chimeracheckcommand.cpp

   1 /*
   2  *  chimeracheckcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/31/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeracheckcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> ChimeraCheckCommand::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 pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  18                 CommandParameter psvg("svg", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psvg);
  19                 CommandParameter pincrement("increment", "Number", "", "10", "", "", "",false,false); parameters.push_back(pincrement);
  20                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  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, "ChimeraCheckCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraCheckCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.check command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  39                 helpString += "This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n";
  40                 helpString += "The chimera.check command parameters are fasta, reference, processors, ksize, increment, svg and name.\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 processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  45 #ifdef USE_MPI
  46                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  47 #endif
  48                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n";
  49                 helpString += "The ksize parameter allows you to input kmersize, default is 7. \n";
  50                 helpString += "The svg parameter allows you to specify whether or not you would like a svg file outputted for each query sequence, default is False.\n";
  51                 helpString += "The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n";
  52                 helpString += "You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n";
  53                 helpString += "The chimera.check command should be in the following format: \n";
  54                 helpString += "chimera.check(fasta=yourFastaFile, reference=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n";
  55                 helpString += "Example: chimera.check(fasta=AD.fasta, reference=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  57                 return helpString;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "ChimeraCheckCommand", "getHelpString");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 ChimeraCheckCommand::ChimeraCheckCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["chimera"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  74                 exit(1);
  75         }
  76 }
  77 //***************************************************************************************************************
  78 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); 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.check");
  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
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 it = parameters.find("reference");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         string path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 113                                 }
 114                         }
 115
 116                         //check for required parameters
 117                         fastafile = validParameter.validFile(parameters, "fasta", false);
 118                         if (fastafile == "not found") {
 119                                 //if there is a current fasta file, use it
 120                                 string filename = m->getFastaFile();
 121                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 122                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 123                         }else {
 124                                 m->splitAtDash(fastafile, fastaFileNames);
 125
 126                                 //go through files and make sure they are good, if not, then disregard them
 127                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 128
 129                                         bool ignore = false;
 130                                         if (fastaFileNames[i] == "current") {
 131                                                 fastaFileNames[i] = m->getFastaFile();
 132                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 133                                                 else {
 134                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 135                                                         //erase from file list
 136                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 137                                                         i--;
 138                                                 }
 139                                         }
 140
 141                                         if (!ignore) {
 142
 143
 144                                                 if (inputDir != "") {
 145                                                         string path = m->hasPath(fastaFileNames[i]);
 146                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 148                                                 }
 149
 150                                                 int ableToOpen;
 151                                                 ifstream in;
 152
 153                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 154
 155                                                 //if you can't open it, try default location
 156                                                 if (ableToOpen == 1) {
 157                                                         if (m->getDefaultPath() != "") { //default path is set
 158                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 159                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 160                                                                 ifstream in2;
 161                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 162                                                                 in2.close();
 163                                                                 fastaFileNames[i] = tryPath;
 164                                                         }
 165                                                 }
 166
 167                                                 //if you can't open it, try default location
 168                                                 if (ableToOpen == 1) {
 169                                                         if (m->getOutputDir() != "") { //default path is set
 170                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 171                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 172                                                                 ifstream in2;
 173                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 174                                                                 in2.close();
 175                                                                 fastaFileNames[i] = tryPath;
 176                                                         }
 177                                                 }
 178
 179                                                 in.close();
 180
 181                                                 if (ableToOpen == 1) {
 182                                                         m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 183                                                         //erase from file list
 184                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 185                                                         i--;
 186                                                 }else {
 187                                                         m->setFastaFile(fastaFileNames[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.check command."); m->mothurOutEndLine(); abort = true;  }
 202
 203                         namefile = validParameter.validFile(parameters, "name", false);
 204                         if (namefile == "not found") { namefile = ""; }
 205                         else {
 206                                 m->splitAtDash(namefile, nameFileNames);
 207
 208                                 //go through files and make sure they are good, if not, then disregard them
 209                                 for (int i = 0; i < nameFileNames.size(); i++) {
 210
 211                                         bool ignore = false;
 212                                         if (nameFileNames[i] == "current") {
 213                                                 nameFileNames[i] = m->getNameFile();
 214                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 215                                                 else {
 216                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 217                                                         //erase from file list
 218                                                         nameFileNames.erase(nameFileNames.begin()+i);
 219                                                         i--;
 220                                                 }
 221                                         }
 222
