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 string ChimeraCcodeCommand::getOutputFileNameTag(string type, string inputName=""){
  68         try {
  69         string outputFileName = "";
  70                 map<string, vector<string> >::iterator it;
  71
  72         //is this a type this command creates
  73         it = outputTypes.find(type);
  74         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  75         else {
  76             if (type == "chimera") {  outputFileName =  "ccode.chimeras"; }
  77             else if (type == "mapinfo") {  outputFileName =  "mapinfo"; }
  78             else if (type == "accnos") {  outputFileName =  "ccode.accnos"; }
  79             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  80         }
  81         return outputFileName;
  82         }
  83         catch(exception& e) {
  84                 m->errorOut(e, "ChimeraCcodeCommand", "getOutputFileNameTag");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 ChimeraCcodeCommand::ChimeraCcodeCommand(){
  90         try {
  91                 abort = true; calledHelp = true;
  92                 setParameters();
  93                 vector<string> tempOutNames;
  94                 outputTypes["chimera"] = tempOutNames;
  95                 outputTypes["mapinfo"] = tempOutNames;
  96                 outputTypes["accnos"] = tempOutNames;
  97         }
  98         catch(exception& e) {
  99                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 100                 exit(1);
 101         }
 102 }
 103 //***************************************************************************************************************
 104 ChimeraCcodeCommand::ChimeraCcodeCommand(string option)  {
 105         try {
 106                 abort = false; calledHelp = false;
 107                 ReferenceDB* rdb = ReferenceDB::getInstance();
 108
 109                 //allow user to run help
 110                 if(option == "help") { help(); abort = true; calledHelp = true; }
 111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 112
 113                 else {
 114                         vector<string> myArray = setParameters();
 115
 116                         OptionParser parser(option);
 117                         map<string,string> parameters = parser.getParameters();
 118
 119                         ValidParameters validParameter("chimera.ccode");
 120                         map<string,string>::iterator it;
 121
 122                         //check to make sure all parameters are valid for command
 123                         for (it = parameters.begin(); it != parameters.end(); it++) {
 124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 125                         }
 126
 127                         vector<string> tempOutNames;
 128                         outputTypes["chimera"] = tempOutNames;
 129                         outputTypes["mapinfo"] = tempOutNames;
 130                         outputTypes["accnos"] = tempOutNames;
 131
 132                         //if the user changes the input directory command factory will send this info to us in the output parameter
 133                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 134                         if (inputDir == "not found"){   inputDir = "";          }
 135                         else {
 136                                 string path;
 137                                 it = parameters.find("reference");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 143                                 }
 144                         }
 145
 146                         //check for required parameters
 147                         fastafile = validParameter.validFile(parameters, "fasta", false);
 148                         if (fastafile == "not found") {                                 //if there is a current fasta file, use it
 149                                 string filename = m->getFastaFile();
 150                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 151                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 152                         }else {
 153                                 m->splitAtDash(fastafile, fastaFileNames);
 154
 155                                 //go through files and make sure they are good, if not, then disregard them
 156                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 157
 158                                         bool ignore = false;
 159                                         if (fastaFileNames[i] == "current") {
 160                                                 fastaFileNames[i] = m->getFastaFile();
 161                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 162                                                 else {
 163                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 164                                                         //erase from file list
 165                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 166                                                         i--;
 167                                                 }
 168                                         }
 169
 170                                         if (!ignore) {
 171
 172                                                 if (inputDir != "") {
 173                                                         string path = m->hasPath(fastaFileNames[i]);
 174                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 176                                                 }
 177
 178                                                 int ableToOpen;
 179                                                 ifstream in;
 180
 181                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 182
 183                                                 //if you can't open it, try default location
 184                                                 if (ableToOpen == 1) {
 185                                                         if (m->getDefaultPath() != "") { //default path is set
 186                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 187                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 188                                                                 ifstream in2;
 189                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 190                                                                 in2.close();
 191                                                                 fastaFileNames[i] = tryPath;
 192                                                         }
 193                                                 }
 194
 195                                                 //if you can't open it, try default location
 196                                                 if (ableToOpen == 1) {
 197                                                         if (m->getOutputDir() != "") { //default path is set
 198                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 199                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 200                                                                 ifstream in2;
 201                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 202                                                                 in2.close();
 203                                                                 fastaFileNames[i] = tryPath;
 204                                                         }
 205                                                 }
 206
 207                                                 in.close();
 208
 209                                                 if (ableToOpen == 1) {
 210                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 211                                                         //erase from file list
 212                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 213                                                         i--;
 214                                                 }else {
 215                                                         m->setFastaFile(fastaFileNames[i]);
 216                                                 }
 217                                         }
 218                                 }
 219
 220                                 //make sure there is at least one valid file left
 221                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 222                         }
 223
 224                         //if the user changes the output directory command factory will send this info to us in the output parameter
 225                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 226
 227                         maskfile = validParameter.validFile(parameters, "mask", false);
 228                         if (maskfile == "not found") { maskfile = "";  }
 229                         else if (maskfile != "default")  {
 230                                 if (inputDir != "") {
 231                                         string path = m->hasPath(maskfile);
 232                                         //if the user has not given a path then, add inputdir. else leave path alone.
