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
  85                 else {
  86                         vector<string> myArray = setParameters();
  87
  88                         OptionParser parser(option);
  89                         map<string,string> parameters = parser.getParameters();
  90
  91                         ValidParameters validParameter("chimera.check");
  92                         map<string,string>::iterator it;
  93
  94                         //check to make sure all parameters are valid for command
  95                         for (it = parameters.begin(); it != parameters.end(); it++) {
  96                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  97                         }
  98
  99                         vector<string> tempOutNames;
 100                         outputTypes["chimera"] = tempOutNames;
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 it = parameters.find("reference");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         string path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 112                                 }
 113                         }
 114
 115                         //check for required parameters
 116                         fastafile = validParameter.validFile(parameters, "fasta", false);
 117                         if (fastafile == "not found") {
 118                                 //if there is a current fasta file, use it
 119                                 string filename = m->getFastaFile();
 120                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 121                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 122                         }else {
 123                                 m->splitAtDash(fastafile, fastaFileNames);
 124
 125                                 //go through files and make sure they are good, if not, then disregard them
 126                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 127
 128                                         bool ignore = false;
 129                                         if (fastaFileNames[i] == "current") {
 130                                                 fastaFileNames[i] = m->getFastaFile();
 131                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 132                                                 else {
 133                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 134                                                         //erase from file list
 135                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 136                                                         i--;
 137                                                 }
 138                                         }
 139
 140                                         if (!ignore) {
 141
 142
 143                                                 if (inputDir != "") {
 144                                                         string path = m->hasPath(fastaFileNames[i]);
 145                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 147                                                 }
 148
 149                                                 int ableToOpen;
 150                                                 ifstream in;
 151
 152                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 153
 154                                                 //if you can't open it, try default location
 155                                                 if (ableToOpen == 1) {
 156                                                         if (m->getDefaultPath() != "") { //default path is set
 157                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 158                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 159                                                                 ifstream in2;
 160                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 161                                                                 in2.close();
 162                                                                 fastaFileNames[i] = tryPath;
 163                                                         }
 164                                                 }
 165
 166                                                 //if you can't open it, try default location
 167                                                 if (ableToOpen == 1) {
 168                                                         if (m->getOutputDir() != "") { //default path is set
 169                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 170                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 171                                                                 ifstream in2;
 172                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 173                                                                 in2.close();
 174                                                                 fastaFileNames[i] = tryPath;
 175                                                         }
 176                                                 }
 177
 178                                                 in.close();
 179
 180                                                 if (ableToOpen == 1) {
 181                                                         m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 182                                                         //erase from file list
 183                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 184                                                         i--;
 185                                                 }
 186                                         }
 187                                 }
 188
 189                                 //make sure there is at least one valid file left
 190                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 191                         }
 192
 193                         //if the user changes the output directory command factory will send this info to us in the output parameter
 194                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 195
 196                         templatefile = validParameter.validFile(parameters, "reference", true);
 197                         if (templatefile == "not open") { abort = true; }
 198                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("reference is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
 199
 200                         namefile = validParameter.validFile(parameters, "name", false);
 201                         if (namefile == "not found") { namefile = ""; }
 202                         else {
 203                                 m->splitAtDash(namefile, nameFileNames);
 204
 205                                 //go through files and make sure they are good, if not, then disregard them
 206                                 for (int i = 0; i < nameFileNames.size(); i++) {
 207
 208                                         bool ignore = false;
 209                                         if (nameFileNames[i] == "current") {
 210                                                 nameFileNames[i] = m->getNameFile();
 211                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 212                                                 else {
 213                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 214                                                         //erase from file list
 215                                                         nameFileNames.erase(nameFileNames.begin()+i);
 216                                                         i--;
 217                                                 }
 218                                         }
 219
 220                                         if (!ignore) {
 221
 222                                                 if (inputDir != "") {
 223                                                         string path = m->hasPath(nameFileNames[i]);
 224                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 225                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 226                                                 }
 227
 228                                                 int ableToOpen;
 229                                                 ifstream in;
 230
 231                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 232