 223                                         if (!ignore) {
 224
 225                                                 if (inputDir != "") {
 226                                                         string path = m->hasPath(nameFileNames[i]);
 227                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 228                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 229                                                 }
 230
 231                                                 int ableToOpen;
 232                                                 ifstream in;
 233
 234                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 235
 236                                                 //if you can't open it, try default location
 237                                                 if (ableToOpen == 1) {
 238                                                         if (m->getDefaultPath() != "") { //default path is set
 239                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 240                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 241                                                                 ifstream in2;
 242                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 243                                                                 in2.close();
 244                                                                 nameFileNames[i] = tryPath;
 245                                                         }
 246                                                 }
 247
 248                                                 //if you can't open it, try default location
 249                                                 if (ableToOpen == 1) {
 250                                                         if (m->getOutputDir() != "") { //default path is set
 251                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 252                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 253                                                                 ifstream in2;
 254                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 255                                                                 in2.close();
 256                                                                 nameFileNames[i] = tryPath;
 257                                                         }
 258                                                 }
 259
 260                                                 in.close();
 261
 262                                                 if (ableToOpen == 1) {
 263                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 264                                                         //erase from file list
 265                                                         nameFileNames.erase(nameFileNames.begin()+i);
 266                                                         i--;
 267                                                 }else {
 268                                                         m->setNameFile(nameFileNames[i]);
 269                                                 }
 270                                         }
 271                                 }
 272
 273                                 //make sure there is at least one valid file left
 274                                 if (nameFileNames.size() != 0) {
 275                                         if (nameFileNames.size() != fastaFileNames.size()) {
 276                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 277                                                  abort = true;
 278                                         }
 279                                 }
 280                         }
 281
 282                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 283                         m->setProcessors(temp);
 284                         convert(temp, processors);
 285
 286                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 287                         convert(temp, ksize);
 288
 289                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 290                         svg = m->isTrue(temp);
 291                         if (nameFileNames.size() != 0) { svg = true; }
 292
 293                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 294                         convert(temp, increment);
 295                 }
 296         }
 297         catch(exception& e) {
 298                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 299                 exit(1);
 300         }
 301 }
 302 //***************************************************************************************************************
 303
 304 int ChimeraCheckCommand::execute(){
 305         try{
 306
 307                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 308
 309                 for (int i = 0; i < fastaFileNames.size(); i++) {
 310
 311                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 312
 313                         int start = time(NULL);
 314
 315                         string thisNameFile = "";
 316                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 317
 318                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 319
 320                         if (m->control_pressed) { delete chimera;       return 0;       }
 321
 322                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 323                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 324                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 325
 326                 #ifdef USE_MPI
 327
 328                                 int pid, numSeqsPerProcessor;
 329                                 int tag = 2001;
 330                                 vector<unsigned long int> MPIPos;
 331
 332                                 MPI_Status status;
 333                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 334                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 335
 336                                 MPI_File inMPI;
 337                                 MPI_File outMPI;
 338
 339                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 340                                 int inMode=MPI_MODE_RDONLY;
 341
 342                                 char outFilename[1024];
 343                                 strcpy(outFilename, outputFileName.c_str());
 344
 345                                 char inFileName[1024];
 346                                 strcpy(inFileName, fastaFileNames[i].c_str());
 347
 348                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 349                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 350
 351                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } outputTypes.clear(); delete chimera; return 0;  }
 352
 353                                 if (pid == 0) { //you are the root process
 354                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 355
 356                                         //send file positions to all processes
 357                                         for(int j = 1; j < processors; j++) {
 358                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 359                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 360                                         }
 361
 362                                         //figure out how many sequences you have to align
 363                                         numSeqsPerProcessor = numSeqs / processors;
 364                                         int startIndex =  pid * numSeqsPerProcessor;
 365                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 366
 367
 368                                         //align your part
 369                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 370
 371                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); }   outputTypes.clear(); delete chimera; return 0;  }
 372
 373                                         //wait on chidren
 374                                         for(int j = 1; j < processors; j++) {
 375                                                 char buf[5];
 376                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 377                                         }
 378                                 }else{ //you are a child process
 379                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 380                                         MPIPos.resize(numSeqs+1);
 381                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 382
 383                                         //figure out how many sequences you have to align
 384                                         numSeqsPerProcessor = numSeqs / processors;
 385                                         int startIndex =  pid * numSeqsPerProcessor;
 386                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 387
 388                                         //align your part
 389                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 390
 391                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  outputTypes.clear(); delete chimera; return 0;  }
 392
 393                                         //tell parent you are done.
