chimeraslayercommand.cpp

   1 /*
   2  *  chimeraslayercommand.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 "chimeraslayercommand.h"
  11 #include "bellerophon.h"
  12 #include "pintail.h"
  13 #include "ccode.h"
  14 #include "chimeracheckrdp.h"
  15 #include "chimeraslayer.h"
  16
  17
  18 //***************************************************************************************************************
  19
  20 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
  21         try {
  22                 abort = false;
  23
  24                 //allow user to run help
  25                 if(option == "help") { help(); abort = true; }
  26
  27                 else {
  28                         //valid paramters for this command
  29                         string Array[] =  {"fasta", "processors", "window", "template","numwanted", "ksize", "match","mismatch",
  30                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  31                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  32
  33                         OptionParser parser(option);
  34                         map<string,string> parameters = parser.getParameters();
  35
  36                         ValidParameters validParameter;
  37                         map<string,string>::iterator it;
  38
  39                         //check to make sure all parameters are valid for command
  40                         for (it = parameters.begin(); it != parameters.end(); it++) {
  41                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  42                         }
  43
  44                         //if the user changes the input directory command factory will send this info to us in the output parameter
  45                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  46                         if (inputDir == "not found"){   inputDir = "";          }
  47                         else {
  48                                 string path;
  49                                 it = parameters.find("fasta");
  50                                 //user has given a template file
  51                                 if(it != parameters.end()){
  52                                         path = hasPath(it->second);
  53                                         //if the user has not given a path then, add inputdir. else leave path alone.
  54                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  55                                 }
  56
  57                                 it = parameters.find("template");
  58                                 //user has given a template file
  59                                 if(it != parameters.end()){
  60                                         path = hasPath(it->second);
  61                                         //if the user has not given a path then, add inputdir. else leave path alone.
  62                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  63                                 }
  64                         }
  65
  66
  67                         //check for required parameters
  68                         fastafile = validParameter.validFile(parameters, "fasta", true);
  69                         if (fastafile == "not open") { abort = true; }
  70                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
  71
  72                         //if the user changes the output directory command factory will send this info to us in the output parameter
  73                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  74                                 outputDir = "";
  75                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
  76                         }
  77
  78                         templatefile = validParameter.validFile(parameters, "template", true);
  79                         if (templatefile == "not open") { abort = true; }
  80                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
  81
  82                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
  83                         convert(temp, processors);
  84
  85                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
  86                         convert(temp, ksize);
  87
  88                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
  89                         convert(temp, window);
  90
  91                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
  92                         convert(temp, match);
  93
  94                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
  95                         convert(temp, mismatch);
  96
  97                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
  98                         convert(temp, divR);
  99
 100                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 101                         convert(temp, minSimilarity);
 102
 103                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 104                         convert(temp, minCoverage);
 105
 106                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 107                         convert(temp, minBS);
 108
 109                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "10"; }
 110                         convert(temp, minSNP);
 111
 112                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 113                         convert(temp, parents);
 114
 115                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 116                         realign = isTrue(temp);
 117
 118                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 119
 120                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "100"; }
 121                         convert(temp, iters);
 122
 123                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 124                         convert(temp, increment);
 125
 126                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 127                         convert(temp, numwanted);
 128
 129                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 130                 }
 131         }
 132         catch(exception& e) {
 133                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 134                 exit(1);
 135         }
 136 }
 137 //**********************************************************************************************************************
 138
 139 void ChimeraSlayerCommand::help(){
 140         try {
 141
 142                 m->mothurOut("The chimera.slayer command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 143                 m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
 144                 m->mothurOut("The chimera.slayer command parameters are fasta, template, filter, mask, processors, ksize, window, match, mismatch, divergence. minsim, mincov, minbs, minsnp, parents, search, iters, increment and numwanted.\n"); //realign,
 145                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 146                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 147                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 148                 #ifdef USE_MPI
 149                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 150                 #endif
 151                 m->mothurOut("The mask parameter allows you to specify a file containing one sequence you wish to use as a mask for the your sequences. \n");
