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 vector<string> ChimeraCcodeCommand::getValidParameters(){
  15         try {
  16                 string AlignArray[] =  {"fasta", "filter", "processors", "window", "template", "mask", "numwanted", "outputdir","inputdir" };
  17                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "ChimeraCcodeCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 ChimeraCcodeCommand::ChimeraCcodeCommand(){
  27         try {
  28                 vector<string> tempOutNames;
  29                 outputTypes["chimera"] = tempOutNames;
  30                 outputTypes["mapinfo"] = tempOutNames;
  31                 outputTypes["accnos"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> ChimeraCcodeCommand::getRequiredParameters(){
  40         try {
  41                 string AlignArray[] =  {"template","fasta"};
  42                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "ChimeraCcodeCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> ChimeraCcodeCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "ChimeraCcodeCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 //***************************************************************************************************************
  62 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
  63         try {
  64                 abort = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta", "filter", "processors", "window", "template", "mask", "numwanted", "outputdir","inputdir" };
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  73
  74                         OptionParser parser(option);
  75                         map<string,string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter("chimera.ccode");
  78                         map<string,string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         vector<string> tempOutNames;
  86                         outputTypes["chimera"] = tempOutNames;
  87                         outputTypes["mapinfo"] = tempOutNames;
  88                         outputTypes["accnos"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("template");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
 101                                 }
 102                         }
 103
 104                         //check for required parameters
 105                         fastafile = validParameter.validFile(parameters, "fasta", false);
 106                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 107                         else {
 108                                 m->splitAtDash(fastafile, fastaFileNames);
 109
 110                                 //go through files and make sure they are good, if not, then disregard them
 111                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 112                                         if (inputDir != "") {
 113                                                 string path = m->hasPath(fastaFileNames[i]);
 114                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 115                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 116                                         }
 117
 118                                         int ableToOpen;
 119                                         ifstream in;
 120
 121                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 122
 123                                         //if you can't open it, try default location
 124                                         if (ableToOpen == 1) {
 125                                                 if (m->getDefaultPath() != "") { //default path is set
 126                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 127                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 128                                                         ifstream in2;
 129                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 130                                                         in2.close();
 131                                                         fastaFileNames[i] = tryPath;
 132                                                 }
 133                                         }
 134
 135                                         //if you can't open it, try default location
 136                                         if (ableToOpen == 1) {
 137                                                 if (m->getOutputDir() != "") { //default path is set
 138                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 139                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 140                                                         ifstream in2;
 141                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 142                                                         in2.close();
 143                                                         fastaFileNames[i] = tryPath;
 144                                                 }
 145                                         }
 146
 147                                         in.close();
 148
 149                                         if (ableToOpen == 1) {
 150                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 151                                                 //erase from file list
 152                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 153                                                 i--;
 154                                         }
 155                                 }
 156
 157                                 //make sure there is at least one valid file left
 158                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 159                         }
 160
 161                         //if the user changes the output directory command factory will send this info to us in the output parameter
 162                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 163
 164                         templatefile = validParameter.validFile(parameters, "template", true);
 165                         if (templatefile == "not open") { abort = true; }
 166                         else if (templatefile == "not found") { templatefile = ""; m->mothurOut("template is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 167
 168                         maskfile = validParameter.validFile(parameters, "mask", false);
 169                         if (maskfile == "not found") { maskfile = "";  }
 170                         else if (maskfile != "default")  {
 171                                 if (inputDir != "") {
 172                                         string path = m->hasPath(maskfile);
 173                                         //if the user has not given a path then, add inputdir. else leave path alone.