 233                                                 //if you can't open it, try default location
 234                                                 if (ableToOpen == 1) {
 235                                                         if (m->getDefaultPath() != "") { //default path is set
 236                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 237                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 238                                                                 ifstream in2;
 239                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 240                                                                 in2.close();
 241                                                                 nameFileNames[i] = tryPath;
 242                                                         }
 243                                                 }
 244
 245                                                 //if you can't open it, try default location
 246                                                 if (ableToOpen == 1) {
 247                                                         if (m->getOutputDir() != "") { //default path is set
 248                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 249                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 250                                                                 ifstream in2;
 251                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 252                                                                 in2.close();
 253                                                                 nameFileNames[i] = tryPath;
 254                                                         }
 255                                                 }
 256
 257                                                 in.close();
 258
 259                                                 if (ableToOpen == 1) {
 260                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 261                                                         //erase from file list
 262                                                         nameFileNames.erase(nameFileNames.begin()+i);
 263                                                         i--;
 264                                                 }
 265                                         }
 266                                 }
 267
 268                                 //make sure there is at least one valid file left
 269                                 if (nameFileNames.size() != 0) {
 270                                         if (nameFileNames.size() != fastaFileNames.size()) {
 271                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 272                                                  abort = true;
 273                                         }
 274                                 }
 275                         }
 276
 277                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 278                         m->setProcessors(temp);
 279                         convert(temp, processors);
 280
 281                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 282                         convert(temp, ksize);
 283
 284                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 285                         svg = m->isTrue(temp);
 286                         if (nameFileNames.size() != 0) { svg = true; }
 287
 288                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 289                         convert(temp, increment);
 290                 }
 291         }
 292         catch(exception& e) {
 293                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 294                 exit(1);
 295         }
 296 }
 297 //***************************************************************************************************************
 298
 299 int ChimeraCheckCommand::execute(){
 300         try{
 301
 302                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 303
 304                 for (int i = 0; i < fastaFileNames.size(); i++) {
 305
 306                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 307
 308                         int start = time(NULL);
 309
 310                         string thisNameFile = "";
 311                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 312
 313                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 314
 315                         if (m->control_pressed) { delete chimera;       return 0;       }
 316
 317                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 318                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 319                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 320
 321                 #ifdef USE_MPI
 322
 323                                 int pid, numSeqsPerProcessor;
 324                                 int tag = 2001;
 325                                 vector<unsigned long int> MPIPos;
 326
 327                                 MPI_Status status;
 328                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 329                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 330
 331                                 MPI_File inMPI;
 332                                 MPI_File outMPI;
 333
 334                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 335                                 int inMode=MPI_MODE_RDONLY;
 336
 337                                 char outFilename[1024];
 338                                 strcpy(outFilename, outputFileName.c_str());
 339
 340                                 char inFileName[1024];
 341                                 strcpy(inFileName, fastaFileNames[i].c_str());
 342
 343                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 344                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 345
 346                                 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;  }
 347
 348                                 if (pid == 0) { //you are the root process
 349                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 350
 351                                         //send file positions to all processes
 352                                         for(int j = 1; j < processors; j++) {
 353                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 354                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 355                                         }
 356
 357                                         //figure out how many sequences you have to align
 358                                         numSeqsPerProcessor = numSeqs / processors;
 359                                         int startIndex =  pid * numSeqsPerProcessor;
 360                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 361
 362
 363                                         //align your part
 364                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 365
 366                                         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;  }
 367
 368                                         //wait on chidren
 369                                         for(int j = 1; j < processors; j++) {
 370                                                 char buf[5];
 371                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 372                                         }
 373                                 }else{ //you are a child process
 374                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 375                                         MPIPos.resize(numSeqs+1);
 376                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 377
 378                                         //figure out how many sequences you have to align
 379                                         numSeqsPerProcessor = numSeqs / processors;
 380                                         int startIndex =  pid * numSeqsPerProcessor;
 381                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 382
 383                                         //align your part
 384                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 385
 386                                         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;  }
 387
 388                                         //tell parent you are done.
