chimeraccodecommand.cpp

   1 /*
   2  *  chimeraccodecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/30/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraccodecommand.h"
  11 #include "ccode.h"
  12
  13 //***************************************************************************************************************
  14
  15 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"fasta", "filter", "processors", "window", "template", "mask", "numwanted", "outputdir","inputdir", };
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31                         map<string,string>::iterator it;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //if the user changes the input directory command factory will send this info to us in the output parameter
  39                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  40                         if (inputDir == "not found"){   inputDir = "";          }
  41                         else {
  42                                 string path;
  43                                 it = parameters.find("template");
  44                                 //user has given a template file
  45                                 if(it != parameters.end()){
  46                                         path = hasPath(it->second);
  47                                         //if the user has not given a path then, add inputdir. else leave path alone.
  48                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  49                                 }
  50                         }
  51
  52                         //check for required parameters
  53                         fastafile = validParameter.validFile(parameters, "fasta", false);
  54                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
  55                         else {
  56                                 splitAtDash(fastafile, fastaFileNames);
  57
  58                                 //go through files and make sure they are good, if not, then disregard them
  59                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  60                                         if (inputDir != "") {
  61                                                 string path = hasPath(fastaFileNames[i]);
  62                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  63                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
  64                                         }
  65
  66                                         int ableToOpen;
  67                                         ifstream in;
  68
  69                                         #ifdef USE_MPI
  70                                                 int pid;
  71                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
  72                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
  73
  74                                                 if (pid == 0) {
  75                                         #endif
  76
  77                                         ableToOpen = openInputFile(fastaFileNames[i], in);
  78                                         in.close();
  79
  80                                         #ifdef USE_MPI
  81                                                         for (int j = 1; j < processors; j++) {
  82                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
  83                                                         }
  84                                                 }else{
  85                                                         MPI_Status status;
  86                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
  87                                                 }
  88
  89                                         #endif
  90
  91                                         if (ableToOpen == 1) {
  92                                                 m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
  93                                                 //erase from file list
  94                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
  95                                                 i--;
  96                                         }
  97                                 }
  98
  99                                 //make sure there is at least one valid file left
 100                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 101                         }
 102
 103                         //if the user changes the output directory command factory will send this info to us in the output parameter
 104                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 105                                 outputDir = "";
 106                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
 107                         }
 108
 109                         templatefile = validParameter.validFile(parameters, "template", true);
 110                         if (templatefile == "not open") { abort = true; }
 111                         else if (templatefile == "not found") { templatefile = ""; m->mothurOut("template is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 112
 113                         maskfile = validParameter.validFile(parameters, "mask", false);
 114                         if (maskfile == "not found") { maskfile = "";  }
 115                         else if (maskfile != "default")  {
 116                                 if (inputDir != "") {
 117                                         string path = hasPath(maskfile);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 120                                 }
 121
 122                                 ifstream in;
 123                                 int     ableToOpen = openInputFile(maskfile, in);
 124                                 if (ableToOpen == 1) { abort = true; }
 125                                 in.close();
 126                         }
 127
 128                         string temp;
 129                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 130                         filter = isTrue(temp);
 131
 132                         temp = validParameter.validFile(parameters, "processors", false);               if (temp == "not found") { temp = "1"; }
 133                         convert(temp, processors);
 134
 135                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 136                         convert(temp, window);
 137
 138                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 139                         convert(temp, numwanted);
 140
 141                 }
 142         }
 143         catch(exception& e) {
 144                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 145                 exit(1);
 146         }
 147 }
 148 //**********************************************************************************************************************
 149
 150 void ChimeraCcodeCommand::help(){
 151         try {
 152
 153                 m->mothurOut("The chimera.ccode command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 154                 m->mothurOut("This command was created using the algorythms described in the 'Evaluating putative chimeric sequences from PCR-amplified products' paper by Juan M. Gonzalez, Johannes Zimmerman and Cesareo Saiz-Jimenez.\n");
 155                 m->mothurOut("The chimera.ccode command parameters are fasta, template, filter, mask, processors, window and numwanted.\n");
 156                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 157                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 158                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 159                 m->mothurOut("The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n");
 160                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 161                 #ifdef USE_MPI
 162                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 163                 #endif
 164                 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");
 165                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras. \n");
 166                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n");
 167                 m->mothurOut("The chimera.ccode command should be in the following format: \n");
 168                 m->mothurOut("chimera.ccode(fasta=yourFastaFile, template=yourTemplate) \n");
 169                 m->mothurOut("Example: chimera.ccode(fasta=AD.align, template=core_set_aligned.imputed.fasta) \n");
 170                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "ChimeraCcodeCommand", "help");
 174                 exit(1);
 175         }
 176 }
 177
 178 //***************************************************************************************************************
 179
 180 ChimeraCcodeCommand::~ChimeraCcodeCommand(){    /*      do nothing      */      }
 181
 182 //***************************************************************************************************************
 183
 184 int ChimeraCcodeCommand::execute(){
 185         try{
 186
 187                 if (abort == true) { return 0; }
 188
 189                 for (int s = 0; s < fastaFileNames.size(); s++) {
 190
 191                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 192
 193                         int start = time(NULL);
 194
 195                         //set user options
 196                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 197
 198                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 199
 200                         //is your template aligned?
