chimeraccodecommand.cpp

   1 /*
   2  *  chimeraccodecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/30/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraccodecommand.h"
  11 #include "ccode.h"
  12 #include "referencedb.h"
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraCcodeCommand::setParameters(){
  15         try {
  16                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
  19                 CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
  20                 CommandParameter pnumwanted("numwanted", "Number", "", "20", "", "", "",false,false); parameters.push_back(pnumwanted);
  21                 CommandParameter pmask("mask", "String", "", "", "", "", "",false,false); parameters.push_back(pmask);
  22                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  26
  27                 vector<string> myArray;
  28                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  29                 return myArray;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "ChimeraCcodeCommand", "setParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 string ChimeraCcodeCommand::getHelpString(){
  38         try {
  39                 string helpString = "";
  40                 helpString += "The chimera.ccode command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  41                 helpString += "This command was created using the algorythms described in the 'Evaluating putative chimeric sequences from PCR-amplified products' paper by Juan M. Gonzalez, Johannes Zimmerman and Cesareo Saiz-Jimenez.\n";
  42                 helpString += "The chimera.ccode command parameters are fasta, reference, filter, mask, processors, window and numwanted.\n";
  43                 helpString += "The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required unless you have a valid current fasta file. \n";
  44                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  45                 helpString += "The reference parameter allows you to enter a reference file containing known non-chimeric sequences, and is required. \n";
  46                 helpString += "The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n";
  47                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  48 #ifdef USE_MPI
  49                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  50 #endif
  51                 helpString += "The mask parameter allows you to specify a file containing one sequence you wish to use as a mask for the your sequences. \n";
  52                 helpString += "The window parameter allows you to specify the window size for searching for chimeras. \n";
  53                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n";
  54                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  55                 helpString += "The chimera.ccode command should be in the following format: \n";
  56                 helpString += "chimera.ccode(fasta=yourFastaFile, reference=yourTemplate) \n";
  57                 helpString += "Example: chimera.ccode(fasta=AD.align, reference=core_set_aligned.imputed.fasta) \n";
  58                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  59                 return helpString;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "ChimeraCcodeCommand", "getHelpString");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 ChimeraCcodeCommand::ChimeraCcodeCommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["chimera"] = tempOutNames;
  73                 outputTypes["mapinfo"] = tempOutNames;
  74                 outputTypes["accnos"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
  78                 exit(1);
  79         }
  80 }
  81 //***************************************************************************************************************
  82 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
  83         try {
  84                 abort = false; calledHelp = false;
  85                 ReferenceDB* rdb = ReferenceDB::getInstance();
  86
  87                 //allow user to run help
  88                 if(option == "help") { help(); abort = true; calledHelp = true; }
  89                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  90
  91                 else {
  92                         vector<string> myArray = setParameters();
  93
  94                         OptionParser parser(option);
  95                         map<string,string> parameters = parser.getParameters();
  96
  97                         ValidParameters validParameter("chimera.ccode");
  98                         map<string,string>::iterator it;
  99
 100                         //check to make sure all parameters are valid for command
 101                         for (it = parameters.begin(); it != parameters.end(); it++) {
 102                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 103                         }
 104
 105                         vector<string> tempOutNames;
 106                         outputTypes["chimera"] = tempOutNames;
 107                         outputTypes["mapinfo"] = tempOutNames;
 108                         outputTypes["accnos"] = tempOutNames;
 109
 110                         //if the user changes the input directory command factory will send this info to us in the output parameter
 111                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 112                         if (inputDir == "not found"){   inputDir = "";          }
 113                         else {
 114                                 string path;
 115                                 it = parameters.find("reference");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 121                                 }
 122                         }
 123
 124                         //check for required parameters
 125                         fastafile = validParameter.validFile(parameters, "fasta", false);
 126                         if (fastafile == "not found") {                                 //if there is a current fasta file, use it
 127                                 string filename = m->getFastaFile();
 128                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 129                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 130                         }else {
 131                                 m->splitAtDash(fastafile, fastaFileNames);
 132
 133                                 //go through files and make sure they are good, if not, then disregard them
 134                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 135
 136                                         bool ignore = false;
 137                                         if (fastaFileNames[i] == "current") {
 138                                                 fastaFileNames[i] = m->getFastaFile();
 139                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 140                                                 else {
 141                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 142                                                         //erase from file list
 143                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 144                                                         i--;
 145                                                 }
