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 vector<string> ChimeraSlayerCommand::getValidParameters(){
  15         try {
  16                 string AlignArray[] =  {"fasta", "processors", "window", "template","numwanted", "ksize", "match","mismatch",
  17                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  18                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "ChimeraSlayerCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 ChimeraSlayerCommand::ChimeraSlayerCommand(){
  28         try {
  29                 vector<string> tempOutNames;
  30                 outputTypes["chimera"] = tempOutNames;
  31                 outputTypes["accnos"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> ChimeraSlayerCommand::getRequiredParameters(){
  40         try {
  41                 string AlignArray[] =  {"template","fasta"};
  42                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> ChimeraSlayerCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 //***************************************************************************************************************
  62 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
  63         try {
  64                 abort = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta", "processors", "window", "template","numwanted", "ksize", "match","mismatch",
  72                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  73                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  74
  75                         OptionParser parser(option);
  76                         map<string,string> parameters = parser.getParameters();
  77
  78                         ValidParameters validParameter("chimera.slayer");
  79                         map<string,string>::iterator it;
  80
  81                         //check to make sure all parameters are valid for command
  82                         for (it = parameters.begin(); it != parameters.end(); it++) {
  83                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  84                         }
  85
  86                         vector<string> tempOutNames;
  87                         outputTypes["chimera"] = tempOutNames;
  88                         outputTypes["accnos"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("template");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
 101                                 }
 102                         }
 103
 104
 105                         //check for required parameters
 106                         fastafile = validParameter.validFile(parameters, "fasta", false);
 107                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 108                         else {
 109                                 m->splitAtDash(fastafile, fastaFileNames);
 110
 111                                 //go through files and make sure they are good, if not, then disregard them
 112                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 113                                         if (inputDir != "") {
 114                                                 string path = m->hasPath(fastaFileNames[i]);
 115                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 116                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 117                                         }
 118
 119                                         int ableToOpen;
 120                                         ifstream in;
 121
 122                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 123
 124                                         //if you can't open it, try default location
 125                                         if (ableToOpen == 1) {
 126                                                 if (m->getDefaultPath() != "") { //default path is set
 127                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 128                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 129                                                         ifstream in2;
 130                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 131                                                         in2.close();
 132                                                         fastaFileNames[i] = tryPath;
 133                                                 }
 134                                         }
 135
 136                                         if (ableToOpen == 1) {
 137                                                 if (m->getOutputDir() != "") { //default path is set
 138                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 139                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 140                                                         ifstream in2;
 141                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 142                                                         in2.close();
 143                                                         fastaFileNames[i] = tryPath;
 144                                                 }
 145                                         }
 146
 147                                         in.close();
 148
 149                                         if (ableToOpen == 1) {
 150                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 151                                                 //erase from file list
 152                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 153                                                 i--;
 154                                         }
 155                                 }
 156
 157                                 //make sure there is at least one valid file left
 158                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 159                         }
 160
 161                         //if the user changes the output directory command factory will send this info to us in the output parameter
 162                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 163
 164                         templatefile = validParameter.validFile(parameters, "template", true);
 165                         if (templatefile == "not open") { abort = true; }
 166                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 167
 168                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 169                         convert(temp, processors);
 170
 171                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 172                         convert(temp, ksize);
 173
 174                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 175                         convert(temp, window);
 176
 177                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 178                         convert(temp, match);
 179
 180                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 181                         convert(temp, mismatch);
 182
 183                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 184                         convert(temp, divR);
 185
 186                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 187                         convert(temp, minSimilarity);
 188
 189                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 190                         convert(temp, minCoverage);
 191
 192                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 193                         convert(temp, minBS);
 194
 195                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "100"; }
 196                         convert(temp, minSNP);
 197
 198                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 199                         convert(temp, parents);
 200
 201                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 202                         realign = m->isTrue(temp);
 203
 204                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 205
 206                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "100"; }
 207                         convert(temp, iters);
 208
 209                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 210                         convert(temp, increment);
 211
 212                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 213                         convert(temp, numwanted);
 214
 215                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 216                 }
 217         }
 218         catch(exception& e) {
 219                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 220                 exit(1);
 221         }
 222 }
 223 //**********************************************************************************************************************
 224
 225 void ChimeraSlayerCommand::help(){
 226         try {
 227
 228                 m->mothurOut("The chimera.slayer command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 229                 m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
 230                 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,
 231                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 232                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n");
 233                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 234                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 235                 #ifdef USE_MPI
