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 #include "referencedb.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraCheckCommand::setParameters(){
  15         try {
  16                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter psvg("svg", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psvg);
  20                 CommandParameter pincrement("increment", "Number", "", "10", "", "", "",false,false); parameters.push_back(pincrement);
  21                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  22                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  26
  27                 vector<string> myArray;
  28                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  29                 return myArray;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "ChimeraCheckCommand", "setParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 string ChimeraCheckCommand::getHelpString(){
  38         try {
  39                 string helpString = "";
  40                 helpString += "The chimera.check command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  41                 helpString += "This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n";
  42                 helpString += "The chimera.check command parameters are fasta, reference, processors, ksize, increment, svg and name.\n";
  43                 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";
  44                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  45                 helpString += "The reference parameter allows you to enter a reference file containing known non-chimeric sequences, and is required. \n";
  46                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  47 #ifdef USE_MPI
  48                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  49 #endif
  50                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n";
  51                 helpString += "The ksize parameter allows you to input kmersize, default is 7. \n";
  52                 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";
  53                 helpString += "The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n";
  54                 helpString += "You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n";
  55                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  56                 helpString += "The chimera.check command should be in the following format: \n";
  57                 helpString += "chimera.check(fasta=yourFastaFile, reference=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n";
  58                 helpString += "Example: chimera.check(fasta=AD.fasta, reference=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n";
  59                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  60                 return helpString;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "ChimeraCheckCommand", "getHelpString");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 ChimeraCheckCommand::ChimeraCheckCommand(){
  69         try {
  70                 abort = true; calledHelp = true;
  71                 setParameters();
  72                 vector<string> tempOutNames;
  73                 outputTypes["chimera"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  77                 exit(1);
  78         }
  79 }
  80 //***************************************************************************************************************
  81 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  82         try {
  83                 abort = false; calledHelp = false;
  84                 ReferenceDB* rdb = ReferenceDB::getInstance();
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string,string> parameters = parser.getParameters();
  95
  96                         ValidParameters validParameter("chimera.check");
  97                         map<string,string>::iterator it;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it = parameters.begin(); it != parameters.end(); it++) {
 101                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 102                         }
 103
 104                         vector<string> tempOutNames;
 105                         outputTypes["chimera"] = tempOutNames;
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 it = parameters.find("reference");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         string 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") {
 123                                 //if there is a current fasta file, use it
 124                                 string filename = m->getFastaFile();
 125                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 126                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 127                         }else {
 128                                 m->splitAtDash(fastafile, fastaFileNames);
 129
 130                                 //go through files and make sure they are good, if not, then disregard them
 131                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 132
 133                                         bool ignore = false;
 134                                         if (fastaFileNames[i] == "current") {
 135                                                 fastaFileNames[i] = m->getFastaFile();
 136                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 137                                                 else {
 138                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 139                                                         //erase from file list
 140                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 141                                                         i--;
 142                                                 }
 143                                         }
 144
 145                                         if (!ignore) {
 146
 147
 148                                                 if (inputDir != "") {
 149                                                         string path = m->hasPath(fastaFileNames[i]);
 150                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 151                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 152                                                 }
 153
 154                                                 int ableToOpen;
 155                                                 ifstream in;
 156
 157                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 158
 159                                                 //if you can't open it, try default location
 160                                                 if (ableToOpen == 1) {
 161                                                         if (m->getDefaultPath() != "") { //default path is set
 162                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 163                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 164                                                                 ifstream in2;
 165                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 166                                                                 in2.close();
 167                                                                 fastaFileNames[i] = tryPath;
 168                                                         }
 169                                                 }
 170
 171                                                 //if you can't open it, try default location
 172                                                 if (ableToOpen == 1) {
 173                                                         if (m->getOutputDir() != "") { //default path is set
 174                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 175                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 176                                                                 ifstream in2;
 177                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                                 in2.close();