 146                                         }
 147
 148                                         if (!ignore) {
 149
 150                                                 if (inputDir != "") {
 151                                                         string path = m->hasPath(fastaFileNames[i]);
 152                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 154                                                 }
 155
 156                                                 int ableToOpen;
 157                                                 ifstream in;
 158
 159                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 160
 161                                                 //if you can't open it, try default location
 162                                                 if (ableToOpen == 1) {
 163                                                         if (m->getDefaultPath() != "") { //default path is set
 164                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 165                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 166                                                                 ifstream in2;
 167                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 168                                                                 in2.close();
 169                                                                 fastaFileNames[i] = tryPath;
 170                                                         }
 171                                                 }
 172
 173                                                 //if you can't open it, try default location
 174                                                 if (ableToOpen == 1) {
 175                                                         if (m->getOutputDir() != "") { //default path is set
 176                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 177                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 178                                                                 ifstream in2;
 179                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 180                                                                 in2.close();
 181                                                                 fastaFileNames[i] = tryPath;
 182                                                         }
 183                                                 }
 184
 185                                                 in.close();
 186
 187                                                 if (ableToOpen == 1) {
 188                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 189                                                         //erase from file list
 190                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 191                                                         i--;
 192                                                 }else {
 193                                                         m->setFastaFile(fastaFileNames[i]);
 194                                                 }
 195                                         }
 196                                 }
 197
 198                                 //make sure there is at least one valid file left
 199                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 200                         }
 201
 202                         //if the user changes the output directory command factory will send this info to us in the output parameter
 203                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 204
 205                         maskfile = validParameter.validFile(parameters, "mask", false);
 206                         if (maskfile == "not found") { maskfile = "";  }
 207                         else if (maskfile != "default")  {
 208                                 if (inputDir != "") {
 209                                         string path = m->hasPath(maskfile);
 210                                         //if the user has not given a path then, add inputdir. else leave path alone.
 211                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 212                                 }
 213
 214                                 ifstream in;
 215                                 int     ableToOpen = m->openInputFile(maskfile, in);
 216                                 if (ableToOpen == 1) { abort = true; }
 217                                 in.close();
 218                         }
 219
 220                         string temp;
 221                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 222                         filter = m->isTrue(temp);
 223
 224                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 225                         m->setProcessors(temp);
 226                         convert(temp, processors);
 227
 228                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 229                         convert(temp, window);
 230
 231                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 232                         convert(temp, numwanted);
 233
 234                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 235                         save = m->isTrue(temp);
 236                         rdb->save = save;
 237                         if (save) { //clear out old references
 238                                 rdb->clearMemory();
 239                         }
 240
 241                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 242                         templatefile = validParameter.validFile(parameters, "reference", true);
 243                         if (templatefile == "not found") {
 244                                 //check for saved reference sequences
 245                                 if (rdb->referenceSeqs.size() != 0) {
 246                                         templatefile = "saved";
 247                                 }else {
 248                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 249                                         m->mothurOutEndLine();
 250                                         abort = true;
 251                                 }
 252                         }else if (templatefile == "not open") { abort = true; }
 253                         else {  if (save) {     rdb->setSavedReference(templatefile);   }       }
 254
 255
 256                 }
 257         }
 258         catch(exception& e) {
 259                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 260                 exit(1);
 261         }
 262 }
 263 //***************************************************************************************************************
 264 int ChimeraCcodeCommand::execute(){
 265         try{
 266
 267                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 268
 269                 for (int s = 0; s < fastaFileNames.size(); s++) {
 270
 271                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 272
 273                         int start = time(NULL);
 274
 275                         //set user options
 276                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 277
 278                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 279
 280                         //is your template aligned?