 152                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=50. \n");
 153                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n");
 154                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n");
 155                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n");
 156                 m->mothurOut("The match parameter allows you to reward matched bases in blast search, default is 5. \n");
 157                 m->mothurOut("The parents parameter allows you to select the number of potential parents to investigate from the numwanted best matches after rating them, default is 3. \n");
 158                 m->mothurOut("The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n");
 159                 m->mothurOut("The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n");
 160                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=100.\n");
 161                 m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
 162                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 163                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 164                 m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 10) \n");
 165                 m->mothurOut("The search parameter allows you to specify search method for finding the closest parent. Choices are distance, blast, and kmer, default distance. \n");
 166                 //m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false. Found to make results worse. \n");
 167                 m->mothurOut("NOT ALL PARAMETERS ARE USED BY ALL METHODS. Please look below for method specifics.\n\n");
 168                 m->mothurOut("The chimera.slayer command should be in the following format: \n");
 169                 m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
 170                 m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
 171                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "ChimeraSlayerCommand", "help");
 175                 exit(1);
 176         }
 177 }
 178
 179 //***************************************************************************************************************
 180
 181 ChimeraSlayerCommand::~ChimeraSlayerCommand(){  /*      do nothing      */      }
 182
 183 //***************************************************************************************************************
 184
 185 int ChimeraSlayerCommand::execute(){
 186         try{
 187
 188                 if (abort == true) { return 0; }
 189
 190                 int start = time(NULL);
 191
 192                 chimera = new ChimeraSlayer(fastafile, templatefile, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 193
 194                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "slayer.chimeras";
 195                 string accnosFileName = outputDir + getRootName(getSimpleName(fastafile))  + "slayer.accnos";
 196                 bool hasAccnos = true;
 197
 198                 if (m->control_pressed) { delete chimera;       return 0;       }
 199
 200                 if (chimera->getUnaligned()) {
 201                         m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 202                         delete chimera;
 203                         return 0;
 204                 }
 205                 templateSeqsLength = chimera->getLength();
 206
 207         #ifdef USE_MPI
 208                 int pid, end, numSeqsPerProcessor;
 209                         int tag = 2001;
 210                         vector<long> MPIPos;
 211                         MPIWroteAccnos = false;
 212
 213                         MPI_Status status;
 214                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 215                         MPI_Comm_size(MPI_COMM_WORLD, &processors);
 216
 217                         MPI_File inMPI;
 218                         MPI_File outMPI;
 219                         MPI_File outMPIAccnos;
 220
 221                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 222                         int inMode=MPI_MODE_RDONLY;
 223
 224                         char outFilename[outputFileName.length()];
 225                         strcpy(outFilename, outputFileName.c_str());
 226
 227                         char outAccnosFilename[accnosFileName.length()];
 228                         strcpy(outAccnosFilename, accnosFileName.c_str());
 229
 230                         char inFileName[fastafile.length()];
 231                         strcpy(inFileName, fastafile.c_str());
 232
 233                         MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 234                         MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 235                         MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 236
 237                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  delete chimera; return 0;  }
 238
 239
 240                         if (pid == 0) { //you are the root process
 241                                 m->mothurOutEndLine();
 242                                 m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 243                                 m->mothurOutEndLine();
 244
 245                                 string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 246
 247                                 //print header
 248                                 int length = outTemp.length();
 249                                 char buf2[length];
 250                                 strcpy(buf2, outTemp.c_str());
 251                                 MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 252
 253                                 MPIPos = setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
 254
 255                                 //send file positions to all processes
 256                                 MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD);  //send numSeqs
 257                                 MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
 258
 259                                 //figure out how many sequences you have to align
 260                                 numSeqsPerProcessor = numSeqs / processors;
 261                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 262                                 int startIndex =  pid * numSeqsPerProcessor;
 263
 264                                 //align your part
 265                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 266
 267                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  delete chimera; return 0;  }
 268
 269                                 for (int i = 1; i < processors; i++) {
 270                                         bool tempResult;
 271                                         MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 272                                         if (tempResult != 0) { MPIWroteAccnos = true; }
 273                                 }
 274                         }else{ //you are a child process
 275                                 MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
 276                                 MPIPos.resize(numSeqs+1);
 277                                 MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
 278
 279                                 //figure out how many sequences you have to align
 280                                 numSeqsPerProcessor = numSeqs / processors;
 281                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 282                                 int startIndex =  pid * numSeqsPerProcessor;
