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 //**********************************************************************************************************************
  13 vector<string> ChimeraCcodeCommand::setParameters(){
  14         try {
  15                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
  18                 CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
  19                 CommandParameter pnumwanted("numwanted", "Number", "", "20", "", "", "",false,false); parameters.push_back(pnumwanted);
  20                 CommandParameter pmask("mask", "String", "", "", "", "", "",false,false); parameters.push_back(pmask);
  21                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ChimeraCcodeCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraCcodeCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.ccode command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  39                 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";
  40                 helpString += "The chimera.ccode command parameters are fasta, reference, filter, mask, processors, window and numwanted.\n";
  41                 helpString += "The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required unless you have a valid current fasta file. \n";
  42                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  43                 helpString += "The reference parameter allows you to enter a reference file containing known non-chimeric sequences, and is required. \n";
  44                 helpString += "The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n";
  45                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  46 #ifdef USE_MPI
  47                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  48 #endif
  49                 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";
  50                 helpString += "The window parameter allows you to specify the window size for searching for chimeras. \n";
  51                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n";
  52                 helpString += "The chimera.ccode command should be in the following format: \n";
  53                 helpString += "chimera.ccode(fasta=yourFastaFile, reference=yourTemplate) \n";
  54                 helpString += "Example: chimera.ccode(fasta=AD.align, reference=core_set_aligned.imputed.fasta) \n";
  55                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  56                 return helpString;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "ChimeraCcodeCommand", "getHelpString");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 ChimeraCcodeCommand::ChimeraCcodeCommand(){
  65         try {
  66                 abort = true; calledHelp = true;
  67                 setParameters();
  68                 vector<string> tempOutNames;
  69                 outputTypes["chimera"] = tempOutNames;
  70                 outputTypes["mapinfo"] = tempOutNames;
  71                 outputTypes["accnos"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
  75                 exit(1);
  76         }
  77 }
  78 //***************************************************************************************************************
  79 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82
  83                 //allow user to run help
  84                 if(option == "help") { help(); abort = true; calledHelp = true; }
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter("chimera.ccode");
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         vector<string> tempOutNames;
 101                         outputTypes["chimera"] = tempOutNames;
 102                         outputTypes["mapinfo"] = tempOutNames;
 103                         outputTypes["accnos"] = tempOutNames;
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("reference");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 116                                 }
 117                         }
 118
 119                         //check for required parameters
 120                         fastafile = validParameter.validFile(parameters, "fasta", false);
 121                         if (fastafile == "not found") {                                 //if there is a current fasta file, use it
 122                                 string filename = m->getFastaFile();
 123                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 124                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 125                         }else {
 126                                 m->splitAtDash(fastafile, fastaFileNames);
 127
 128                                 //go through files and make sure they are good, if not, then disregard them
 129                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 130
 131                                         bool ignore = false;
 132                                         if (fastaFileNames[i] == "current") {
 133                                                 fastaFileNames[i] = m->getFastaFile();
 134                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 135                                                 else {
 136                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 137                                                         //erase from file list
 138                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 139                                                         i--;
 140                                                 }
 141                                         }
 142
 143                                         if (!ignore) {
 144
 145                                                 if (inputDir != "") {
 146                                                         string path = m->hasPath(fastaFileNames[i]);
 147                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 148                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 149                                                 }
 150
 151                                                 int ableToOpen;
 152                                                 ifstream in;
 153
 154                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 155
 156                                                 //if you can't open it, try default location
 157                                                 if (ableToOpen == 1) {
 158                                                         if (m->getDefaultPath() != "") { //default path is set
 159                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 160                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 161                                                                 ifstream in2;
 162                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 163                                                                 in2.close();
 164                                                                 fastaFileNames[i] = tryPath;
 165                                                         }
 166                                                 }
 167
 168                                                 //if you can't open it, try default location
 169                                                 if (ableToOpen == 1) {
 170                                                         if (m->getOutputDir() != "") { //default path is set
 171                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 172                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 173                                                                 ifstream in2;
 174                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 175                                                                 in2.close();
 176                                                                 fastaFileNames[i] = tryPath;
 177                                                         }
 178                                                 }
 179
 180                                                 in.close();
 181
 182                                                 if (ableToOpen == 1) {
 183                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 184                                                         //erase from file list
 185                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 186                                                         i--;
 187                                                 }
 188                                         }
 189                                 }
 190
 191                                 //make sure there is at least one valid file left
 192                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 193                         }
 194
 195                         //if the user changes the output directory command factory will send this info to us in the output parameter
 196                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 197
 198                         templatefile = validParameter.validFile(parameters, "reference", true);
 199                         if (templatefile == "not open") { abort = true; }
 200                         else if (templatefile == "not found") { templatefile = ""; m->mothurOut("reference is a required parameter for the chimera.ccode command."); m->mothurOutEndLine(); abort = true;  }
 201
 202                         maskfile = validParameter.validFile(parameters, "mask", false);
 203                         if (maskfile == "not found") { maskfile = "";  }
 204                         else if (maskfile != "default")  {
 205                                 if (inputDir != "") {
 206                                         string path = m->hasPath(maskfile);
 207                                         //if the user has not given a path then, add inputdir. else leave path alone.
