chimeraslayercommand.cpp

   1 /*
   2  *  chimeraslayercommand.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 "chimeraslayercommand.h"
  11 #include "chimeraslayer.h"
  12
  13
  14 //***************************************************************************************************************
  15
  16 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
  17         try {
  18                 abort = false;
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string Array[] =  {"fasta", "processors", "window", "template","numwanted", "ksize", "match","mismatch",
  26                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  27                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  28
  29                         OptionParser parser(option);
  30                         map<string,string> parameters = parser.getParameters();
  31
  32                         ValidParameters validParameter("chimera.slayer");
  33                         map<string,string>::iterator it;
  34
  35                         //check to make sure all parameters are valid for command
  36                         for (it = parameters.begin(); it != parameters.end(); it++) {
  37                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  38                         }
  39
  40                         //if the user changes the input directory command factory will send this info to us in the output parameter
  41                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  42                         if (inputDir == "not found"){   inputDir = "";          }
  43                         else {
  44                                 string path;
  45                                 it = parameters.find("template");
  46                                 //user has given a template file
  47                                 if(it != parameters.end()){
  48                                         path = hasPath(it->second);
  49                                         //if the user has not given a path then, add inputdir. else leave path alone.
  50                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  51                                 }
  52                         }
  53
  54
  55                         //check for required parameters
  56                         fastafile = validParameter.validFile(parameters, "fasta", false);
  57                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
  58                         else {
  59                                 splitAtDash(fastafile, fastaFileNames);
  60
  61                                 //go through files and make sure they are good, if not, then disregard them
  62                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  63                                         if (inputDir != "") {
  64                                                 string path = hasPath(fastaFileNames[i]);
  65                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  66                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
  67                                         }
  68
  69                                         int ableToOpen;
  70                                         ifstream in;
  71
  72                                         #ifdef USE_MPI
  73                                                 int pid;
  74                                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
  75                                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
  76
  77                                                 if (pid == 0) {
  78                                         #endif
  79
  80                                         ableToOpen = openInputFile(fastaFileNames[i], in, "noerror");
  81
  82                                         //if you can't open it, try default location
  83                                         if (ableToOpen == 1) {
  84                                                 if (m->getDefaultPath() != "") { //default path is set
  85                                                         string tryPath = m->getDefaultPath() + getSimpleName(fastaFileNames[i]);
  86                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
  87                                                         ableToOpen = openInputFile(tryPath, in, "noerror");
  88                                                         fastaFileNames[i] = tryPath;
  89                                                 }
  90                                         }
  91                                         in.close();
  92
  93                                         #ifdef USE_MPI
  94                                                         for (int j = 1; j < processors; j++) {
  95                                                                 MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
  96                                                         }
  97                                                 }else{
  98                                                         MPI_Status status;
  99                                                         MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 100                                                 }
 101
 102                                         #endif
 103
 104                                         if (ableToOpen == 1) {
 105                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 106                                                 //erase from file list
 107                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 108                                                 i--;
 109                                         }
 110                                 }
 111
 112                                 //make sure there is at least one valid file left
 113                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 114                         }
 115
 116                         //if the user changes the output directory command factory will send this info to us in the output parameter
 117                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 118                                 outputDir = "";
 119                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
 120                         }
 121
 122                         templatefile = validParameter.validFile(parameters, "template", true);
 123                         if (templatefile == "not open") { abort = true; }
 124                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 125
 126                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 127                         convert(temp, processors);
 128
 129                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 130                         convert(temp, ksize);
 131
 132                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 133                         convert(temp, window);
 134
 135                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 136                         convert(temp, match);
 137
 138                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 139                         convert(temp, mismatch);
 140
 141                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 142                         convert(temp, divR);
 143
 144                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 145                         convert(temp, minSimilarity);
 146
 147                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 148                         convert(temp, minCoverage);
 149
 150                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 151                         convert(temp, minBS);
 152
 153                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "100"; }