 236                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 237                 #endif
 238                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=50. \n");
 239                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n");
 240                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n");
 241                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n");
 242                 m->mothurOut("The match parameter allows you to reward matched bases in blast search, default is 5. \n");
 243                 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");
 244                 m->mothurOut("The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n");
 245                 m->mothurOut("The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n");
 246                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=100.\n");
 247                 m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
 248                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 249                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 250                 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");
 251                 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");
 252                 m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false.  \n");
 253                 m->mothurOut("The chimera.slayer command should be in the following format: \n");
 254                 m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
 255                 m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
 256                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 257         }
 258         catch(exception& e) {
 259                 m->errorOut(e, "ChimeraSlayerCommand", "help");
 260                 exit(1);
 261         }
 262 }
 263
 264 //***************************************************************************************************************
 265
 266 ChimeraSlayerCommand::~ChimeraSlayerCommand(){  /*      do nothing      */      }
 267
 268 //***************************************************************************************************************
 269
 270 int ChimeraSlayerCommand::execute(){
 271         try{
 272
 273                 if (abort == true) { return 0; }
 274
 275                 for (int s = 0; s < fastaFileNames.size(); s++) {
 276
 277                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 278
 279                         int start = time(NULL);
 280
 281                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 282
 283                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 284                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras";
 285                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 286
 287                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0;    }
 288
 289                         if (chimera->getUnaligned()) {
 290                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 291                                 delete chimera;
 292                                 return 0;
 293                         }
 294                         templateSeqsLength = chimera->getLength();
 295
 296                 #ifdef USE_MPI
 297                         int pid, end, numSeqsPerProcessor;
 298                                 int tag = 2001;
 299                                 vector<unsigned long int> MPIPos;
 300
 301                                 MPI_Status status;
 302                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 303                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 304
 305                                 MPI_File inMPI;
 306                                 MPI_File outMPI;
 307                                 MPI_File outMPIAccnos;
 308
 309                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 310                                 int inMode=MPI_MODE_RDONLY;
 311
 312                                 char outFilename[1024];
 313                                 strcpy(outFilename, outputFileName.c_str());
 314
 315                                 char outAccnosFilename[1024];
 316                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 317
 318                                 char inFileName[1024];
 319                                 strcpy(inFileName, fastaFileNames[s].c_str());
 320
 321                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 322                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 323                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 324
 325                                 if (m->control_pressed) { outputTypes.clear();  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;  }
 326
 327                                 if (pid == 0) { //you are the root process
 328                                         m->mothurOutEndLine();
 329                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 330                                         m->mothurOutEndLine();
 331
 332                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 333
 334                                         //print header
 335                                         int length = outTemp.length();
 336                                         char* buf2 = new char[length];
 337                                         memcpy(buf2, outTemp.c_str(), length);
 338
 339                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 340                                         delete buf2;
 341
 342                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 343
 344                                         //send file positions to all processes
 345                                         for(int i = 1; i < processors; i++) {
 346                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 347                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 348                                         }
 349
 350                                         //figure out how many sequences you have to align
 351                                         numSeqsPerProcessor = numSeqs / processors;
 352                                         int startIndex =  pid * numSeqsPerProcessor;
 353                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 354
 355                                         //do your part
 356                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 357
 358                                         if (m->control_pressed) { outputTypes.clear();  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;  }
 359
 360                                 }else{ //you are a child process
 361                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 362                                         MPIPos.resize(numSeqs+1);
 363                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 364
 365                                         //figure out how many sequences you have to align
 366                                         numSeqsPerProcessor = numSeqs / processors;
 367                                         int startIndex =  pid * numSeqsPerProcessor;
 368                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 369
 370                                         //do your part
 371                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 372
 373                                         if (m->control_pressed) { outputTypes.clear();  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;  }
 374                                 }
 375
 376                                 //close files
 377                                 MPI_File_close(&inMPI);
 378                                 MPI_File_close(&outMPI);
 379                                 MPI_File_close(&outMPIAccnos);
 380                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 381
 382                 #else
 383                         ofstream outHeader;
 384                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
 385                         m->openOutputFile(tempHeader, outHeader);
 386
 387                         chimera->printHeader(outHeader);
 388                         outHeader.close();
 389
 390                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 391
 392                         for (int i = 0; i < (positions.size()-1); i++) {
 393                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 394                         }
 395
 396                         //break up file
 397                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 398                                 if(processors == 1){
 399                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 400
 401                                         if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 402
 403                                 }else{
 404                                         processIDS.resize(0);
 405
 406                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 407
 408                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 409                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 410
 411                                         //append output files
 412                                         for(int i=1;i<processors;i++){
