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