 281                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 282                         templateSeqsLength = chimera->getLength();
 283
 284                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 285                         string outputFileName, accnosFileName;
 286                         if (maskfile != "") {
 287                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 288                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 289                         }else {
 290                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 291                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 292                         }
 293
 294                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "mapinfo";
 295
 296                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } outputTypes.clear(); return 0;        }
 297
 298                 #ifdef USE_MPI
 299
 300                                 int pid, numSeqsPerProcessor;
 301                                 int tag = 2001;
 302                                 vector<unsigned long int> MPIPos;
 303
 304                                 MPI_Status status;
 305                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 306                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 307
 308                                 MPI_File inMPI;
 309                                 MPI_File outMPI;
 310                                 MPI_File outMPIAccnos;
 311
 312                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 313                                 int inMode=MPI_MODE_RDONLY;
 314
 315                                 char outFilename[1024];
 316                                 strcpy(outFilename, outputFileName.c_str());
 317
 318                                 char outAccnosFilename[1024];
 319                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 320
 321                                 char inFileName[1024];
 322                                 strcpy(inFileName, fastaFileNames[s].c_str());
 323
 324                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 325                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 326                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 327
 328                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        } outputTypes.clear();  delete chimera; return 0;  }
 329
 330                                 if (pid == 0) { //you are the root process
 331                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n";
 332
 333                                         //print header
 334                                         int length = outTemp.length();
 335                                         char* buf2 = new char[length];
 336                                         memcpy(buf2, outTemp.c_str(), length);
 337
 338                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 339                                         delete buf2;
 340
 341                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 342
 343                                         //send file positions to all processes
 344                                         for(int i = 1; i < processors; i++) {
 345                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 346                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 347                                         }
 348
 349                                         //figure out how many sequences you have to align
 350                                         numSeqsPerProcessor = numSeqs / processors;
 351                                         int startIndex =  pid * numSeqsPerProcessor;
 352                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 353
 354
 355                                         //align your part
 356                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 357
 358                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  m->mothurRemove(outputFileName);  m->mothurRemove(accnosFileName);  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } outputTypes.clear();  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
 371                                         //align your part
 372                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 373
 374                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  outputTypes.clear(); delete chimera; return 0;  }
 375                                 }
 376
 377                                 //close files
 378                                 MPI_File_close(&inMPI);
 379                                 MPI_File_close(&outMPI);
 380                                 MPI_File_close(&outMPIAccnos);
 381
 382                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 383
 384                 #else
 385                         ofstream outHeader;
 386                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 387                         m->openOutputFile(tempHeader, outHeader);
 388
 389                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 390
 391                         outHeader.close();
 392
 393                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 394
 395                         for (int i = 0; i < (positions.size()-1); i++) {
 396                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 397                         }
 398
 399                         //break up file
 400                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 401                                 if(processors == 1){
 402
 403                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 404
 405                                         if (m->control_pressed) { m->mothurRemove(outputFileName); m->mothurRemove(tempHeader); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 406
 407                                 }else{
 408                                         processIDS.resize(0);
 409
 410                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 411
 412                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 413                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 414
 415                                         //append output files
 416                                         for(int i=1;i<processors;i++){
 417                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 418                                                 m->mothurRemove((outputFileName + toString(processIDS[i]) + ".temp"));
 419                                         }
 420
 421                                         //append output files
 422                                         for(int i=1;i<processors;i++){
 423                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 424                                                 m->mothurRemove((accnosFileName + toString(processIDS[i]) + ".temp"));
 425                                         }
 426
 427                                         if (m->control_pressed) {
 428                                                 m->mothurRemove(outputFileName);
 429                                                 m->mothurRemove(accnosFileName);
 430                                                 for (int j = 0; j < outputNames.size(); j++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear();
 431                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 432                                                 delete chimera;
 433                                                 return 0;
 434                                         }
 435
 436                                 }
 437
 438                         #else
 439                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 440
 441                                 if (m->control_pressed) { m->mothurRemove(outputFileName); m->mothurRemove(tempHeader); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 442
 443                         #endif
 444
 445                         m->appendFiles(outputFileName, tempHeader);
 446
 447                         m->mothurRemove(outputFileName);
 448                         rename(tempHeader.c_str(), outputFileName.c_str());
 449                 #endif
 450
 451                         delete chimera;
 452