 154                         convert(temp, minSNP);
 155
 156                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 157                         convert(temp, parents);
 158
 159                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 160                         realign = isTrue(temp);
 161
 162                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 163
 164                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "100"; }
 165                         convert(temp, iters);
 166
 167                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 168                         convert(temp, increment);
 169
 170                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 171                         convert(temp, numwanted);
 172
 173                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 174                 }
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 178                 exit(1);
 179         }
 180 }
 181 //**********************************************************************************************************************
 182
 183 void ChimeraSlayerCommand::help(){
 184         try {
 185
 186                 m->mothurOut("The chimera.slayer command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 187                 m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
 188                 m->mothurOut("The chimera.slayer command parameters are fasta, template, processors, ksize, window, match, mismatch, divergence. minsim, mincov, minbs, minsnp, parents, search, iters, increment and numwanted.\n"); //realign,
 189                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 190                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n");
 191                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 192                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 193                 #ifdef USE_MPI
 194                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 195                 #endif
 196                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=50. \n");
 197                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n");
 198                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n");
 199                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n");
 200                 m->mothurOut("The match parameter allows you to reward matched bases in blast search, default is 5. \n");
 201                 m->mothurOut("The parents parameter allows you to select the number of potential parents to investigate from the numwanted best matches after rating them, default is 3. \n");
 202                 m->mothurOut("The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n");
 203                 m->mothurOut("The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n");
 204                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=100.\n");
 205                 m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
 206                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 207                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 208                 m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 100) \n");
 209                 m->mothurOut("The search parameter allows you to specify search method for finding the closest parent. Choices are distance, blast, and kmer, default distance. \n");
 210                 m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false.  \n");
 211                 m->mothurOut("The chimera.slayer command should be in the following format: \n");
 212                 m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
 213                 m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
 214                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "ChimeraSlayerCommand", "help");
 218                 exit(1);
 219         }
 220 }
 221
 222 //***************************************************************************************************************
 223
 224 ChimeraSlayerCommand::~ChimeraSlayerCommand(){  /*      do nothing      */      }
 225
 226 //***************************************************************************************************************
 227
 228 int ChimeraSlayerCommand::execute(){
 229         try{
 230
 231                 if (abort == true) { return 0; }
 232
 233                 for (int s = 0; s < fastaFileNames.size(); s++) {
 234
 235                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 236
 237                         int start = time(NULL);
 238
 239                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 240
 241                         string outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "slayer.chimeras";
 242                         string accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 243
 244                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0;    }
 245
 246                         if (chimera->getUnaligned()) {
 247                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 248                                 delete chimera;
 249                                 return 0;
 250                         }
 251                         templateSeqsLength = chimera->getLength();
 252
 253                 #ifdef USE_MPI
 254                         int pid, end, numSeqsPerProcessor;
 255                                 int tag = 2001;
 256                                 vector<unsigned long int> MPIPos;
 257
 258                                 MPI_Status status;
 259                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 260                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 261
 262                                 MPI_File inMPI;
 263                                 MPI_File outMPI;
 264                                 MPI_File outMPIAccnos;
 265
 266                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 267                                 int inMode=MPI_MODE_RDONLY;
 268
 269                                 char outFilename[1024];
 270                                 strcpy(outFilename, outputFileName.c_str());
 271
 272                                 char outAccnosFilename[1024];
 273                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 274
 275                                 char inFileName[1024];
 276                                 strcpy(inFileName, fastaFileNames[s].c_str());
 277
 278                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 279                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 280                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 281
 282                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); }   delete chimera; return 0;  }
 283
 284                                 if (pid == 0) { //you are the root process
 285                                         m->mothurOutEndLine();
 286                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 287                                         m->mothurOutEndLine();
 288
 289                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 290
 291                                         //print header
 292                                         int length = outTemp.length();
 293                                         char* buf2 = new char[length];
 294                                         memcpy(buf2, outTemp.c_str(), length);
 295
 296                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 297                                         delete buf2;
 298
 299                                         MPIPos = setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 300