 179                                                                 fastaFileNames[i] = tryPath;
 180                                                         }
 181                                                 }
 182
 183                                                 in.close();
 184
 185                                                 if (ableToOpen == 1) {
 186                                                         m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 187                                                         //erase from file list
 188                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 189                                                         i--;
 190                                                 }else {
 191                                                         m->setFastaFile(fastaFileNames[i]);
 192                                                 }
 193                                         }
 194                                 }
 195
 196                                 //make sure there is at least one valid file left
 197                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 198                         }
 199
 200                         //if the user changes the output directory command factory will send this info to us in the output parameter
 201                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 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, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 287                         save = m->isTrue(temp);
 288                         rdb->save = save;
 289                         if (save) { //clear out old references
 290                                 rdb->clearMemory();
 291                         }
 292
 293                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 294                         templatefile = validParameter.validFile(parameters, "reference", true);
 295                         if (templatefile == "not found") {
 296                                 //check for saved reference sequences
 297                                 if (rdb->referenceSeqs.size() != 0) {
 298                                         templatefile = "saved";
 299                                 }else {
 300                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 301                                         m->mothurOutEndLine();
 302                                         abort = true;
 303                                 }
 304                         }else if (templatefile == "not open") { abort = true; }
 305                         else {  if (save) {     rdb->setSavedReference(templatefile);   }       }
 306
 307
 308                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 309                         convert(temp, ksize);
 310
 311                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 312                         svg = m->isTrue(temp);
 313                         if (nameFileNames.size() != 0) { svg = true; }
 314
 315                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 316                         convert(temp, increment);
 317                 }
 318         }
 319         catch(exception& e) {
 320                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 321                 exit(1);
 322         }
 323 }
 324 //***************************************************************************************************************
 325
 326 int ChimeraCheckCommand::execute(){
 327         try{
 328
 329                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 330
 331                 for (int i = 0; i < fastaFileNames.size(); i++) {
 332
 333                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 334
 335                         int start = time(NULL);
 336
 337                         string thisNameFile = "";
 338                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 339
 340                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 341
 342                         if (m->control_pressed) { delete chimera;       return 0;       }
 343
 344                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 345                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 346                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 347
 348                 #ifdef USE_MPI
 349
 350                                 int pid, numSeqsPerProcessor;
 351                                 int tag = 2001;
 352                                 vector<unsigned long int> MPIPos;
 353
 354                                 MPI_Status status;
 355                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 356                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 357
 358                                 MPI_File inMPI;
 359                                 MPI_File outMPI;
 360
 361                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 362                                 int inMode=MPI_MODE_RDONLY;
 363
 364                                 char outFilename[1024];
 365                                 strcpy(outFilename, outputFileName.c_str());
 366
 367                                 char inFileName[1024];
 368                                 strcpy(inFileName, fastaFileNames[i].c_str());
 369
 370                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 371                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 372
 373                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        } outputTypes.clear(); delete chimera; return 0;  }
 374
 375                                 if (pid == 0) { //you are the root process
 376                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 377
 378                                         //send file positions to all processes
 379                                         for(int j = 1; j < processors; j++) {
 380                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 381                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 382                                         }
 383
 384                                         //figure out how many sequences you have to align
 385                                         numSeqsPerProcessor = numSeqs / processors;
 386                                         int startIndex =  pid * numSeqsPerProcessor;
 387                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 388
 389
 390                                         //align your part
 391                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 392
 393                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        }   outputTypes.clear(); delete chimera; return 0;  }
 394
 395                                         //wait on chidren
 396                                         for(int j = 1; j < processors; j++) {
 397                                                 char buf[5];
 398                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 399                                         }
 400                                 }else{ //you are a child process
 401                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 402                                         MPIPos.resize(numSeqs+1);
 403                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 404
 405                                         //figure out how many sequences you have to align
 406                                         numSeqsPerProcessor = numSeqs / processors;
 407                                         int startIndex =  pid * numSeqsPerProcessor;
 408                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 409
 410                                         //align your part
 411                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 412
 413                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  outputTypes.clear(); delete chimera; return 0;  }
 414
 415                                         //tell parent you are done.