 201                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 202                         templateSeqsLength = chimera->getLength();
 203
 204                         string outputFileName, accnosFileName;
 205                         if (maskfile != "") {
 206                                 outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 207                                 accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 208                         }else {
 209                                 outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 210                                 accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 211                         }
 212
 213                         string mapInfo = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "mapinfo";
 214                         bool hasAccnos = true;
 215
 216                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); }  return 0;    }
 217
 218                 #ifdef USE_MPI
 219
 220                                 int pid, end, numSeqsPerProcessor;
 221                                 int tag = 2001;
 222                                 vector<long> MPIPos;
 223                                 MPIWroteAccnos = false;
 224
 225                                 MPI_Status status;
 226                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 227                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 228
 229                                 MPI_File inMPI;
 230                                 MPI_File outMPI;
 231                                 MPI_File outMPIAccnos;
 232
 233                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 234                                 int inMode=MPI_MODE_RDONLY;
 235
 236                                 char outFilename[1024];
 237                                 strcpy(outFilename, outputFileName.c_str());
 238
 239                                 char outAccnosFilename[1024];
 240                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 241
 242                                 char inFileName[1024];
 243                                 strcpy(inFileName, fastaFileNames[s].c_str());
 244
 245                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 246                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 247                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 248
 249                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 250
 251                                 if (pid == 0) { //you are the root process
 252                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n\n";
 253
 254                                         //print header
 255                                         int length = outTemp.length();
 256                                         char* buf2 = new char[length];
 257                                         memcpy(buf2, outTemp.c_str(), length);
 258
 259                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 260                                         delete buf2;
 261
 262                                         MPIPos = setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 263
 264                                         //send file positions to all processes
 265                                         for(int i = 1; i < processors; i++) {
 266                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 267                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 268                                         }
 269
 270                                         //figure out how many sequences you have to align
 271                                         numSeqsPerProcessor = numSeqs / processors;
 272                                         int startIndex =  pid * numSeqsPerProcessor;
 273                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 274
 275
 276                                         //align your part
 277                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 278
 279                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 280
 281                                         for (int i = 1; i < processors; i++) {
 282                                                 bool tempResult;
 283                                                 MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 284                                                 if (tempResult != 0) { MPIWroteAccnos = true; }
 285                                         }
 286                                 }else{ //you are a child process
 287                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 288                                         MPIPos.resize(numSeqs+1);
 289                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 290
 291                                         //figure out how many sequences you have to align
 292                                         numSeqsPerProcessor = numSeqs / processors;
 293                                         int startIndex =  pid * numSeqsPerProcessor;
 294                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 295
 296
 297                                         //align your part
 298                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 299
 300                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 301
 302                                         MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 303                                 }
 304
 305                                 //close files
 306                                 MPI_File_close(&inMPI);
 307                                 MPI_File_close(&outMPI);
 308                                 MPI_File_close(&outMPIAccnos);
 309
 310                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 311
 312                                 //delete accnos file if blank
 313                                 if (pid == 0) {
 314                                         if (!MPIWroteAccnos) {
 315                                                 hasAccnos = false;
 316                                                 remove(accnosFileName.c_str());
 317                                         }
 318                                 }
 319
 320                 #else
 321                         ofstream outHeader;
 322                         string tempHeader = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 323                         openOutputFile(tempHeader, outHeader);
 324
 325                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 326
 327                         outHeader.close();
 328
 329                         //break up file
 330                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 331                                 if(processors == 1){
 332                                         ifstream inFASTA;
 333                                         openInputFile(fastaFileNames[s], inFASTA);