 301                                         //send file positions to all processes
 302                                         for(int i = 1; i < processors; i++) {
 303                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 304                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 305                                         }
 306
 307                                         //figure out how many sequences you have to align
 308                                         numSeqsPerProcessor = numSeqs / processors;
 309                                         int startIndex =  pid * numSeqsPerProcessor;
 310                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 311
 312                                         //do your part
 313                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 314
 315                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  delete chimera; return 0;  }
 316
 317                                 }else{ //you are a child process
 318                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 319                                         MPIPos.resize(numSeqs+1);
 320                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 321
 322                                         //figure out how many sequences you have to align
 323                                         numSeqsPerProcessor = numSeqs / processors;
 324                                         int startIndex =  pid * numSeqsPerProcessor;
 325                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 326
 327                                         //do your part
 328                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 329
 330                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 331                                 }
 332
 333                                 //close files
 334                                 MPI_File_close(&inMPI);
 335                                 MPI_File_close(&outMPI);
 336                                 MPI_File_close(&outMPIAccnos);
 337                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 338
 339                 #else
 340                         ofstream outHeader;
 341                         string tempHeader = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
 342                         openOutputFile(tempHeader, outHeader);
 343
 344                         chimera->printHeader(outHeader);
 345                         outHeader.close();
 346
 347                         //break up file
 348                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 349                                 if(processors == 1){
 350                                         ifstream inFASTA;
 351                                         openInputFile(fastaFileNames[s], inFASTA);
 352                                         getNumSeqs(inFASTA, numSeqs);
 353                                         inFASTA.close();
 354
 355                                         lines.push_back(new linePair(0, numSeqs));
 356
 357                                         driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 358
 359                                         if (m->control_pressed) {
 360                                                 remove(outputFileName.c_str());
 361                                                 remove(tempHeader.c_str());
 362                                                 remove(accnosFileName.c_str());
 363                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 364                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 365                                                 delete chimera;
 366                                                 return 0;
 367                                         }
 368
 369                                 }else{
 370                                         vector<unsigned long int> positions;
 371                                         processIDS.resize(0);
 372
 373                                         ifstream inFASTA;
 374                                         openInputFile(fastaFileNames[s], inFASTA);
 375
 376                                         string input;
 377                                         while(!inFASTA.eof()){
 378                                                 input = getline(inFASTA);
 379                                                 if (input.length() != 0) {
 380                                                         if(input[0] == '>'){    unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
 381                                                 }
 382                                         }
 383                                         inFASTA.close();
 384
 385                                         numSeqs = positions.size();
 386
 387                                         int numSeqsPerProcessor = numSeqs / processors;
 388
 389                                         for (int i = 0; i < processors; i++) {
 390                                                 unsigned long int startPos = positions[ i * numSeqsPerProcessor ];
 391                                                 if(i == processors - 1){
 392                                                         numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
 393                                                 }
 394                                                 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 395                                         }
 396
 397                                         createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 398
 399                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 400                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 401
 402                                         //append output files
 403                                         for(int i=1;i<processors;i++){
 404                                                 appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 405                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 406                                         }
 407
 408                                         //append output files
 409                                         for(int i=1;i<processors;i++){
 410                                                 appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 411                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 412                                         }
 413
 414                                         if (m->control_pressed) {
 415                                                 remove(outputFileName.c_str());
 416                                                 remove(accnosFileName.c_str());
 417                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 418                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 419                                                 delete chimera;
 420                                                 return 0;
 421                                         }
 422
 423                                 }
 424
 425                         #else
 426                                 ifstream inFASTA;
 427                                 openInputFile(fastaFileNames[s], inFASTA);
 428                                 getNumSeqs(inFASTA, numSeqs);
 429                                 inFASTA.close();
 430                                 lines.push_back(new linePair(0, numSeqs));
 431
 432                                 driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 433
 434                                 if (m->control_pressed) {
 435                                                 remove(outputFileName.c_str());