 233                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 234                                 }
 235
 236                                 ifstream in;
 237                                 int     ableToOpen = m->openInputFile(maskfile, in);
 238                                 if (ableToOpen == 1) { abort = true; }
 239                                 in.close();
 240                         }
 241
 242                         string temp;
 243                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 244                         filter = m->isTrue(temp);
 245
 246                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 247                         m->setProcessors(temp);
 248                         m->mothurConvert(temp, processors);
 249
 250                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 251                         m->mothurConvert(temp, window);
 252
 253                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "20"; }
 254                         m->mothurConvert(temp, numwanted);
 255
 256                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 257                         save = m->isTrue(temp);
 258                         rdb->save = save;
 259                         if (save) { //clear out old references
 260                                 rdb->clearMemory();
 261                         }
 262
 263                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 264                         templatefile = validParameter.validFile(parameters, "reference", true);
 265                         if (templatefile == "not found") {
 266                                 //check for saved reference sequences
 267                                 if (rdb->referenceSeqs.size() != 0) {
 268                                         templatefile = "saved";
 269                                 }else {
 270                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 271                                         m->mothurOutEndLine();
 272                                         abort = true;
 273                                 }
 274                         }else if (templatefile == "not open") { abort = true; }
 275                         else {  if (save) {     rdb->setSavedReference(templatefile);   }       }
 276
 277
 278                 }
 279         }
 280         catch(exception& e) {
 281                 m->errorOut(e, "ChimeraCcodeCommand", "ChimeraCcodeCommand");
 282                 exit(1);
 283         }
 284 }
 285 //***************************************************************************************************************
 286 int ChimeraCcodeCommand::execute(){
 287         try{
 288
 289                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 290
 291                 for (int s = 0; s < fastaFileNames.size(); s++) {
 292
 293                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 294
 295                         int start = time(NULL);
 296
 297                         //set user options
 298                         if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 299
 300                         chimera = new Ccode(fastaFileNames[s], templatefile, filter, maskfile, window, numwanted, outputDir);
 301
 302                         //is your template aligned?
 303                         if (chimera->getUnaligned()) { m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); delete chimera; return 0; }
 304                         templateSeqsLength = chimera->getLength();
 305
 306                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 307                         string outputFileName, accnosFileName;
 308                         if (maskfile != "") {
 309                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + getOutputFileNameTag("chimera");
 310                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + getOutputFileNameTag("accnos");
 311                         }else {
 312                                 outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + getOutputFileNameTag("chimera");
 313                                 accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + getOutputFileNameTag("accnos");
 314
 315                         }
 316
 317                         string mapInfo = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("mapinfo");
 318
 319
 320                         if (m->control_pressed) { delete chimera;  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } outputTypes.clear(); return 0;        }
 321
 322                 #ifdef USE_MPI
 323
 324                                 int pid, numSeqsPerProcessor;
 325                                 int tag = 2001;
 326                                 vector<unsigned long long> MPIPos;
 327
 328                                 MPI_Status status;
 329                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 330                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 331
 332                                 MPI_File inMPI;
 333                                 MPI_File outMPI;
 334                                 MPI_File outMPIAccnos;
 335
 336                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 337                                 int inMode=MPI_MODE_RDONLY;
 338
 339                                 char outFilename[1024];
 340                                 strcpy(outFilename, outputFileName.c_str());
 341
 342                                 char outAccnosFilename[1024];
 343                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 344
 345                                 char inFileName[1024];
 346                                 strcpy(inFileName, fastaFileNames[s].c_str());
 347
 348                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 349                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 350                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 351
 352                                 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;  }
 353
 354                                 if (pid == 0) { //you are the root process
 355                                         string outTemp = "For full window mapping info refer to " + mapInfo + "\n";
 356
 357                                         //print header
 358                                         int length = outTemp.length();
 359                                         char* buf2 = new char[length];
 360                                         memcpy(buf2, outTemp.c_str(), length);
 361
 362                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 363                                         delete buf2;
 364
 365                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 366
 367                                         //send file positions to all processes
 368                                         for(int i = 1; i < processors; i++) {
 369                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 370                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 371                                         }