 334                                         getNumSeqs(inFASTA, numSeqs);
 335                                         inFASTA.close();
 336
 337                                         lines.push_back(new linePair(0, numSeqs));
 338
 339                                         driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 340
 341                                         if (m->control_pressed) {
 342                                                 remove(outputFileName.c_str());
 343                                                 remove(tempHeader.c_str());
 344                                                 remove(accnosFileName.c_str());
 345                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 346                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 347                                                 delete chimera;
 348                                                 return 0;
 349                                         }
 350
 351                                         //delete accnos file if its blank
 352                                         if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 353
 354                                 }else{
 355                                         vector<int> positions;
 356                                         processIDS.resize(0);
 357
 358                                         ifstream inFASTA;
 359                                         openInputFile(fastaFileNames[s], inFASTA);
 360
 361                                         string input;
 362                                         while(!inFASTA.eof()){
 363                                                 input = getline(inFASTA);
 364                                                 if (input.length() != 0) {
 365                                                         if(input[0] == '>'){    long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);  }
 366                                                 }
 367                                         }
 368                                         inFASTA.close();
 369
 370                                         numSeqs = positions.size();
 371
 372                                         int numSeqsPerProcessor = numSeqs / processors;
 373
 374                                         for (int i = 0; i < processors; i++) {
 375                                                 long int startPos = positions[ i * numSeqsPerProcessor ];
 376                                                 if(i == processors - 1){
 377                                                         numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
 378                                                 }
 379                                                 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 380                                         }
 381
 382
 383                                         createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 384
 385                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 386
 387                                         //append output files
 388                                         for(int i=1;i<processors;i++){
 389                                                 appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 390                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 391                                         }
 392
 393                                         vector<string> nonBlankAccnosFiles;
 394                                         //delete blank accnos files generated with multiple processes
 395                                         for(int i=0;i<processors;i++){
 396                                                 if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
 397                                                         nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
 398                                                 }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());  }
 399                                         }
 400
 401                                         //append accnos files
 402                                         if (nonBlankAccnosFiles.size() != 0) {
 403                                                 rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
 404
 405                                                 for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 406                                                         appendFiles(nonBlankAccnosFiles[h], accnosFileName);
 407                                                         remove(nonBlankAccnosFiles[h].c_str());
 408                                                 }
 409                                         }else{ hasAccnos = false;  }
 410
 411                                         if (m->control_pressed) {
 412                                                 remove(outputFileName.c_str());
 413                                                 remove(accnosFileName.c_str());
 414                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 415                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 416                                                 delete chimera;
 417                                                 return 0;
 418                                         }
 419
 420                                 }
 421
 422                         #else
 423                                 ifstream inFASTA;
 424                                 openInputFile(fastaFileNames[s], inFASTA);
 425                                 getNumSeqs(inFASTA, numSeqs);
 426                                 inFASTA.close();
 427                                 lines.push_back(new linePair(0, numSeqs));
 428
 429                                 driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 430
 431                                 if (m->control_pressed) {
 432                                                 remove(outputFileName.c_str());
 433                                                 remove(tempHeader.c_str());
 434                                                 remove(accnosFileName.c_str());
 435                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 436                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 437                                                 delete chimera;
 438                                                 return 0;
 439                                 }
 440
 441                                 //delete accnos file if its blank
 442                                 if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 443                         #endif
 444
 445                         appendFiles(outputFileName, tempHeader);
 446
 447                         remove(outputFileName.c_str());
 448                         rename(tempHeader.c_str(), outputFileName.c_str());
 449                 #endif
 450
 451                         delete chimera;
 452
 453                         outputNames.push_back(outputFileName);