 283
 284                                 //align your part
 285                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 286
 287                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  delete chimera; return 0;  }
 288
 289                                 MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 290                         }
 291
 292                         //close files
 293                         MPI_File_close(&inMPI);
 294                         MPI_File_close(&outMPI);
 295                         MPI_File_close(&outMPIAccnos);
 296
 297                         //delete accnos file if blank
 298                         if (pid == 0) {
 299                                 if (!MPIWroteAccnos) {
 300                                         //MPI_Info info;
 301                                         //MPI_File_delete(outAccnosFilename, info);
 302                                         hasAccnos = false;
 303                                         remove(accnosFileName.c_str());
 304                                 }
 305                         }
 306
 307         #else
 308                 ofstream outHeader;
 309                 string tempHeader = outputDir + getRootName(getSimpleName(fastafile)) + "slayer.chimeras.tempHeader";
 310                 openOutputFile(tempHeader, outHeader);
 311
 312                 chimera->printHeader(outHeader);
 313                 outHeader.close();
 314
 315                 //break up file
 316                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 317                         if(processors == 1){
 318                                 ifstream inFASTA;
 319                                 openInputFile(fastafile, inFASTA);
 320                                 numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 321                                 inFASTA.close();
 322
 323                                 lines.push_back(new linePair(0, numSeqs));
 324
 325                                 driver(lines[0], outputFileName, fastafile, accnosFileName);
 326
 327                                 if (m->control_pressed) {
 328                                         remove(outputFileName.c_str());
 329                                         remove(tempHeader.c_str());
 330                                         remove(accnosFileName.c_str());
 331                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 332                                         delete chimera;
 333                                         return 0;
 334                                 }
 335
 336                                 //delete accnos file if its blank
 337                                 if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 338
 339                         }else{
 340                                 vector<int> positions;
 341                                 processIDS.resize(0);
 342
 343                                 ifstream inFASTA;
 344                                 openInputFile(fastafile, inFASTA);
 345
 346                                 string input;
 347                                 while(!inFASTA.eof()){
 348                                         input = getline(inFASTA);
 349                                         if (input.length() != 0) {
 350                                                 if(input[0] == '>'){    long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);  }
 351                                         }
 352                                 }
 353                                 inFASTA.close();
 354
 355                                 numSeqs = positions.size();
 356
 357                                 int numSeqsPerProcessor = numSeqs / processors;
 358
 359                                 for (int i = 0; i < processors; i++) {
 360                                         long int startPos = positions[ i * numSeqsPerProcessor ];
 361                                         if(i == processors - 1){
 362                                                 numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
 363                                         }
 364                                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 365                                 }
 366
 367
 368                                 createProcesses(outputFileName, fastafile, accnosFileName);
 369
 370                                 rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 371
 372                                 //append output files
 373                                 for(int i=1;i<processors;i++){
 374                                         appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 375                                         remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 376                                 }
 377
 378                                 vector<string> nonBlankAccnosFiles;
 379                                 //delete blank accnos files generated with multiple processes
 380                                 for(int i=0;i<processors;i++){
 381                                         if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
 382                                                 nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
 383                                         }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());  }
 384                                 }
 385
 386                                 //append accnos files
 387                                 if (nonBlankAccnosFiles.size() != 0) {
 388                                         rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
 389
 390                                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 391                                                 appendFiles(nonBlankAccnosFiles[h], accnosFileName);
 392                                                 remove(nonBlankAccnosFiles[h].c_str());
 393                                         }
 394                                 }else{ hasAccnos = false;  }
 395
 396                                 if (m->control_pressed) {
 397                                         remove(outputFileName.c_str());
 398                                         remove(accnosFileName.c_str());
 399                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 400                                         delete chimera;
 401                                         return 0;
 402                                 }
 403
 404                         }
 405
 406                 #else
 407                         ifstream inFASTA;
 408                         openInputFile(fastafile, inFASTA);
 409                         numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 410                         inFASTA.close();
 411                         lines.push_back(new linePair(0, numSeqs));
 412
 413                         driver(lines[0], outputFileName, fastafile, accnosFileName);
 414
 415                         if (m->control_pressed) {
 416                                         remove(outputFileName.c_str());
 417                                         remove(tempHeader.c_str());
 418                                         remove(accnosFileName.c_str());
 419                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 420                                         delete chimera;