 372
 373                                         //figure out how many sequences you have to align
 374                                         numSeqsPerProcessor = numSeqs / processors;
 375                                         int startIndex =  pid * numSeqsPerProcessor;
 376                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 377
 378
 379                                         //align your part
 380                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 381
 382                                         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;  }
 383
 384                                 }else{ //you are a child process
 385                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 386                                         MPIPos.resize(numSeqs+1);
 387                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 388
 389                                         //figure out how many sequences you have to align
 390                                         numSeqsPerProcessor = numSeqs / processors;
 391                                         int startIndex =  pid * numSeqsPerProcessor;
 392                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 393
 394
 395                                         //align your part
 396                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
 397
 398                                         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;  }
 399                                 }
 400
 401                                 //close files
 402                                 MPI_File_close(&inMPI);
 403                                 MPI_File_close(&outMPI);
 404                                 MPI_File_close(&outMPIAccnos);
 405
 406                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 407
 408                 #else
 409                         ofstream outHeader;
 410                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + maskfile + "ccode.chimeras.tempHeader";
 411                         m->openOutputFile(tempHeader, outHeader);
 412
 413                         outHeader << "For full window mapping info refer to " << mapInfo << endl << endl;
 414
 415                         outHeader.close();
 416
 417
 418
 419                         //break up file
 420                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 421                                 vector<unsigned long long> positions = m->divideFile(fastaFileNames[s], processors);
 422
 423                                 for (int i = 0; i < (positions.size()-1); i++) {
 424                                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 425                                 }
 426
 427                                 if(processors == 1){
 428
 429                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 430
 431                                         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; }
 432
 433                                 }else{
 434                                         processIDS.resize(0);
 435
 436                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
 437
 438                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 439                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 440
 441                                         //append output files
 442                                         for(int i=1;i<processors;i++){
 443                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 444                                                 m->mothurRemove((outputFileName + toString(processIDS[i]) + ".temp"));
 445                                         }
 446
 447                                         //append output files
 448                                         for(int i=1;i<processors;i++){
 449                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 450                                                 m->mothurRemove((accnosFileName + toString(processIDS[i]) + ".temp"));
 451                                         }
 452
 453                                         if (m->control_pressed) {
 454                                                 m->mothurRemove(outputFileName);
 455                                                 m->mothurRemove(accnosFileName);
 456                                                 for (int j = 0; j < outputNames.size(); j++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear();
 457                                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 458                                                 delete chimera;
 459                                                 return 0;
 460                                         }
 461
 462                                 }
 463
 464                         #else
 465                                 lines.push_back(new linePair(0, 1000));
 466                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
 467
 468                                 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; }
 469
 470                         #endif
 471
 472                         m->appendFiles(outputFileName, tempHeader);
 473
 474                         m->mothurRemove(outputFileName);
 475                         rename(tempHeader.c_str(), outputFileName.c_str());
 476                 #endif
 477
 478                         delete chimera;
 479
 480                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 481                         outputNames.push_back(mapInfo); outputTypes["mapinfo"].push_back(mapInfo);
 482                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 483
 484                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 485
 486                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 487                 }
 488
 489
 490                 //set accnos file as new current accnosfile
 491                 string current = "";
 492                 itTypes = outputTypes.find("accnos");
 493                 if (itTypes != outputTypes.end()) {
 494                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 495                 }
 496
 497                 m->mothurOutEndLine();
 498                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 499                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 500                 m->mothurOutEndLine();
 501
 502                 return 0;
 503
 504         }
 505         catch(exception& e) {
 506                 m->errorOut(e, "ChimeraCcodeCommand", "execute");
 507                 exit(1);
 508         }
 509 }
 510 //**********************************************************************************************************************
 511
 512 int ChimeraCcodeCommand::driver(linePair* filePos, string outputFName, string filename, string accnos){
 513         try {
 514                 ofstream out;
 515                 m->openOutputFile(outputFName, out);
 516
 517                 ofstream out2;