 413                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 414                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 415                                         }
 416
 417                                         //append output files
 418                                         for(int i=1;i<processors;i++){
 419                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 420                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 421                                         }
 422
 423                                         if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 424                                 }
 425
 426                         #else
 427                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 428
 429                                 if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 430
 431                         #endif
 432
 433                         m->appendFiles(outputFileName, tempHeader);
 434
 435                         remove(outputFileName.c_str());
 436                         rename(tempHeader.c_str(), outputFileName.c_str());
 437
 438                 #endif
 439                         delete chimera;
 440
 441
 442                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 443
 444                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 445                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 446
 447                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 448                 }
 449
 450                 m->mothurOutEndLine();
 451                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 452                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 453                 m->mothurOutEndLine();
 454
 455                 return 0;
 456
 457         }
 458         catch(exception& e) {
 459                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 460                 exit(1);
 461         }
 462 }
 463 //**********************************************************************************************************************
 464
 465 int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 466         try {
 467                 ofstream out;
 468                 m->openOutputFile(outputFName, out);
 469
 470                 ofstream out2;
 471                 m->openOutputFile(accnos, out2);
 472
 473                 ifstream inFASTA;
 474                 m->openInputFile(filename, inFASTA);
 475
 476                 inFASTA.seekg(filePos->start);
 477
 478                 bool done = false;
 479                 int count = 0;
 480
 481                 while (!done) {
 482
 483                         if (m->control_pressed) {       return 1;       }
 484
 485                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 486
 487                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 488
 489                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 490                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 491                                 }else{
 492                                         //find chimeras
 493                                         chimera->getChimeras(candidateSeq);
 494
 495                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 496
 497                                         //print results
 498                                         chimera->print(out, out2);
 499                                 }
 500                         count++;
 501                         }
 502                         delete candidateSeq;
 503
 504                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 505                                 unsigned long int pos = inFASTA.tellg();
 506                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 507                         #else
 508                                 if (inFASTA.eof()) { break; }
 509                         #endif
 510
 511                         //report progress
 512                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 513                 }
 514                 //report progress
 515                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 516
 517                 out.close();
 518                 out2.close();
 519                 inFASTA.close();
 520
 521                 return count;
 522         }
 523         catch(exception& e) {
 524                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 525                 exit(1);
 526         }
 527 }
 528 //**********************************************************************************************************************
 529 #ifdef USE_MPI
 530 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 531         try {
 532                 MPI_Status status;
 533                 int pid;
 534                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 535
 536                 for(int i=0;i<num;i++){
 537
 538                         if (m->control_pressed) {       return 1;       }
 539
 540                         //read next sequence
 541                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 542
 543                         char* buf4 = new char[length];
 544                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 545
 546                         string tempBuf = buf4;
 547                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 548                         istringstream iss (tempBuf,istringstream::in);
 549
 550                         delete buf4;
 551
 552                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 553
 554                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 555
 556                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 557                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 558                                 }else{
 559
 560                                         //find chimeras
 561                                         chimera->getChimeras(candidateSeq);
 562
 563                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 564                 //cout << "about to print" << endl;
 565                                         //print results
 566                                         bool isChimeric = chimera->print(outMPI, outAccMPI);
 567                                 }
 568                         }
 569                         delete candidateSeq;
 570
 571                         //report progress
 572                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 573                 }
 574                 //report progress
 575                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 576
 577
 578                 return 0;
 579         }
 580         catch(exception& e) {
 581                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 582                 exit(1);
 583         }
 584 }
 585 #endif
 586
 587 /**************************************************************************************************/
 588
 589 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos) {
 590         try {
 591 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 592                 int process = 0;
 593                 int num = 0;
 594
 595                 //loop through and create all the processes you want
 596                 while (process != processors) {
 597                         int pid = fork();
 598
 599                         if (pid > 0) {
 600                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 601                                 process++;
 602                         }else if (pid == 0){
 603                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 604
 605                                 //pass numSeqs to parent
 606                                 ofstream out;
 607                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 608                                 m->openOutputFile(tempFile, out);
 609                                 out << num << endl;
 610                                 out.close();
 611
 612                                 exit(0);
 613                         }else {
 614                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 615                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 616                                 exit(0);
 617                         }
 618                 }
 619
 620                 //force parent to wait until all the processes are done
 621                 for (int i=0;i<processors;i++) {
 622                         int temp = processIDS[i];
 623                         wait(&temp);
 624                 }
 625
 626                 for (int i = 0; i < processIDS.size(); i++) {
 627                         ifstream in;
 628                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 629                         m->openInputFile(tempFile, in);
 630                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 631                         in.close(); remove(tempFile.c_str());
 632                 }
 633
 634                 return num;
 635 #endif
 636         }
 637         catch(exception& e) {
 638                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 639                 exit(1);
 640         }
 641 }
 642
 643 /**************************************************************************************************/
 644
 645