 421                                         return 0;
 422                         }
 423
 424                         //delete accnos file if its blank
 425                         if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 426                 #endif
 427
 428                 appendFiles(tempHeader, outputFileName);
 429
 430                 remove(outputFileName.c_str());
 431                 rename(tempHeader.c_str(), outputFileName.c_str());
 432
 433         #endif
 434                 delete chimera;
 435
 436                 m->mothurOutEndLine();
 437                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 438                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 439                 if (hasAccnos) {  m->mothurOut(accnosFileName); m->mothurOutEndLine();  }
 440                 m->mothurOutEndLine();
 441
 442                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 443
 444                 m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 445
 446                 return 0;
 447
 448         }
 449         catch(exception& e) {
 450                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 451                 exit(1);
 452         }
 453 }
 454 //**********************************************************************************************************************
 455
 456 int ChimeraSlayerCommand::driver(linePair* line, string outputFName, string filename, string accnos){
 457         try {
 458                 ofstream out;
 459                 openOutputFile(outputFName, out);
 460
 461                 ofstream out2;
 462                 openOutputFile(accnos, out2);
 463
 464                 ifstream inFASTA;
 465                 openInputFile(filename, inFASTA);
 466
 467                 inFASTA.seekg(line->start);
 468
 469                 for(int i=0;i<line->numSeqs;i++){
 470
 471                         if (m->control_pressed) {       return 1;       }
 472
 473                         Sequence* candidateSeq = new Sequence(inFASTA);  gobble(inFASTA);
 474
 475                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 476
 477                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 478                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 479                                 }else{
 480                                         //find chimeras
 481                                         chimera->getChimeras(candidateSeq);
 482
 483                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 484
 485                                         //print results
 486                                         chimera->print(out, out2);
 487                                 }
 488                         }
 489                         delete candidateSeq;
 490
 491                         //report progress
 492                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1)); m->mothurOutEndLine();           }
 493                 }
 494                 //report progress
 495                 if((line->numSeqs) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(line->numSeqs)); m->mothurOutEndLine();         }
 496
 497                 out.close();
 498                 out2.close();
 499                 inFASTA.close();
 500
 501                 return 0;
 502         }
 503         catch(exception& e) {
 504                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 505                 exit(1);
 506         }
 507 }
 508 //**********************************************************************************************************************
 509 #ifdef USE_MPI
 510 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<long>& MPIPos){
 511         try {
 512
 513                 MPI_Status status;
 514                 int pid;
 515                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 516
 517                 for(int i=0;i<num;i++){
 518
 519                         if (m->control_pressed) {       return 1;       }
 520
 521                         //read next sequence
 522                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 523
 524                         char buf4[length];
 525                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 526
 527                         string tempBuf = buf4;
 528                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 529                         istringstream iss (tempBuf,istringstream::in);
 530
 531                         Sequence* candidateSeq = new Sequence(iss);  gobble(iss);
 532
 533                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 534
 535                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 536                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 537                                 }else{
 538                                         //find chimeras
 539                                         chimera->getChimeras(candidateSeq);
 540
 541                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 542
 543                                         //print results
 544                                         bool isChimeric = chimera->print(outMPI, outAccMPI);
 545                                         if (isChimeric) { MPIWroteAccnos = true;  }
 546                                 }
 547                         }
 548                         delete candidateSeq;
 549
 550                         //report progress
 551                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 552                 }
 553                 //report progress
 554                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 555
 556
 557                 return 0;
 558         }
 559         catch(exception& e) {
 560                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 561                 exit(1);
 562         }
 563 }
 564 #endif
 565
 566 /**************************************************************************************************/
 567
 568 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos) {
 569         try {
 570 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 571                 int process = 0;
 572                 //              processIDS.resize(0);
 573
 574                 //loop through and create all the processes you want
 575                 while (process != processors) {
 576                         int pid = fork();
 577
 578                         if (pid > 0) {
 579                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 580                                 process++;
 581                         }else if (pid == 0){
 582                                 driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 583                                 exit(0);
 584                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 585                 }
 586
 587                 //force parent to wait until all the processes are done
 588                 for (int i=0;i<processors;i++) {
 589                         int temp = processIDS[i];
 590                         wait(&temp);
 591                 }
 592
 593                 return 0;
 594 #endif
 595         }
 596         catch(exception& e) {
 597                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 598                 exit(1);
 599         }
 600 }
 601
 602 /**************************************************************************************************/
 603
 604