 174                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 175                                 }
 176
 177                                 ifstream in;
 178                                 int     ableToOpen = m->openInputFile(maskfile, in);
 179                                 if (ableToOpen == 1) { abort = true; }
 180                                 in.close();
 181                         }
 182
 183                         string temp;
 184                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 185                         filter = m->isTrue(temp);
 186
 187                         temp = validParameter.validFile(parameters, "processors", false);               if (temp == "not found") { temp = "1"; }
 188                         convert(temp, processors);
 189
 190                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 191                         convert(temp, window);
 192
 193                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 194                         convert(temp, numwanted);
 195
 196                 }
 197         }
 198         catch(exception& e) {
 199                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 200                 exit(1);
 201         }
 202 }
 203 //**********************************************************************************************************************
 204
 205 void ChimeraCcodeCommand::help(){
 206         try {
 207
 208                 m->mothurOut("The chimera.ccode command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 209                 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");
 210                 m->mothurOut("The chimera.ccode command parameters are fasta, template, filter, mask, processors, window and numwanted.\n");
 211                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 212                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 213                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 214                 m->mothurOut("The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n");
 215                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 216                 #ifdef USE_MPI
 217                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 218                 #endif
 219                 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");
 220                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras. \n");
 221                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n");
 222                 m->mothurOut("The chimera.ccode command should be in the following format: \n");
 223                 m->mothurOut("chimera.ccode(fasta=yourFastaFile, template=yourTemplate) \n");
 224                 m->mothurOut("Example: chimera.ccode(fasta=AD.align, template=core_set_aligned.imputed.fasta) \n");
 225                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 226         }
 227         catch(exception& e) {
 228                 m->errorOut(e, "ChimeraCcodeCommand", "help");
 229                 exit(1);
 230         }
 231 }
 232
 233 //***************************************************************************************************************
 234
 235 ChimeraCcodeCommand::~ChimeraCcodeCommand(){    /*      do nothing      */      }
 236
 237 //***************************************************************************************************************
 238
 239 int ChimeraCcodeCommand::execute(){
 240         try{
 241
 242                 if (abort == true) { return 0; }
 243
 244                 for (int s = 0; s < fastaFileNames.size(); s++) {
 245
 246                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 247
 248                         int start = time(NULL);
 249
 250                         //set user options
 251                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 252
 253                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 254
 255                         //is your template aligned?
 256                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 257                         templateSeqsLength = chimera->getLength();
 258
 259                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 260                         string outputFileName, accnosFileName;
 261                         if (maskfile != "") {
 262                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 263                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 264                         }else {
 265                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 266                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 267                         }
 268
 269                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "mapinfo";
 270
 271                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } outputTypes.clear(); return 0;        }
 272
 273                 #ifdef USE_MPI
 274
 275                                 int pid, end, numSeqsPerProcessor;
 276                                 int tag = 2001;
 277                                 vector<unsigned long int> MPIPos;
 278
 279                                 MPI_Status status;
 280                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 281                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 282
 283                                 MPI_File inMPI;
 284                                 MPI_File outMPI;
 285                                 MPI_File outMPIAccnos;
 286
 287                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 288                                 int inMode=MPI_MODE_RDONLY;
 289
 290                                 char outFilename[1024];
 291                                 strcpy(outFilename, outputFileName.c_str());
 292
 293                                 char outAccnosFilename[1024];
 294                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 295
 296                                 char inFileName[1024];
 297                                 strcpy(inFileName, fastaFileNames[s].c_str());
 298
 299                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 300                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 301                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 302
 303                                 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;  }
 304
 305                                 if (pid == 0) { //you are the root process
 306                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n\n";
 307
 308                                         //print header
 309                                         int length = outTemp.length();
 310                                         char* buf2 = new char[length];
 311                                         memcpy(buf2, outTemp.c_str(), length);
 312
 313                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 314                                         delete buf2;
 315
 316                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 317
 318                                         //send file positions to all processes
 319                                         for(int i = 1; i < processors; i++) {
 320                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 321                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 322                                         }
 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);  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;  }
 334
 335                                 }else{ //you are a child process
 336                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 337                                         MPIPos.resize(numSeqs+1);
 338                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 339
 340                                         //figure out how many sequences you have to align
 341                                         numSeqsPerProcessor = numSeqs / processors;
 342                                         int startIndex =  pid * numSeqsPerProcessor;
 343                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 344
 345
 346                                         //align your part
 347                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 348
 349                                         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;  }
 350                                 }
 351
 352                                 //close files
 353                                 MPI_File_close(&inMPI);
 354                                 MPI_File_close(&outMPI);
 355                                 MPI_File_close(&outMPIAccnos);
 356
 357                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 358
 359                 #else
 360                         ofstream outHeader;
 361                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 362                         m->openOutputFile(tempHeader, outHeader);
 363
 364                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 365
 366                         outHeader.close();
 367
 368                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 369
 370                         for (int i = 0; i < (positions.size()-1); i++) {
 371                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 372                         }
 373
 374                         //break up file
 375                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 376                                 if(processors == 1){
 377
 378                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 379
 380                                         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; }
 381
 382                                 }else{
 383                                         processIDS.resize(0);
 384
 385                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 386