 518                 m->openOutputFile(accnos, out2);
 519
 520                 ifstream inFASTA;
 521                 m->openInputFile(filename, inFASTA);
 522
 523                 inFASTA.seekg(filePos->start);
 524
 525                 bool done = false;
 526                 int count = 0;
 527
 528                 while (!done) {
 529
 530                         if (m->control_pressed) {       return 1;       }
 531
 532                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 533
 534                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 535
 536                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 537                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 538                                 }else{
 539                                         //find chimeras
 540                                         chimera->getChimeras(candidateSeq);
 541
 542                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 543
 544                                         //print results
 545                                         chimera->print(out, out2);
 546                                 }
 547                                 count++;
 548                         }
 549                         delete candidateSeq;
 550
 551                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 552                                 unsigned long long pos = inFASTA.tellg();
 553                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 554                         #else
 555                                 if (inFASTA.eof()) { break; }
 556                         #endif
 557
 558                         //report progress
 559                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 560                 }
 561                 //report progress
 562                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 563
 564                 out.close();
 565                 out2.close();
 566                 inFASTA.close();
 567
 568                 return count;
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "ChimeraCcodeCommand", "driver");
 572                 exit(1);
 573         }
 574 }
 575 //**********************************************************************************************************************
 576 #ifdef USE_MPI
 577 int ChimeraCcodeCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long long>& MPIPos){
 578         try {
 579
 580                 MPI_Status status;
 581                 int pid;
 582                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 583
 584                 for(int i=0;i<num;i++){
 585
 586                         if (m->control_pressed) { return 0; }
 587
 588                         //read next sequence
 589                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 590
 591                         char* buf4 = new char[length];
 592
 593                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 594
 595                         string tempBuf = buf4;
 596                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 597                         istringstream iss (tempBuf,istringstream::in);
 598                         delete buf4;
 599
 600                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 601
 602                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 603
 604                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 605                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 606                                 }else{
 607                                         //find chimeras
 608                                         chimera->getChimeras(candidateSeq);
 609
 610                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 611
 612                                         //print results
 613                                         chimera->print(outMPI, outAccMPI);
 614                                 }
 615                         }
 616                         delete candidateSeq;
 617
 618                         //report progress
 619                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 620                 }
 621                 //report progress
 622                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 623
 624
 625                 return 0;
 626         }
 627         catch(exception& e) {
 628                 m->errorOut(e, "ChimeraCcodeCommand", "driverMPI");
 629                 exit(1);
 630         }
 631 }
 632 #endif
 633
 634 /**************************************************************************************************/
 635
 636 int ChimeraCcodeCommand::createProcesses(string outputFileName, string filename, string accnos) {
 637         try {
 638 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 639                 int process = 0;
 640                 int num = 0;
 641
 642                 //loop through and create all the processes you want
 643                 while (process != processors) {
 644                         int pid = fork();
 645
 646                         if (pid > 0) {
 647                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 648                                 process++;
 649                         }else if (pid == 0){
 650                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 651
 652                                 //pass numSeqs to parent
 653                                 ofstream out;
 654                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 655                                 m->openOutputFile(tempFile, out);
 656                                 out << num << endl;
 657                                 out.close();
 658
 659                                 exit(0);
 660                         }else {
 661                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 662                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 663                                 exit(0);
 664                         }
 665                 }
 666
 667                 //force parent to wait until all the processes are done
 668                 for (int i=0;i<processors;i++) {
 669                         int temp = processIDS[i];
 670                         wait(&temp);
 671                 }
 672
 673                 for (int i = 0; i < processIDS.size(); i++) {
 674                         ifstream in;
 675                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 676                         m->openInputFile(tempFile, in);
 677                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 678                         in.close(); m->mothurRemove(tempFile);
 679                 }
 680
 681                 return num;
 682 #endif
 683         }
 684         catch(exception& e) {
 685                 m->errorOut(e, "ChimeraCcodeCommand", "createProcesses");
 686                 exit(1);
 687         }
 688 }
 689 //**********************************************************************************************************************
 690