 454                         outputNames.push_back(mapInfo);
 455                         if (hasAccnos) { outputNames.push_back(accnosFileName); }
 456
 457                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 458
 459                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 460                 }
 461
 462                 m->mothurOutEndLine();
 463                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 464                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 465                 m->mothurOutEndLine();
 466
 467                 return 0;
 468
 469         }
 470         catch(exception& e) {
 471                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 472                 exit(1);
 473         }
 474 }
 475 //**********************************************************************************************************************
 476
 477 int ChimeraCcodeCommand::driver(linePair* line, string outputFName, string filename, string accnos){
 478         try {
 479                 ofstream out;
 480                 openOutputFile(outputFName, out);
 481
 482                 ofstream out2;
 483                 openOutputFile(accnos, out2);
 484
 485                 ifstream inFASTA;
 486                 openInputFile(filename, inFASTA);
 487
 488                 inFASTA.seekg(line->start);
 489
 490                 for(int i=0;i<line->numSeqs;i++){
 491
 492                         if (m->control_pressed) {       return 1;       }
 493
 494                         Sequence* candidateSeq = new Sequence(inFASTA);  gobble(inFASTA);
 495
 496                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 497
 498                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 499                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 500                                 }else{
 501                                         //find chimeras
 502                                         chimera->getChimeras(candidateSeq);
 503
 504                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 505
 506                                         //print results
 507                                         chimera->print(out, out2);
 508                                 }
 509                         }
 510                         delete candidateSeq;
 511
 512                         //report progress
 513                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1)); m->mothurOutEndLine();           }
 514                 }
 515                 //report progress
 516                 if((line->numSeqs) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(line->numSeqs)); m->mothurOutEndLine();         }
 517
 518                 out.close();
 519                 out2.close();
 520                 inFASTA.close();
 521
 522                 return 0;
 523         }
 524         catch(exception& e) {
 525                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 526                 exit(1);
 527         }
 528 }
 529 //**********************************************************************************************************************
 530 #ifdef USE_MPI
 531 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<long>& MPIPos){
 532         try {
 533
 534                 MPI_Status status;
 535                 int pid;
 536                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 537
 538                 for(int i=0;i<num;i++){
 539
 540                         if (m->control_pressed) { return 0; }
 541
 542                         //read next sequence
 543                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 544
 545                         char* buf4 = new char[length];
 546
 547                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 548
 549                         string tempBuf = buf4;
 550                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 551                         istringstream iss (tempBuf,istringstream::in);
 552                         delete buf4;
 553
 554                         Sequence* candidateSeq = new Sequence(iss);  gobble(iss);
 555
 556                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 557
 558                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 559                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 560                                 }else{
 561                                         //find chimeras
 562                                         chimera->getChimeras(candidateSeq);
 563
 564                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 565
 566                                         //print results
 567                                         bool isChimeric = chimera->print(outMPI, outAccMPI);
 568                                         if (isChimeric) { MPIWroteAccnos = true;  }
 569                                 }
 570                         }
 571                         delete candidateSeq;
 572
 573                         //report progress
 574                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 575                 }
 576                 //report progress
 577                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 578
 579
 580                 return 0;
 581         }
 582         catch(exception& e) {
 583                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 584                 exit(1);
 585         }
 586 }
 587 #endif
 588
 589 /**************************************************************************************************/
 590
 591 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 592         try {
 593 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 594                 int process = 0;
 595                 //              processIDS.resize(0);
 596
 597                 //loop through and create all the processes you want
 598                 while (process != processors) {
 599                         int pid = fork();
 600
 601                         if (pid > 0) {
 602                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 603                                 process++;
 604                         }else if (pid == 0){
 605                                 driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 606                                 exit(0);
 607                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 608                 }
 609
 610                 //force parent to wait until all the processes are done
 611                 for (int i=0;i<processors;i++) {
 612                         int temp = processIDS[i];
 613                         wait(&temp);
 614                 }
 615
 616                 return 0;
 617 #endif
 618         }
 619         catch(exception& e) {
 620                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 621                 exit(1);
 622         }
 623 }
 624 //**********************************************************************************************************************
 625