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 vector<string> ChimeraCcodeCommand::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 pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
  18                 CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
  19                 CommandParameter pnumwanted("numwanted", "Number", "", "20", "", "", "",false,false); parameters.push_back(pnumwanted);
  20                 CommandParameter pmask("mask", "String", "", "", "", "", "",false,false); parameters.push_back(pmask);
  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, "ChimeraCcodeCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraCcodeCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.ccode command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  39                 helpString += "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";
  40                 helpString += "The chimera.ccode command parameters are fasta, reference, filter, mask, processors, window and numwanted.\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 filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n";
  45                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  46 #ifdef USE_MPI
  47                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  48 #endif
  49                 helpString += "The mask parameter allows you to specify a file containing one sequence you wish to use as a mask for the your sequences. \n";
  50                 helpString += "The window parameter allows you to specify the window size for searching for chimeras. \n";
  51                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n";
  52                 helpString += "The chimera.ccode command should be in the following format: \n";
  53                 helpString += "chimera.ccode(fasta=yourFastaFile, reference=yourTemplate) \n";
  54                 helpString += "Example: chimera.ccode(fasta=AD.align, reference=core_set_aligned.imputed.fasta) \n";
  55                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  56                 return helpString;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "ChimeraCcodeCommand", "getHelpString");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 ChimeraCcodeCommand::ChimeraCcodeCommand(){
  65         try {
  66                 abort = true; calledHelp = true;
  67                 setParameters();
  68                 vector<string> tempOutNames;
  69                 outputTypes["chimera"] = tempOutNames;
  70                 outputTypes["mapinfo"] = tempOutNames;
  71                 outputTypes["accnos"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
  75                 exit(1);
  76         }
  77 }
  78 //***************************************************************************************************************
  79 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82
  83                 //allow user to run help
  84                 if(option == "help") { help(); abort = true; calledHelp = true; }
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter("chimera.ccode");
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         vector<string> tempOutNames;
 101                         outputTypes["chimera"] = tempOutNames;
 102                         outputTypes["mapinfo"] = tempOutNames;
 103                         outputTypes["accnos"] = tempOutNames;
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("reference");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 116                                 }
 117                         }
 118
 119                         //check for required parameters
 120                         fastafile = validParameter.validFile(parameters, "fasta", false);
 121                         if (fastafile == "not found") {                                 //if there is a current fasta file, use it
 122                                 string filename = m->getFastaFile();
 123                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 124                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 125                         }else {
 126                                 m->splitAtDash(fastafile, fastaFileNames);
 127
 128                                 //go through files and make sure they are good, if not, then disregard them
 129                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 130                                         if (inputDir != "") {
 131                                                 string path = m->hasPath(fastaFileNames[i]);
 132                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 133                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 134                                         }
 135
 136                                         int ableToOpen;
 137                                         ifstream in;
 138
 139                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 140
 141                                         //if you can't open it, try default location
 142                                         if (ableToOpen == 1) {
 143                                                 if (m->getDefaultPath() != "") { //default path is set
 144                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 145                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 146                                                         ifstream in2;
 147                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 148                                                         in2.close();
 149                                                         fastaFileNames[i] = tryPath;
 150                                                 }
 151                                         }
 152
 153                                         //if you can't open it, try default location
 154                                         if (ableToOpen == 1) {
 155                                                 if (m->getOutputDir() != "") { //default path is set
 156                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 157                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 158                                                         ifstream in2;
 159                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 160                                                         in2.close();
 161                                                         fastaFileNames[i] = tryPath;
 162                                                 }
 163                                         }
 164
 165                                         in.close();
 166
 167                                         if (ableToOpen == 1) {
 168                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 169                                                 //erase from file list
 170                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 171                                                 i--;
 172                                         }
 173                                 }
 174
 175                                 //make sure there is at least one valid file left
 176                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 177                         }
 178
 179                         //if the user changes the output directory command factory will send this info to us in the output parameter
 180                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 181
 182                         templatefile = validParameter.validFile(parameters, "reference", true);
 183                         if (templatefile == "not open") { abort = true; }
 184                         else if (templatefile == "not found") { templatefile = ""; m->mothurOut("reference is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 185
 186                         maskfile = validParameter.validFile(parameters, "mask", false);
 187                         if (maskfile == "not found") { maskfile = "";  }
 188                         else if (maskfile != "default")  {
 189                                 if (inputDir != "") {
 190                                         string path = m->hasPath(maskfile);
 191                                         //if the user has not given a path then, add inputdir. else leave path alone.