 436                                                 remove(tempHeader.c_str());
 437                                                 remove(accnosFileName.c_str());
 438                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 439                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 440                                                 delete chimera;
 441                                                 return 0;
 442                                 }
 443
 444                         #endif
 445
 446                         appendFiles(outputFileName, tempHeader);
 447
 448                         remove(outputFileName.c_str());
 449                         rename(tempHeader.c_str(), outputFileName.c_str());
 450
 451                 #endif
 452                         delete chimera;
 453
 454
 455                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 456
 457                         outputNames.push_back(outputFileName);
 458                         outputNames.push_back(accnosFileName);
 459
 460                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 461                 }
 462
 463                 m->mothurOutEndLine();
 464                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 465                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 466                 m->mothurOutEndLine();
 467
 468                 return 0;
 469
 470         }
 471         catch(exception& e) {
 472                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 473                 exit(1);
 474         }
 475 }
 476 //**********************************************************************************************************************
 477
 478 int ChimeraSlayerCommand::driver(linePair* line, string outputFName, string filename, string accnos){
 479         try {
 480                 ofstream out;
 481                 openOutputFile(outputFName, out);
 482
 483                 ofstream out2;
 484                 openOutputFile(accnos, out2);
 485
 486                 ifstream inFASTA;
 487                 openInputFile(filename, inFASTA);
 488
 489                 inFASTA.seekg(line->start);
 490
 491                 for(int i=0;i<line->numSeqs;i++){
 492
 493                         if (m->control_pressed) {       return 1;       }
 494
 495                         Sequence* candidateSeq = new Sequence(inFASTA);  gobble(inFASTA);
 496
 497                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 498
 499                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 500                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 501                                 }else{
 502                                         //find chimeras
 503                                         chimera->getChimeras(candidateSeq);
 504
 505                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 506
 507                                         //print results
 508                                         chimera->print(out, out2);
 509                                 }
 510                         }
 511                         delete candidateSeq;
 512
 513                         //report progress
 514                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1)); m->mothurOutEndLine();           }
 515                 }
 516                 //report progress
 517                 if((line->numSeqs) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(line->numSeqs)); m->mothurOutEndLine();         }
 518
 519                 out.close();
 520                 out2.close();
 521                 inFASTA.close();
 522
 523                 return 0;
 524         }
 525         catch(exception& e) {
 526                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 527                 exit(1);
 528         }
 529 }
 530 //**********************************************************************************************************************
 531 #ifdef USE_MPI
 532 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 533         try {
 534                 MPI_Status status;
 535                 int pid;
 536                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 537
 538                 for(int i=0;i<num;i++){
 539
 540                         if (m->control_pressed) {       return 1;       }
 541
 542                         //read next sequence
 543                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 544
 545                         char* buf4 = new char[length];
 546                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 547
 548                         string tempBuf = buf4;
 549                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 550                         istringstream iss (tempBuf,istringstream::in);
 551
 552                         delete buf4;
 553
 554                         Sequence* candidateSeq = new Sequence(iss);  gobble(iss);
 555
 556                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 557
 558                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 559                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 560                                 }else{
 561
 562                                         //find chimeras
 563                                         chimera->getChimeras(candidateSeq);
 564
 565                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 566                 //cout << "about to print" << endl;
 567                                         //print results
 568                                         bool isChimeric = chimera->print(outMPI, outAccMPI);
 569                                 }
 570                         }
 571                         delete candidateSeq;
 572
 573                         //report progress
 574                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 575                 }
 576                 //report progress
 577                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 578
 579
 580                 return 0;
 581         }
 582         catch(exception& e) {
 583                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 584                 exit(1);
 585         }
 586 }
 587 #endif
 588
 589 /**************************************************************************************************/
 590
 591 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos) {
 592         try {
 593 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 594                 int process = 0;
 595                 //              processIDS.resize(0);
 596
 597                 //loop through and create all the processes you want
 598                 while (process != processors) {
 599                         int pid = fork();
 600
 601                         if (pid > 0) {
 602                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 603                                 process++;
 604                         }else if (pid == 0){
 605                                 driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 606                                 exit(0);
 607                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 608                 }
 609
 610                 //force parent to wait until all the processes are done
 611                 for (int i=0;i<processors;i++) {
 612                         int temp = processIDS[i];
 613                         wait(&temp);
 614                 }
 615
 616                 return 0;
 617 #endif
 618         }
 619         catch(exception& e) {
 620                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 621                 exit(1);
 622         }
 623 }
 624
 625 /**************************************************************************************************/
 626
 627