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                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  86
  87                 else {
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters = parser.getParameters();
  92
  93                         ValidParameters validParameter("chimera.ccode");
  94                         map<string,string>::iterator it;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         vector<string> tempOutNames;
 102                         outputTypes["chimera"] = tempOutNames;
 103                         outputTypes["mapinfo"] = tempOutNames;
 104                         outputTypes["accnos"] = tempOutNames;
 105
 106                         //if the user changes the input directory command factory will send this info to us in the output parameter
 107                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 108                         if (inputDir == "not found"){   inputDir = "";          }
 109                         else {
 110                                 string path;
 111                                 it = parameters.find("reference");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 117                                 }
 118                         }
 119
 120                         //check for required parameters
 121                         fastafile = validParameter.validFile(parameters, "fasta", false);
 122                         if (fastafile == "not found") {                                 //if there is a current fasta file, use it
 123                                 string filename = m->getFastaFile();
 124                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 125                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 126                         }else {
 127                                 m->splitAtDash(fastafile, fastaFileNames);
 128
 129                                 //go through files and make sure they are good, if not, then disregard them
 130                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 131
 132                                         bool ignore = false;
 133                                         if (fastaFileNames[i] == "current") {
 134                                                 fastaFileNames[i] = m->getFastaFile();
 135                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 136                                                 else {
 137                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 138                                                         //erase from file list
 139                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 140                                                         i--;
 141                                                 }
 142                                         }
 143
 144                                         if (!ignore) {
 145
 146                                                 if (inputDir != "") {
 147                                                         string path = m->hasPath(fastaFileNames[i]);
 148                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 150                                                 }
 151
 152                                                 int ableToOpen;
 153                                                 ifstream in;
 154
 155                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 156
 157                                                 //if you can't open it, try default location
 158                                                 if (ableToOpen == 1) {
 159                                                         if (m->getDefaultPath() != "") { //default path is set
 160                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 161                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 162                                                                 ifstream in2;
 163                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 164                                                                 in2.close();
 165                                                                 fastaFileNames[i] = tryPath;
 166                                                         }
 167                                                 }
 168
 169                                                 //if you can't open it, try default location
 170                                                 if (ableToOpen == 1) {
 171                                                         if (m->getOutputDir() != "") { //default path is set
 172                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 173                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 174                                                                 ifstream in2;
 175                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 176                                                                 in2.close();
 177                                                                 fastaFileNames[i] = tryPath;
 178                                                         }
 179                                                 }
 180
 181                                                 in.close();
 182
 183                                                 if (ableToOpen == 1) {
 184                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 185                                                         //erase from file list
 186                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 187                                                         i--;
 188                                                 }
 189                                         }
 190                                 }
 191
 192                                 //make sure there is at least one valid file left
 193                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 194                         }
 195
 196                         //if the user changes the output directory command factory will send this info to us in the output parameter
 197                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 198
 199                         templatefile = validParameter.validFile(parameters, "reference", true);
 200                         if (templatefile == "not open") { abort = true; }
 201                         else if (templatefile == "not found") { templatefile = ""; m->mothurOut("reference is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 202
 203                         maskfile = validParameter.validFile(parameters, "mask", false);
 204                         if (maskfile == "not found") { maskfile = "";  }
 205                         else if (maskfile != "default")  {
 206                                 if (inputDir != "") {
 207                                         string path = m->hasPath(maskfile);
 208                                         //if the user has not given a path then, add inputdir. else leave path alone.
 209                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 210                                 }
 211
 212                                 ifstream in;
 213                                 int     ableToOpen = m->openInputFile(maskfile, in);
 214                                 if (ableToOpen == 1) { abort = true; }
 215                                 in.close();
 216                         }
 217
 218                         string temp;
 219                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 220                         filter = m->isTrue(temp);
 221
 222                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 223                         m->setProcessors(temp);
 224                         convert(temp, processors);
 225
 226                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 227                         convert(temp, window);
 228
 229                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 230                         convert(temp, numwanted);
 231
 232                 }
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 236                 exit(1);
 237         }
 238 }
 239 //***************************************************************************************************************
 240 int ChimeraCcodeCommand::execute(){
 241         try{
 242
 243                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 244
 245                 for (int s = 0; s < fastaFileNames.size(); s++) {
 246
 247                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 248
 249                         int start = time(NULL);
 250
 251                         //set user options
 252                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 253
 254                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 255
 256                         //is your template aligned?