 387                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 388                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 389
 390                                         //append output files
 391                                         for(int i=1;i<processors;i++){
 392                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 393                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 394                                         }
 395
 396                                         //append output files
 397                                         for(int i=1;i<processors;i++){
 398                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 399                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 400                                         }
 401
 402                                         if (m->control_pressed) {
 403                                                 remove(outputFileName.c_str());
 404                                                 remove(accnosFileName.c_str());
 405                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 406                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 407                                                 delete chimera;
 408                                                 return 0;
 409                                         }
 410
 411                                 }
 412
 413                         #else
 414                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 415
 416                                 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; }
 417
 418                         #endif
 419
 420                         m->appendFiles(outputFileName, tempHeader);
 421
 422                         remove(outputFileName.c_str());
 423                         rename(tempHeader.c_str(), outputFileName.c_str());
 424                 #endif
 425
 426                         delete chimera;
 427
 428                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 429                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 430                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 431
 432                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 433
 434                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 435                 }
 436
 437                 m->mothurOutEndLine();
 438                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 439                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 440                 m->mothurOutEndLine();
 441
 442                 return 0;
 443
 444         }
 445         catch(exception& e) {
 446                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 447                 exit(1);
 448         }
 449 }
 450 //**********************************************************************************************************************
 451
 452 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 453         try {
 454                 ofstream out;
 455                 m->openOutputFile(outputFName, out);
 456
 457                 ofstream out2;
 458                 m->openOutputFile(accnos, out2);
 459
 460                 ifstream inFASTA;
 461                 m->openInputFile(filename, inFASTA);
 462
 463                 inFASTA.seekg(filePos->start);
 464
 465                 bool done = false;
 466                 int count = 0;
 467
 468                 while (!done) {
 469
 470                         if (m->control_pressed) {       return 1;       }
 471
 472                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 473
 474                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 475
 476                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 477                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 478                                 }else{
 479                                         //find chimeras
 480                                         chimera->getChimeras(candidateSeq);
 481
 482                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 483
 484                                         //print results
 485                                         chimera->print(out, out2);
 486                                 }
 487                                 count++;
 488                         }
 489                         delete candidateSeq;
 490
 491                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 492                                 unsigned long int pos = inFASTA.tellg();
 493                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 494                         #else
 495                                 if (inFASTA.eof()) { break; }
 496                         #endif
 497
 498                         //report progress
 499                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 500                 }
 501                 //report progress
 502                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 503
 504                 out.close();
 505                 out2.close();
 506                 inFASTA.close();
 507
 508                 return count;
 509         }
 510         catch(exception& e) {
 511                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 512                 exit(1);
 513         }
 514 }
 515 //**********************************************************************************************************************
 516 #ifdef USE_MPI
 517 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 518         try {
 519
 520                 MPI_Status status;
 521                 int pid;
 522                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 523
 524                 for(int i=0;i<num;i++){
 525
 526                         if (m->control_pressed) { return 0; }
 527
 528                         //read next sequence
 529                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 530
 531                         char* buf4 = new char[length];
 532
 533                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 534
 535                         string tempBuf = buf4;
 536                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 537                         istringstream iss (tempBuf,istringstream::in);
 538                         delete buf4;
 539
 540                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 541
 542                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 543
 544                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 545                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 546                                 }else{
 547                                         //find chimeras
 548                                         chimera->getChimeras(candidateSeq);
 549
 550                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 551
 552                                         //print results
 553                                         bool isChimeric = chimera->print(outMPI, outAccMPI);
 554                                 }
 555                         }
 556                         delete candidateSeq;
 557
 558                         //report progress
 559                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 560                 }
 561                 //report progress
 562                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 563
 564
 565                 return 0;
 566         }
 567         catch(exception& e) {
 568                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 569                 exit(1);
 570         }
 571 }
 572 #endif
 573
 574 /**************************************************************************************************/
 575
 576 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 577         try {
 578 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 579                 int process = 0;
 580                 int num = 0;
 581
 582                 //loop through and create all the processes you want
 583                 while (process != processors) {
 584                         int pid = fork();
 585
 586                         if (pid > 0) {
 587                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 588                                 process++;
 589                         }else if (pid == 0){
 590                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 591
 592                                 //pass numSeqs to parent
 593                                 ofstream out;
 594                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 595                                 m->openOutputFile(tempFile, out);
 596                                 out << num << endl;
 597                                 out.close();
 598
 599                                 exit(0);
 600                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 601                 }
 602
 603                 //force parent to wait until all the processes are done
 604                 for (int i=0;i<processors;i++) {
 605                         int temp = processIDS[i];
 606                         wait(&temp);
 607                 }
 608
 609                 for (int i = 0; i < processIDS.size(); i++) {
 610                         ifstream in;
 611                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 612                         m->openInputFile(tempFile, in);
 613                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 614                         in.close(); remove(tempFile.c_str());
 615                 }
 616
 617                 return num;
 618 #endif
 619         }
 620         catch(exception& e) {
 621                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 622                 exit(1);
 623         }
 624 }
 625 //**********************************************************************************************************************
 626