 394                                         char buf[5];
 395                                         strcpy(buf, "done");
 396                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 397                                 }
 398
 399                                 //close files
 400                                 MPI_File_close(&inMPI);
 401                                 MPI_File_close(&outMPI);
 402                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 403                 #else
 404
 405                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 406
 407                         for (int s = 0; s < (positions.size()-1); s++) {
 408                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 409                         }
 410
 411                         //break up file
 412                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 413                                 if(processors == 1){
 414                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 415
 416                                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); } for (int j = 0; j < lines.size(); j++) {  delete lines[j];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 417
 418                                 }else{
 419                                         processIDS.resize(0);
 420
 421                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 422
 423                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 424
 425                                         //append output files
 426                                         for(int j=1;j<processors;j++){
 427                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 428                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 429                                         }
 430
 431                                         if (m->control_pressed) {
 432                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 433                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 434                                                 delete chimera;
 435                                                 return 0;
 436                                         }
 437                                 }
 438
 439                         #else
 440                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 441
 442                                 if (m->control_pressed) { for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear(); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear(); delete chimera; return 0; }
 443                         #endif
 444                 #endif
 445                         delete chimera;
 446                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 447
 448                         m->mothurOutEndLine(); m->mothurOut("This method does not determine if a sequence is chimeric, but allows you to make that determination based on the IS values."); m->mothurOutEndLine();
 449                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 450
 451                 }
 452
 453                 m->mothurOutEndLine();
 454                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 455                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 456                 m->mothurOutEndLine();
 457
 458                 return 0;
 459
 460         }
 461         catch(exception& e) {
 462                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 463                 exit(1);
 464         }
 465 }
 466 //**********************************************************************************************************************
 467
 468 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 469         try {
 470                 ofstream out;
 471                 m->openOutputFile(outputFName, out);
 472
 473                 ofstream out2;
 474
 475                 ifstream inFASTA;
 476                 m->openInputFile(filename, inFASTA);
 477
 478                 inFASTA.seekg(filePos->start);
 479
 480                 bool done = false;
 481                 int count = 0;
 482
 483                 while (!done) {
 484
 485                         if (m->control_pressed) {       return 1;       }
 486
 487                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 488
 489                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 490                                 //find chimeras
 491                                 chimera->getChimeras(candidateSeq);
 492
 493                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 494
 495                                 //print results
 496                                 chimera->print(out, out2);
 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                 inFASTA.close();
 515
 516                 return count;
 517         }
 518         catch(exception& e) {
 519                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 520                 exit(1);
 521         }
 522 }
 523 //**********************************************************************************************************************
 524 #ifdef USE_MPI
 525 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 526         try {
 527                 MPI_File outAccMPI;
 528                 MPI_Status status;
 529                 int pid;
 530                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 531
 532                 for(int i=0;i<num;i++){
 533
 534                         if (m->control_pressed) { return 0; }
 535
 536                         //read next sequence
 537                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 538
 539                         char* buf4 = new char[length];
 540                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 541
 542                         string tempBuf = buf4;
 543                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 544                         istringstream iss (tempBuf,istringstream::in);
 545                         delete buf4;
 546
 547                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 548
 549                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 550                                 //find chimeras
 551                                 chimera->getChimeras(candidateSeq);
 552
 553                                 //print results
 554                                 chimera->print(outMPI, outAccMPI);
 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                 return 0;
 565         }
 566         catch(exception& e) {
 567                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 568                 exit(1);
 569         }
 570 }
 571 #endif
 572
 573 /**************************************************************************************************/
 574
 575 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 576         try {
 577 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 578                 int process = 0;
 579                 int num = 0;
 580
 581                 //loop through and create all the processes you want
 582                 while (process != processors) {
 583                         int pid = fork();
 584
 585                         if (pid > 0) {
 586                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 587                                 process++;
 588                         }else if (pid == 0){
 589                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 590
 591                                 //pass numSeqs to parent
 592                                 ofstream out;
 593                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 594                                 m->openOutputFile(tempFile, out);
 595                                 out << num << endl;
 596                                 out.close();
 597
 598                                 exit(0);
 599                         }else {
 600                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 601                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 602                                 exit(0);
 603                         }
 604                 }
 605
 606                 //force parent to wait until all the processes are done
 607                 for (int i=0;i<processors;i++) {
 608                         int temp = processIDS[i];
 609                         wait(&temp);
 610                 }
 611
 612                 for (int i = 0; i < processIDS.size(); i++) {
 613                         ifstream in;
 614                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 615                         m->openInputFile(tempFile, in);
 616                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 617                         in.close(); remove(tempFile.c_str());
 618                 }
 619
 620                 return num;
 621 #endif
 622         }
 623         catch(exception& e) {
 624                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 625                 exit(1);
 626         }
 627 }
 628 /**************************************************************************************************/
 629
 630