 389                                         char buf[5];
 390                                         strcpy(buf, "done");
 391                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 392                                 }
 393
 394                                 //close files
 395                                 MPI_File_close(&inMPI);
 396                                 MPI_File_close(&outMPI);
 397                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 398                 #else
 399
 400                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[i], processors);
 401
 402                         for (int s = 0; s < (positions.size()-1); s++) {
 403                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 404                         }
 405
 406                         //break up file
 407                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 408                                 if(processors == 1){
 409                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 410
 411                                         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; }
 412
 413                                 }else{
 414                                         processIDS.resize(0);
 415
 416                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 417
 418                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 419
 420                                         //append output files
 421                                         for(int j=1;j<processors;j++){
 422                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 423                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 424                                         }
 425
 426                                         if (m->control_pressed) {
 427                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 428                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 429                                                 delete chimera;
 430                                                 return 0;
 431                                         }
 432                                 }
 433
 434                         #else
 435                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 436
 437                                 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; }
 438                         #endif
 439                 #endif
 440                         delete chimera;
 441                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 442
 443                         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();
 444                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 445
 446                 }
 447
 448                 m->mothurOutEndLine();
 449                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 450                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 451                 m->mothurOutEndLine();
 452
 453                 return 0;
 454
 455         }
 456         catch(exception& e) {
 457                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 458                 exit(1);
 459         }
 460 }
 461 //**********************************************************************************************************************
 462
 463 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 464         try {
 465                 ofstream out;
 466                 m->openOutputFile(outputFName, out);
 467
 468                 ofstream out2;
 469
 470                 ifstream inFASTA;
 471                 m->openInputFile(filename, inFASTA);
 472
 473                 inFASTA.seekg(filePos->start);
 474
 475                 bool done = false;
 476                 int count = 0;
 477
 478                 while (!done) {
 479
 480                         if (m->control_pressed) {       return 1;       }
 481
 482                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 483
 484                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 485                                 //find chimeras
 486                                 chimera->getChimeras(candidateSeq);
 487
 488                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 489
 490                                 //print results
 491                                 chimera->print(out, out2);
 492                         }
 493                         delete candidateSeq;
 494
 495                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 496                                 unsigned long int pos = inFASTA.tellg();
 497                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 498                         #else
 499                                 if (inFASTA.eof()) { break; }
 500                         #endif
 501
 502                         //report progress
 503                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 504                 }
 505                 //report progress
 506                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 507
 508                 out.close();
 509                 inFASTA.close();
 510
 511                 return count;
 512         }
 513         catch(exception& e) {
 514                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 515                 exit(1);
 516         }
 517 }
 518 //**********************************************************************************************************************
 519 #ifdef USE_MPI
 520 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 521         try {
 522                 MPI_File outAccMPI;
 523                 MPI_Status status;
 524                 int pid;
 525                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 526
 527                 for(int i=0;i<num;i++){
 528
 529                         if (m->control_pressed) { return 0; }
 530
 531                         //read next sequence
 532                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 533
 534                         char* buf4 = new char[length];
 535                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 536
 537                         string tempBuf = buf4;
 538                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 539                         istringstream iss (tempBuf,istringstream::in);
 540                         delete buf4;
 541
 542                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 543
 544                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 545                                 //find chimeras
 546                                 chimera->getChimeras(candidateSeq);
 547
 548                                 //print results
 549                                 chimera->print(outMPI, outAccMPI);
 550                         }
 551                         delete candidateSeq;
 552
 553                         //report progress
 554                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 555                 }
 556                 //report progress
 557                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 558
 559                 return 0;
 560         }
 561         catch(exception& e) {
 562                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 563                 exit(1);
 564         }
 565 }
 566 #endif
 567
 568 /**************************************************************************************************/
 569
 570 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 571         try {
 572 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 573                 int process = 0;
 574                 int num = 0;
 575
 576                 //loop through and create all the processes you want
 577                 while (process != processors) {
 578                         int pid = fork();
 579
 580                         if (pid > 0) {
 581                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 582                                 process++;
 583                         }else if (pid == 0){
 584                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 585
 586                                 //pass numSeqs to parent
 587                                 ofstream out;
 588                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 589                                 m->openOutputFile(tempFile, out);
 590                                 out << num << endl;
 591                                 out.close();
 592
 593                                 exit(0);
 594                         }else {
 595                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 596                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 597                                 exit(0);
 598                         }
 599                 }
 600
 601                 //force parent to wait until all the processes are done
 602                 for (int i=0;i<processors;i++) {
 603                         int temp = processIDS[i];
 604                         wait(&temp);
 605                 }
 606
 607                 for (int i = 0; i < processIDS.size(); i++) {
 608                         ifstream in;
 609                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 610                         m->openInputFile(tempFile, in);
 611                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 612                         in.close(); remove(tempFile.c_str());
 613                 }
 614
 615                 return num;
 616 #endif
 617         }
 618         catch(exception& e) {
 619                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 620                 exit(1);
 621         }
 622 }
 623 /**************************************************************************************************/
 624
 625