 208                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 209                                 }
 210
 211                                 ifstream in;
 212                                 int     ableToOpen = m->openInputFile(maskfile, in);
 213                                 if (ableToOpen == 1) { abort = true; }
 214                                 in.close();
 215                         }
 216
 217                         string temp;
 218                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 219                         filter = m->isTrue(temp);
 220
 221                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 222                         m->setProcessors(temp);
 223                         convert(temp, processors);
 224
 225                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 226                         convert(temp, window);
 227
 228                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 229                         convert(temp, numwanted);
 230
 231                 }
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 235                 exit(1);
 236         }
 237 }
 238 //***************************************************************************************************************
 239 int ChimeraCcodeCommand::execute(){
 240         try{
 241
 242                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 243
 244                 for (int s = 0; s < fastaFileNames.size(); s++) {
 245
 246                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 247
 248                         int start = time(NULL);
 249
 250                         //set user options
 251                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 252
 253                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 254
 255                         //is your template aligned?
 256                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 257                         templateSeqsLength = chimera->getLength();
 258
 259                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 260                         string outputFileName, accnosFileName;
 261                         if (maskfile != "") {
 262                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.chimeras";
 263                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + ".ccode.accnos";
 264                         }else {
 265                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.chimeras";
 266                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "ccode.accnos";
 267                         }
 268
 269                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "mapinfo";
 270
 271                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } outputTypes.clear(); return 0;        }
 272
 273                 #ifdef USE_MPI
 274
 275                                 int pid, numSeqsPerProcessor;
 276                                 int tag = 2001;
 277                                 vector<unsigned long int> MPIPos;
 278
 279                                 MPI_Status status;
 280                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 281                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 282
 283                                 MPI_File inMPI;
 284                                 MPI_File outMPI;
 285                                 MPI_File outMPIAccnos;
 286
 287                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 288                                 int inMode=MPI_MODE_RDONLY;
 289
 290                                 char outFilename[1024];
 291                                 strcpy(outFilename, outputFileName.c_str());
 292
 293                                 char outAccnosFilename[1024];
 294                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 295
 296                                 char inFileName[1024];
 297                                 strcpy(inFileName, fastaFileNames[s].c_str());
 298
 299                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 300                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 301                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 302
 303                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); } outputTypes.clear();  delete chimera; return 0;  }
 304
 305                                 if (pid == 0) { //you are the root process
 306                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n";
 307
 308                                         //print header
 309                                         int length = outTemp.length();
 310                                         char* buf2 = new char[length];
 311                                         memcpy(buf2, outTemp.c_str(), length);
 312
 313                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 314                                         delete buf2;
 315
 316                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 317
 318                                         //send file positions to all processes
 319                                         for(int i = 1; i < processors; i++) {
 320                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 321                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 322                                         }
 323
 324                                         //figure out how many sequences you have to align
 325                                         numSeqsPerProcessor = numSeqs / processors;
 326                                         int startIndex =  pid * numSeqsPerProcessor;
 327                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 328
 329
 330                                         //align your part
 331                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 332
 333                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } outputTypes.clear();  delete chimera; return 0;  }
 334
 335                                 }else{ //you are a child process
 336                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 337                                         MPIPos.resize(numSeqs+1);
 338                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 339
 340                                         //figure out how many sequences you have to align
 341                                         numSeqsPerProcessor = numSeqs / processors;
 342                                         int startIndex =  pid * numSeqsPerProcessor;
 343                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 344
 345
 346                                         //align your part
 347                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 348
 349                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  outputTypes.clear(); delete chimera; return 0;  }
 350                                 }
 351
 352                                 //close files
 353                                 MPI_File_close(&inMPI);
 354                                 MPI_File_close(&outMPI);
 355                                 MPI_File_close(&outMPIAccnos);
 356
 357                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 358
 359                 #else
 360                         ofstream outHeader;
 361                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 362                         m->openOutputFile(tempHeader, outHeader);
 363
 364                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 365
 366                         outHeader.close();
 367
 368                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 369
 370                         for (int i = 0; i < (positions.size()-1); i++) {
 371                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 372                         }
 373
 374                         //break up file
 375                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 376                                 if(processors == 1){
 377
 378                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 379
 380                                         if (m->control_pressed) { 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];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 381
 382                                 }else{
 383                                         processIDS.resize(0);
 384
 385                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 386
 387                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 388                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 389
 390                                         //append output files
 391                                         for(int i=1;i<processors;i++){
 392                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 393                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 394                                         }
 395
 396                                         //append output files
 397                                         for(int i=1;i<processors;i++){
 398                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 399                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 400                                         }