 192                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 193                                 }
 194
 195                                 ifstream in;
 196                                 int     ableToOpen = m->openInputFile(maskfile, in);
 197                                 if (ableToOpen == 1) { abort = true; }
 198                                 in.close();
 199                         }
 200
 201                         string temp;
 202                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 203                         filter = m->isTrue(temp);
 204
 205                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 206                         m->setProcessors(temp);
 207                         convert(temp, processors);
 208
 209                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 210                         convert(temp, window);
 211
 212                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 213                         convert(temp, numwanted);
 214
 215                 }
 216         }
 217         catch(exception& e) {
 218                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 219                 exit(1);
 220         }
 221 }
 222 //***************************************************************************************************************
 223 int ChimeraCcodeCommand::execute(){
 224         try{
 225
 226                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 227
 228                 for (int s = 0; s < fastaFileNames.size(); s++) {
 229
 230                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 231
 232                         int start = time(NULL);
 233
 234                         //set user options
 235                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 236
 237                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 238
 239                         //is your template aligned?
 240                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 241                         templateSeqsLength = chimera->getLength();
 242
 243                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 244                         string outputFileName, accnosFileName;
 245                         if (maskfile != "") {
 246                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 247                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 248                         }else {
 249                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 250                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 251                         }
 252
 253                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "mapinfo";
 254
 255                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } outputTypes.clear(); return 0;        }
 256
 257                 #ifdef USE_MPI
 258
 259                                 int pid, numSeqsPerProcessor;
 260                                 int tag = 2001;
 261                                 vector<unsigned long int> MPIPos;
 262
 263                                 MPI_Status status;
 264                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 265                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 266
 267                                 MPI_File inMPI;
 268                                 MPI_File outMPI;
 269                                 MPI_File outMPIAccnos;
 270
 271                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 272                                 int inMode=MPI_MODE_RDONLY;
 273
 274                                 char outFilename[1024];
 275                                 strcpy(outFilename, outputFileName.c_str());
 276
 277                                 char outAccnosFilename[1024];
 278                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 279
 280                                 char inFileName[1024];
 281                                 strcpy(inFileName, fastaFileNames[s].c_str());
 282
 283                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 284                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 285                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 286
 287                                 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()); } outputTypes.clear();  delete chimera; return 0;  }
 288
 289                                 if (pid == 0) { //you are the root process
 290                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n";
 291
 292                                         //print header
 293                                         int length = outTemp.length();
 294                                         char* buf2 = new char[length];
 295                                         memcpy(buf2, outTemp.c_str(), length);
 296
 297                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 298                                         delete buf2;
 299
 300                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 301
 302                                         //send file positions to all processes
 303                                         for(int i = 1; i < processors; i++) {
 304                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 305                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 306                                         }
 307
 308                                         //figure out how many sequences you have to align
 309                                         numSeqsPerProcessor = numSeqs / processors;
 310                                         int startIndex =  pid * numSeqsPerProcessor;
 311                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 312
 313
 314                                         //align your part
 315                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 316
 317                                         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()); } outputTypes.clear();  delete chimera; return 0;  }
 318
 319                                 }else{ //you are a child process
 320                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 321                                         MPIPos.resize(numSeqs+1);
 322                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 323
 324                                         //figure out how many sequences you have to align
 325                                         numSeqsPerProcessor = numSeqs / processors;
 326                                         int startIndex =  pid * numSeqsPerProcessor;
 327                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 328
 329
 330                                         //align your part
 331                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 332
 333                                         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()); }  outputTypes.clear(); delete chimera; return 0;  }
 334                                 }
 335
 336                                 //close files
 337                                 MPI_File_close(&inMPI);
 338                                 MPI_File_close(&outMPI);
 339                                 MPI_File_close(&outMPIAccnos);
 340
 341                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 342
 343                 #else
 344                         ofstream outHeader;
 345                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 346                         m->openOutputFile(tempHeader, outHeader);
 347
 348                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 349
 350                         outHeader.close();
 351
 352                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 353
 354                         for (int i = 0; i < (positions.size()-1); i++) {
 355                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 356                         }
 357
 358                         //break up file
 359                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 360                                 if(processors == 1){
 361
 362                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 363
 364                                         if (m->control_pressed) { remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 365
 366                                 }else{
 367                                         processIDS.resize(0);
 368
 369                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 370
 371                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 372                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 373
 374                                         //append output files
 375                                         for(int i=1;i<processors;i++){
 376                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 377                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 378                                         }