 416                                         char buf[5];
 417                                         strcpy(buf, "done");
 418                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 419                                 }
 420
 421                                 //close files
 422                                 MPI_File_close(&inMPI);
 423                                 MPI_File_close(&outMPI);
 424                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 425                 #else
 426
 427                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 428
 429                         for (int s = 0; s < (positions.size()-1); s++) {
 430                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 431                         }
 432
 433                         //break up file
 434                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 435                                 if(processors == 1){
 436                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 437
 438                                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } for (int j = 0; j < lines.size(); j++) {  delete lines[j];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 439
 440                                 }else{
 441                                         processIDS.resize(0);
 442
 443                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 444
 445                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 446
 447                                         //append output files
 448                                         for(int j=1;j<processors;j++){
 449                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 450                                                 m->mothurRemove((outputFileName + toString(processIDS[j]) + ".temp"));
 451                                         }
 452
 453                                         if (m->control_pressed) {
 454                                                 for (int j = 0; j < outputNames.size(); j++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear();
 455                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 456                                                 delete chimera;
 457                                                 return 0;
 458                                         }
 459                                 }
 460
 461                         #else
 462                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 463
 464                                 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++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear(); delete chimera; return 0; }
 465                         #endif
 466                 #endif
 467                         delete chimera;
 468                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 469
 470                         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();
 471                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 472
 473                 }
 474
 475                 m->mothurOutEndLine();
 476                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 477                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 478                 m->mothurOutEndLine();
 479
 480                 return 0;
 481
 482         }
 483         catch(exception& e) {
 484                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 485                 exit(1);
 486         }
 487 }
 488 //**********************************************************************************************************************
 489
 490 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 491         try {
 492                 ofstream out;
 493                 m->openOutputFile(outputFName, out);
 494
 495                 ofstream out2;
 496
 497                 ifstream inFASTA;
 498                 m->openInputFile(filename, inFASTA);
 499
 500                 inFASTA.seekg(filePos->start);
 501
 502                 bool done = false;
 503                 int count = 0;
 504
 505                 while (!done) {
 506
 507                         if (m->control_pressed) {       return 1;       }
 508
 509                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 510
 511                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 512                                 //find chimeras
 513                                 chimera->getChimeras(candidateSeq);
 514
 515                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 516
 517                                 //print results
 518                                 chimera->print(out, out2);
 519                         }
 520                         delete candidateSeq;
 521
 522                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 523                                 unsigned long int pos = inFASTA.tellg();
 524                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 525                         #else
 526                                 if (inFASTA.eof()) { break; }
 527                         #endif
 528
 529                         //report progress
 530                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 531                 }
 532                 //report progress
 533                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 534
 535                 out.close();
 536                 inFASTA.close();
 537
 538                 return count;
 539         }
 540         catch(exception& e) {
 541                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 542                 exit(1);
 543         }
 544 }
 545 //**********************************************************************************************************************
 546 #ifdef USE_MPI
 547 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 548         try {
 549                 MPI_File outAccMPI;
 550                 MPI_Status status;
 551                 int pid;
 552                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 553
 554                 for(int i=0;i<num;i++){
 555
 556                         if (m->control_pressed) { return 0; }
 557
 558                         //read next sequence
 559                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 560
 561                         char* buf4 = new char[length];
 562                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 563
 564                         string tempBuf = buf4;
 565                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 566                         istringstream iss (tempBuf,istringstream::in);
 567                         delete buf4;
 568
 569                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 570
 571                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 572                                 //find chimeras
 573                                 chimera->getChimeras(candidateSeq);
 574
 575                                 //print results
 576                                 chimera->print(outMPI, outAccMPI);
 577                         }
 578                         delete candidateSeq;
 579
 580                         //report progress
 581                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 582                 }
 583                 //report progress
 584                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 585
 586                 return 0;
 587         }
 588         catch(exception& e) {
 589                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 590                 exit(1);
 591         }
 592 }
 593 #endif
 594
 595 /**************************************************************************************************/
 596
 597 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 598         try {
 599 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 600                 int process = 0;
 601                 int num = 0;
 602
 603                 //loop through and create all the processes you want
 604                 while (process != processors) {
 605                         int pid = fork();
 606
 607                         if (pid > 0) {
 608                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 609                                 process++;
 610                         }else if (pid == 0){
 611                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 612
 613                                 //pass numSeqs to parent
 614                                 ofstream out;
 615                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 616                                 m->openOutputFile(tempFile, out);
 617                                 out << num << endl;
 618                                 out.close();
 619
 620                                 exit(0);
 621                         }else {
 622                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 623                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 624                                 exit(0);
 625                         }
 626                 }
 627
 628                 //force parent to wait until all the processes are done
 629                 for (int i=0;i<processors;i++) {
 630                         int temp = processIDS[i];
 631                         wait(&temp);
 632                 }
 633
 634                 for (int i = 0; i < processIDS.size(); i++) {
 635                         ifstream in;
 636                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 637                         m->openInputFile(tempFile, in);
 638                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 639                         in.close(); m->mothurRemove(tempFile);
 640                 }
 641
 642                 return num;
 643 #endif
 644         }
 645         catch(exception& e) {
 646                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 647                 exit(1);
 648         }
 649 }
 650 /**************************************************************************************************/
 651
 652