 257                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 258                         templateSeqsLength = chimera->getLength();
 259
 260                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 261                         string outputFileName, accnosFileName;
 262                         if (maskfile != "") {
 263                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 264                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 265                         }else {
 266                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 267                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 268                         }
 269
 270                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "mapinfo";
 271
 272                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } outputTypes.clear(); return 0;        }
 273
 274                 #ifdef USE_MPI
 275
 276                                 int pid, numSeqsPerProcessor;
 277                                 int tag = 2001;
 278                                 vector<unsigned long int> MPIPos;
 279
 280                                 MPI_Status status;
 281                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 282                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 283
 284                                 MPI_File inMPI;
 285                                 MPI_File outMPI;
 286                                 MPI_File outMPIAccnos;
 287
 288                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 289                                 int inMode=MPI_MODE_RDONLY;
 290
 291                                 char outFilename[1024];
 292                                 strcpy(outFilename, outputFileName.c_str());
 293
 294                                 char outAccnosFilename[1024];
 295                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 296
 297                                 char inFileName[1024];
 298                                 strcpy(inFileName, fastaFileNames[s].c_str());
 299
 300                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 301                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 302                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 303
 304                                 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;  }
 305
 306                                 if (pid == 0) { //you are the root process
 307                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n";
 308
 309                                         //print header
 310                                         int length = outTemp.length();
 311                                         char* buf2 = new char[length];
 312                                         memcpy(buf2, outTemp.c_str(), length);
 313
 314                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 315                                         delete buf2;
 316
 317                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 318
 319                                         //send file positions to all processes
 320                                         for(int i = 1; i < processors; i++) {
 321                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 322                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 323                                         }
 324
 325                                         //figure out how many sequences you have to align
 326                                         numSeqsPerProcessor = numSeqs / processors;
 327                                         int startIndex =  pid * numSeqsPerProcessor;
 328                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 329
 330
 331                                         //align your part
 332                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 333
 334                                         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;  }
 335
 336                                 }else{ //you are a child process
 337                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 338                                         MPIPos.resize(numSeqs+1);
 339                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 340
 341                                         //figure out how many sequences you have to align
 342                                         numSeqsPerProcessor = numSeqs / processors;
 343                                         int startIndex =  pid * numSeqsPerProcessor;
 344                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 345
 346
 347                                         //align your part
 348                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 349
 350                                         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;  }
 351                                 }
 352
 353                                 //close files
 354                                 MPI_File_close(&inMPI);
 355                                 MPI_File_close(&outMPI);
 356                                 MPI_File_close(&outMPIAccnos);
 357
 358                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 359
 360                 #else
 361                         ofstream outHeader;
 362                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 363                         m->openOutputFile(tempHeader, outHeader);
 364
 365                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 366
 367                         outHeader.close();
 368
 369                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 370
 371                         for (int i = 0; i < (positions.size()-1); i++) {
 372                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 373                         }
 374
 375                         //break up file
 376                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 377                                 if(processors == 1){
 378
 379                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 380
 381                                         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; }
 382
 383                                 }else{
 384                                         processIDS.resize(0);
 385
 386                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 387
 388                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 389                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 390
 391                                         //append output files
 392                                         for(int i=1;i<processors;i++){
 393                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 394                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 395                                         }
 396
 397                                         //append output files
 398                                         for(int i=1;i<processors;i++){
 399                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 400                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 401                                         }
 402
 403                                         if (m->control_pressed) {
 404                                                 remove(outputFileName.c_str());
 405                                                 remove(accnosFileName.c_str());
 406                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 407                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 408                                                 delete chimera;
 409                                                 return 0;
 410                                         }
 411
 412                                 }
 413
 414                         #else
 415                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 416
 417                                 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; }
 418
 419                         #endif
 420
 421                         m->appendFiles(outputFileName, tempHeader);
 422
 423                         remove(outputFileName.c_str());
 424                         rename(tempHeader.c_str(), outputFileName.c_str());
 425                 #endif
 426
 427                         delete chimera;
 428