 453                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 454                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 455                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 456
 457                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 458
 459                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 460                 }
 461
 462
 463                 //set accnos file as new current accnosfile
 464                 string current = "";
 465                 itTypes = outputTypes.find("accnos");
 466                 if (itTypes != outputTypes.end()) {
 467                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 468                 }
 469
 470                 m->mothurOutEndLine();
 471                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 472                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 473                 m->mothurOutEndLine();
 474
 475                 return 0;
 476
 477         }
 478         catch(exception& e) {
 479                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 480                 exit(1);
 481         }
 482 }
 483 //**********************************************************************************************************************
 484
 485 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 486         try {
 487                 ofstream out;
 488                 m->openOutputFile(outputFName, out);
 489
 490                 ofstream out2;
 491                 m->openOutputFile(accnos, out2);
 492
 493                 ifstream inFASTA;
 494                 m->openInputFile(filename, inFASTA);
 495
 496                 inFASTA.seekg(filePos->start);
 497
 498                 bool done = false;
 499                 int count = 0;
 500
 501                 while (!done) {
 502
 503                         if (m->control_pressed) {       return 1;       }
 504
 505                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 506
 507                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 508
 509                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 510                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 511                                 }else{
 512                                         //find chimeras
 513                                         chimera->getChimeras(candidateSeq);
 514
 515                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 516
 517                                         //print results
 518                                         chimera->print(out, out2);
 519                                 }
 520                                 count++;
 521                         }
 522                         delete candidateSeq;
 523
 524                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 525                                 unsigned long int pos = inFASTA.tellg();
 526                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 527                         #else
 528                                 if (inFASTA.eof()) { break; }
 529                         #endif
 530
 531                         //report progress
 532                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 533                 }
 534                 //report progress
 535                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 536
 537                 out.close();
 538                 out2.close();
 539                 inFASTA.close();
 540
 541                 return count;
 542         }
 543         catch(exception& e) {
 544                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 545                 exit(1);
 546         }
 547 }
 548 //**********************************************************************************************************************
 549 #ifdef USE_MPI
 550 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 551         try {
 552
 553                 MPI_Status status;
 554                 int pid;
 555                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 556
 557                 for(int i=0;i<num;i++){
 558
 559                         if (m->control_pressed) { return 0; }
 560
 561                         //read next sequence
 562                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 563
 564                         char* buf4 = new char[length];
 565
 566                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 567
 568                         string tempBuf = buf4;
 569                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 570                         istringstream iss (tempBuf,istringstream::in);
 571                         delete buf4;
 572
 573                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 574
 575                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 576
 577                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 578                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 579                                 }else{
 580                                         //find chimeras
 581                                         chimera->getChimeras(candidateSeq);
 582
 583                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 584
 585                                         //print results
 586                                         chimera->print(outMPI, outAccMPI);
 587                                 }
 588                         }
 589                         delete candidateSeq;
 590
 591                         //report progress
 592                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 593                 }
 594                 //report progress
 595                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 596
 597
 598                 return 0;
 599         }
 600         catch(exception& e) {
 601                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 602                 exit(1);
 603         }
 604 }
 605 #endif
 606
 607 /**************************************************************************************************/
 608
 609 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 610         try {
 611 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 612                 int process = 0;
 613                 int num = 0;
 614
 615                 //loop through and create all the processes you want
 616                 while (process != processors) {
 617                         int pid = fork();
 618
 619                         if (pid > 0) {
 620                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 621                                 process++;
 622                         }else if (pid == 0){
 623                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 624
 625                                 //pass numSeqs to parent
 626                                 ofstream out;
 627                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 628                                 m->openOutputFile(tempFile, out);
 629                                 out << num << endl;
 630                                 out.close();
 631
 632                                 exit(0);
 633                         }else {
 634                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 635                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 636                                 exit(0);
 637                         }
 638                 }
 639
 640                 //force parent to wait until all the processes are done
 641                 for (int i=0;i<processors;i++) {
 642                         int temp = processIDS[i];
 643                         wait(&temp);
 644                 }
 645
 646                 for (int i = 0; i < processIDS.size(); i++) {
 647                         ifstream in;
 648                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 649                         m->openInputFile(tempFile, in);
 650                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 651                         in.close(); m->mothurRemove(tempFile);
 652                 }
 653
 654                 return num;
 655 #endif
 656         }
 657         catch(exception& e) {
 658                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 659                 exit(1);
 660         }
 661 }
 662 //**********************************************************************************************************************
 663