 401
 402                                         if (m->control_pressed) {
 403                                                 remove(outputFileName.c_str());
 404                                                 remove(accnosFileName.c_str());
 405                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } outputTypes.clear();
 406                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 407                                                 delete chimera;
 408                                                 return 0;
 409                                         }
 410
 411                                 }
 412
 413                         #else
 414                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 415
 416                                 if (m->control_pressed) { 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];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 417
 418                         #endif
 419
 420                         m->appendFiles(outputFileName, tempHeader);
 421
 422                         remove(outputFileName.c_str());
 423                         rename(tempHeader.c_str(), outputFileName.c_str());
 424                 #endif
 425
 426                         delete chimera;
 427
 428                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 429                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 430                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 431
 432                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 433
 434                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 435                 }
 436
 437
 438                 //set accnos file as new current accnosfile
 439                 string current = "";
 440                 itTypes = outputTypes.find("accnos");
 441                 if (itTypes != outputTypes.end()) {
 442                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 443                 }
 444
 445                 m->mothurOutEndLine();
 446                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 447                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 448                 m->mothurOutEndLine();
 449
 450                 return 0;
 451
 452         }
 453         catch(exception& e) {
 454                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 455                 exit(1);
 456         }
 457 }
 458 //**********************************************************************************************************************
 459
 460 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 461         try {
 462                 ofstream out;
 463                 m->openOutputFile(outputFName, out);
 464
 465                 ofstream out2;
 466                 m->openOutputFile(accnos, out2);
 467
 468                 ifstream inFASTA;
 469                 m->openInputFile(filename, inFASTA);
 470
 471                 inFASTA.seekg(filePos->start);
 472
 473                 bool done = false;
 474                 int count = 0;
 475
 476                 while (!done) {
 477
 478                         if (m->control_pressed) {       return 1;       }
 479
 480                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 481
 482                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 483
 484                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 485                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 486                                 }else{
 487                                         //find chimeras
 488                                         chimera->getChimeras(candidateSeq);
 489
 490                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 491
 492                                         //print results
 493                                         chimera->print(out, out2);
 494                                 }
 495                                 count++;
 496                         }
 497                         delete candidateSeq;
 498
 499                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 500                                 unsigned long int pos = inFASTA.tellg();
 501                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 502                         #else
 503                                 if (inFASTA.eof()) { break; }
 504                         #endif
 505
 506                         //report progress
 507                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 508                 }
 509                 //report progress
 510                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 511
 512                 out.close();
 513                 out2.close();
 514                 inFASTA.close();
 515
 516                 return count;
 517         }
 518         catch(exception& e) {
 519                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 520                 exit(1);
 521         }
 522 }
 523 //**********************************************************************************************************************
 524 #ifdef USE_MPI
 525 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
 526         try {
 527
 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 0; }
 535
 536                         //read next sequence
 537                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 538
 539                         char* buf4 = new char[length];
 540
 541                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 542
 543                         string tempBuf = buf4;
 544                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 545                         istringstream iss (tempBuf,istringstream::in);
 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                                         //find chimeras
 556                                         chimera->getChimeras(candidateSeq);
 557
 558                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 559
 560                                         //print results
 561                                         chimera->print(outMPI, outAccMPI);
 562                                 }
 563                         }
 564                         delete candidateSeq;
 565
 566                         //report progress
 567                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 568                 }
 569                 //report progress
 570                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 571
 572
 573                 return 0;
 574         }
 575         catch(exception& e) {
 576                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 577                 exit(1);
 578         }
 579 }
 580 #endif
 581
 582 /**************************************************************************************************/
 583
 584 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 585         try {
 586 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 587                 int process = 0;
 588                 int num = 0;
 589
 590                 //loop through and create all the processes you want
 591                 while (process != processors) {
 592                         int pid = fork();
 593
 594                         if (pid > 0) {
 595                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 596                                 process++;
 597                         }else if (pid == 0){
 598                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 599
 600                                 //pass numSeqs to parent
 601                                 ofstream out;
 602                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 603                                 m->openOutputFile(tempFile, out);
 604                                 out << num << endl;
 605                                 out.close();
 606
 607                                 exit(0);
 608                         }else {
 609                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 610                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 611                                 exit(0);
 612                         }
 613                 }
 614
 615                 //force parent to wait until all the processes are done
 616                 for (int i=0;i<processors;i++) {
 617                         int temp = processIDS[i];
 618                         wait(&temp);
 619                 }
 620
 621                 for (int i = 0; i < processIDS.size(); i++) {
 622                         ifstream in;
 623                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 624                         m->openInputFile(tempFile, in);
 625                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 626                         in.close(); remove(tempFile.c_str());
 627                 }
 628
 629                 return num;
 630 #endif
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 634                 exit(1);
 635         }
 636 }
 637 //**********************************************************************************************************************
 638