 379
 380                                         //append output files
 381                                         for(int i=1;i<processors;i++){
 382                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 383                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 384                                         }
 385
 386                                         if (m->control_pressed) {
 387                                                 remove(outputFileName.c_str());
 388                                                 remove(accnosFileName.c_str());
 389                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 390                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 391                                                 delete chimera;
 392                                                 return 0;
 393                                         }
 394
 395                                 }
 396
 397                         #else
 398                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 399
 400                                 if (m->control_pressed) { remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 401
 402                         #endif
 403
 404                         m->appendFiles(outputFileName, tempHeader);
 405
 406                         remove(outputFileName.c_str());
 407                         rename(tempHeader.c_str(), outputFileName.c_str());
 408                 #endif
 409
 410                         delete chimera;
 411
 412                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 413                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 414                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 415
 416                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 417
 418                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 419                 }
 420
 421
 422                 //set accnos file as new current accnosfile
 423                 string current = "";
 424                 itTypes = outputTypes.find("accnos");
 425                 if (itTypes != outputTypes.end()) {
 426                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 427                 }
 428
 429                 m->mothurOutEndLine();
 430                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 431                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 432                 m->mothurOutEndLine();
 433
 434                 return 0;
 435
 436         }
 437         catch(exception& e) {
 438                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 439                 exit(1);
 440         }
 441 }
 442 //**********************************************************************************************************************
 443
 444 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 445         try {
 446                 ofstream out;
 447                 m->openOutputFile(outputFName, out);
 448
 449                 ofstream out2;
 450                 m->openOutputFile(accnos, out2);
 451
 452                 ifstream inFASTA;
 453                 m->openInputFile(filename, inFASTA);
 454
 455                 inFASTA.seekg(filePos->start);
 456
 457                 bool done = false;
 458                 int count = 0;
 459
 460                 while (!done) {
 461
 462                         if (m->control_pressed) {       return 1;       }
 463
 464                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 465
 466                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 467
 468                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 469                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 470                                 }else{
 471                                         //find chimeras
 472                                         chimera->getChimeras(candidateSeq);
 473
 474                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 475
 476                                         //print results
 477                                         chimera->print(out, out2);
 478                                 }
 479                                 count++;
 480                         }
 481                         delete candidateSeq;
 482
 483                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 484                                 unsigned long int pos = inFASTA.tellg();
 485                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 486                         #else
 487                                 if (inFASTA.eof()) { break; }
 488                         #endif
 489
 490                         //report progress
 491                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 492                 }
 493                 //report progress
 494                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 495
 496                 out.close();
 497                 out2.close();
 498                 inFASTA.close();
 499
 500                 return count;
 501         }
 502         catch(exception& e) {
 503                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 504                 exit(1);
 505         }
 506 }
 507 //**********************************************************************************************************************
 508 #ifdef USE_MPI
 509 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 510         try {
 511
 512                 MPI_Status status;
 513                 int pid;
 514                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 515
 516                 for(int i=0;i<num;i++){
 517
 518                         if (m->control_pressed) { return 0; }
 519
 520                         //read next sequence
 521                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 522
 523                         char* buf4 = new char[length];
 524
 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                         delete buf4;
 531
 532                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 533
 534                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 535
 536                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 537                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 538                                 }else{
 539                                         //find chimeras
 540                                         chimera->getChimeras(candidateSeq);
 541
 542                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 543
 544                                         //print results
 545                                         chimera->print(outMPI, outAccMPI);
 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, "ChimeraCcodeCommand", "driverMPI");
 561                 exit(1);
 562         }
 563 }
 564 #endif
 565
 566 /**************************************************************************************************/
 567
 568 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 569         try {
 570 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 571                 int process = 0;
 572                 int num = 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                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 583
 584                                 //pass numSeqs to parent
 585                                 ofstream out;
 586                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 587                                 m->openOutputFile(tempFile, out);
 588                                 out << num << endl;
 589                                 out.close();
 590
 591                                 exit(0);
 592                         }else {
 593                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 594                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 595                                 exit(0);
 596                         }
 597                 }
 598
 599                 //force parent to wait until all the processes are done
 600                 for (int i=0;i<processors;i++) {
 601                         int temp = processIDS[i];
 602                         wait(&temp);
 603                 }
 604
 605                 for (int i = 0; i < processIDS.size(); i++) {
 606                         ifstream in;
 607                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 608                         m->openInputFile(tempFile, in);
 609                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 610                         in.close(); remove(tempFile.c_str());
 611                 }
 612
 613                 return num;
 614 #endif
 615         }
 616         catch(exception& e) {
 617                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 618                 exit(1);
 619         }
 620 }
 621 //**********************************************************************************************************************
 622