         #endif
 407                 #endif
 408                         delete chimera;
 409                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 410
 411                         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();
 412                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 413
 414                 }
 415
 416                 m->mothurOutEndLine();
 417                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 418                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 419                 m->mothurOutEndLine();
 420
 421                 return 0;
 422
 423         }
 424         catch(exception& e) {
 425                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 426                 exit(1);
 427         }
 428 }
 429 //**********************************************************************************************************************
 430
 431 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 432         try {
 433                 ofstream out;
 434                 m->openOutputFile(outputFName, out);
 435
 436                 ofstream out2;
 437
 438                 ifstream inFASTA;
 439                 m->openInputFile(filename, inFASTA);
 440
 441                 inFASTA.seekg(filePos->start);
 442
 443                 bool done = false;
 444                 int count = 0;
 445
 446                 while (!done) {
 447
 448                         if (m->control_pressed) {       return 1;       }
 449
 450                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 451
 452                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 453                                 //find chimeras
 454                                 chimera->getChimeras(candidateSeq);
 455
 456                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 457
 458                                 //print results
 459                                 chimera->print(out, out2);
 460                         }
 461                         delete candidateSeq;
 462
 463                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 464                                 unsigned long int pos = inFASTA.tellg();
 465                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 466                         #else
 467                                 if (inFASTA.eof()) { break; }
 468                         #endif
 469
 470                         //report progress
 471                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 472                 }
 473                 //report progress
 474                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 475
 476                 out.close();
 477                 inFASTA.close();
 478
 479                 return count;
 480         }
 481         catch(exception& e) {
 482                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 483                 exit(1);
 484         }
 485 }
 486 //**********************************************************************************************************************
 487 #ifdef USE_MPI
 488 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 489         try {
 490                 MPI_File outAccMPI;
 491                 MPI_Status status;
 492                 int pid;
 493                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 494
 495                 for(int i=0;i<num;i++){
 496
 497                         if (m->control_pressed) { return 0; }
 498
 499                         //read next sequence
 500                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 501
 502                         char* buf4 = new char[length];
 503                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 504
 505                         string tempBuf = buf4;
 506                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 507                         istringstream iss (tempBuf,istringstream::in);
 508                         delete buf4;
 509
 510                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 511
 512                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 513                                 //find chimeras
 514                                 chimera->getChimeras(candidateSeq);
 515
 516                                 //print results
 517                                 chimera->print(outMPI, outAccMPI);
 518                         }
 519                         delete candidateSeq;
 520
 521                         //report progress
 522                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 523                 }
 524                 //report progress
 525                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 526
 527                 return 0;
 528         }
 529         catch(exception& e) {
 530                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 531                 exit(1);
 532         }
 533 }
 534 #endif
 535
 536 /**************************************************************************************************/
 537
 538 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 539         try {
 540 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 541                 int process = 0;
 542                 int num = 0;
 543
 544                 //loop through and create all the processes you want
 545                 while (process != processors) {
 546                         int pid = fork();
 547
 548                         if (pid > 0) {
 549                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 550                                 process++;
 551                         }else if (pid == 0){
 552                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 553
 554                                 //pass numSeqs to parent
 555                                 ofstream out;
 556                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 557                                 m->openOutputFile(tempFile, out);
 558                                 out << num << endl;
 559                                 out.close();
 560
 561                                 exit(0);
 562                         }else {
 563                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 564                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 565                                 exit(0);
 566                         }
 567                 }
 568
 569                 //force parent to wait until all the processes are done
 570                 for (int i=0;i<processors;i++) {
 571                         int temp = processIDS[i];
 572                         wait(&temp);
 573                 }
 574
 575                 for (int i = 0; i < processIDS.size(); i++) {
 576                         ifstream in;
 577                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 578                         m->openInputFile(tempFile, in);
 579                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 580                         in.close(); remove(tempFile.c_str());
 581                 }
 582
 583                 return num;
 584 #endif
 585         }
 586         catch(exception& e) {
 587                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 588                 exit(1);
 589         }
 590 }
 591 /**************************************************************************************************/
 592
 593