 429                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 430                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 431                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 432
 433                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 434
 435                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 436                 }
 437
 438
 439                 //set accnos file as new current accnosfile
 440                 string current = "";
 441                 itTypes = outputTypes.find("accnos");
 442                 if (itTypes != outputTypes.end()) {
 443                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 444                 }
 445
 446                 m->mothurOutEndLine();
 447                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 448                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 449                 m->mothurOutEndLine();
 450
 451                 return 0;
 452
 453         }
 454         catch(exception& e) {
 455                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 456                 exit(1);
 457         }
 458 }
 459 //**********************************************************************************************************************
 460
 461 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 462         try {
 463                 ofstream out;
 464                 m->openOutputFile(outputFName, out);
 465
 466                 ofstream out2;
 467                 m->openOutputFile(accnos, out2);
 468
 469                 ifstream inFASTA;
 470                 m->openInputFile(filename, inFASTA);
 471
 472                 inFASTA.seekg(filePos->start);
 473
 474                 bool done = false;
 475                 int count = 0;
 476
 477                 while (!done) {
 478
 479                         if (m->control_pressed) {       return 1;       }
 480
 481                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 482
 483                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 484
 485                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 486                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 487                                 }else{
 488                                         //find chimeras
 489                                         chimera->getChimeras(candidateSeq);
 490
 491                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 492
 493                                         //print results
 494                                         chimera->print(out, out2);
 495                                 }
 496                                 count++;
 497                         }
 498                         delete candidateSeq;
 499
 500                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 501                                 unsigned long int pos = inFASTA.tellg();
 502                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 503                         #else
 504                                 if (inFASTA.eof()) { break; }
 505                         #endif
 506
 507                         //report progress
 508                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 509                 }
 510                 //report progress
 511                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 512
 513                 out.close();
 514                 out2.close();
 515                 inFASTA.close();
 516
 517                 return count;
 518         }
 519         catch(exception& e) {
 520                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 521                 exit(1);
 522         }
 523 }
 524 //**********************************************************************************************************************
 525 #ifdef USE_MPI
 526 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 527         try {
 528
 529                 MPI_Status status;
 530                 int pid;
 531                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 532
 533                 for(int i=0;i<num;i++){
 534
 535                         if (m->control_pressed) { return 0; }
 536
 537                         //read next sequence
 538                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 539
 540                         char* buf4 = new char[length];
 541
 542                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 543
 544                         string tempBuf = buf4;
 545                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 546                         istringstream iss (tempBuf,istringstream::in);
 547                         delete buf4;
 548
 549                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 550
 551                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 552
 553                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 554                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 555                                 }else{
 556                                         //find chimeras
 557                                         chimera->getChimeras(candidateSeq);
 558
 559                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 560
 561                                         //print results
 562                                         chimera->print(outMPI, outAccMPI);
 563                                 }
 564                         }
 565                         delete candidateSeq;
 566
 567                         //report progress
 568                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 569                 }
 570                 //report progress
 571                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 572
 573
 574                 return 0;
 575         }
 576         catch(exception& e) {
 577                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 578                 exit(1);
 579         }
 580 }
 581 #endif
 582
 583 /**************************************************************************************************/
 584
 585 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 586         try {
 587 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 588                 int process = 0;
 589                 int num = 0;
 590
 591                 //loop through and create all the processes you want
 592                 while (process != processors) {
 593                         int pid = fork();
 594
 595                         if (pid > 0) {
 596                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 597                                 process++;
 598                         }else if (pid == 0){
 599                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 600
 601                                 //pass numSeqs to parent
 602                                 ofstream out;
 603                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 604                                 m->openOutputFile(tempFile, out);
 605                                 out << num << endl;
 606                                 out.close();
 607
 608                                 exit(0);
 609                         }else {
 610                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 611                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 612                                 exit(0);
 613                         }
 614                 }
 615
 616                 //force parent to wait until all the processes are done
 617                 for (int i=0;i<processors;i++) {
 618                         int temp = processIDS[i];
 619                         wait(&temp);
 620                 }
 621
 622                 for (int i = 0; i < processIDS.size(); i++) {
 623                         ifstream in;
 624                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 625                         m->openInputFile(tempFile, in);
 626                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 627                         in.close(); remove(tempFile.c_str());
 628                 }
 629
 630                 return num;
 631 #endif
 632         }
 633         catch(exception& e) {
 634                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 635                 exit(1);
 636         }
 637 }
 638 //**********************************************************************************************************************
 639