chimeracheckcommand.cpp

   1 /*
   2  *  chimeracheckcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/31/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeracheckcommand.h"
  11 #include "referencedb.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraCheckCommand::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 pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter psvg("svg", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psvg);
  20                 CommandParameter pincrement("increment", "Number", "", "10", "", "", "",false,false); parameters.push_back(pincrement);
  21                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  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, "ChimeraCheckCommand", "setParameters");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 string ChimeraCheckCommand::getHelpString(){
  38         try {
  39                 string helpString = "";
  40                 helpString += "The chimera.check command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  41                 helpString += "This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n";
  42                 helpString += "The chimera.check command parameters are fasta, reference, processors, ksize, increment, svg and name.\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 processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  47 #ifdef USE_MPI
  48                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  49 #endif
  50                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n";
  51                 helpString += "The ksize parameter allows you to input kmersize, default is 7. \n";
  52                 helpString += "The svg parameter allows you to specify whether or not you would like a svg file outputted for each query sequence, default is False.\n";
  53                 helpString += "The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n";
  54                 helpString += "You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n";
  55                 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.";
  56                 helpString += "The chimera.check command should be in the following format: \n";
  57                 helpString += "chimera.check(fasta=yourFastaFile, reference=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n";
  58                 helpString += "Example: chimera.check(fasta=AD.fasta, reference=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n";
  59                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  60                 return helpString;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "ChimeraCheckCommand", "getHelpString");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 string ChimeraCheckCommand::getOutputFileNameTag(string type, string inputName=""){
  69         try {
  70         string outputFileName = "";
  71                 map<string, vector<string> >::iterator it;
  72
  73         //is this a type this command creates
  74         it = outputTypes.find(type);
  75         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  76         else {
  77             if (type == "chimera") {  outputFileName =  "chimeracheck.chimeras"; }
  78             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  79         }
  80         return outputFileName;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "ChimeraCcodeCommand", "getOutputFileNameTag");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 ChimeraCheckCommand::ChimeraCheckCommand(){
  89         try {
  90                 abort = true; calledHelp = true;
  91                 setParameters();
  92                 vector<string> tempOutNames;
  93                 outputTypes["chimera"] = tempOutNames;
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
  97                 exit(1);
  98         }
  99 }
 100 //***************************************************************************************************************
 101 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
 102         try {
 103                 abort = false; calledHelp = false;
 104                 ReferenceDB* rdb = ReferenceDB::getInstance();
 105
 106                 //allow user to run help
 107                 if(option == "help") { help(); abort = true; calledHelp = true; }
 108                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 109
 110                 else {
 111                         vector<string> myArray = setParameters();
 112
 113                         OptionParser parser(option);
 114                         map<string,string> parameters = parser.getParameters();
 115
 116                         ValidParameters validParameter("chimera.check");
 117                         map<string,string>::iterator it;
 118
 119                         //check to make sure all parameters are valid for command
 120                         for (it = parameters.begin(); it != parameters.end(); it++) {
 121                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 122                         }
 123
 124                         vector<string> tempOutNames;
 125                         outputTypes["chimera"] = tempOutNames;
 126
 127                         //if the user changes the input directory command factory will send this info to us in the output parameter
 128                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 129                         if (inputDir == "not found"){   inputDir = "";          }
 130                         else {
 131                                 it = parameters.find("reference");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         string path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 137                                 }
 138                         }
 139
 140                         //check for required parameters
 141                         fastafile = validParameter.validFile(parameters, "fasta", false);
 142                         if (fastafile == "not found") {
 143                                 //if there is a current fasta file, use it
 144                                 string filename = m->getFastaFile();
 145                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 146                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 147                         }else {
 148                                 m->splitAtDash(fastafile, fastaFileNames);
 149
 150                                 //go through files and make sure they are good, if not, then disregard them
 151                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 152
 153                                         bool ignore = false;
 154                                         if (fastaFileNames[i] == "current") {
 155                                                 fastaFileNames[i] = m->getFastaFile();
 156                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 157                                                 else {
 158                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 159                                                         //erase from file list
 160                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 161                                                         i--;
 162                                                 }
 163                                         }
 164
 165                                         if (!ignore) {
 166
 167
 168                                                 if (inputDir != "") {
 169                                                         string path = m->hasPath(fastaFileNames[i]);
 170                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 172                                                 }
 173
 174                                                 int ableToOpen;
 175                                                 ifstream in;
 176
 177                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 178
 179                                                 //if you can't open it, try default location
 180                                                 if (ableToOpen == 1) {
 181                                                         if (m->getDefaultPath() != "") { //default path is set
 182                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 183                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 184                                                                 ifstream in2;
 185                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 186                                                                 in2.close();
 187                                                                 fastaFileNames[i] = tryPath;
 188                                                         }
 189                                                 }
 190
 191                                                 //if you can't open it, try default location
 192                                                 if (ableToOpen == 1) {
 193                                                         if (m->getOutputDir() != "") { //default path is set
 194                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 195                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 196                                                                 ifstream in2;
 197                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 198                                                                 in2.close();
 199                                                                 fastaFileNames[i] = tryPath;
 200                                                         }
 201                                                 }
 202
 203                                                 in.close();
 204
 205                                                 if (ableToOpen == 1) {
 206                                                         m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
 207                                                         //erase from file list
 208                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 209                                                         i--;
 210                                                 }else {
 211                                                         m->setFastaFile(fastaFileNames[i]);
 212                                                 }
 213                                         }
 214                                 }
 215
 216                                 //make sure there is at least one valid file left
 217                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 218                         }
 219
 220                         //if the user changes the output directory command factory will send this info to us in the output parameter
 221                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 222
 223                         namefile = validParameter.validFile(parameters, "name", false);
 224                         if (namefile == "not found") { namefile = ""; }
 225                         else {
 226                                 m->splitAtDash(namefile, nameFileNames);
 227
 228                                 //go through files and make sure they are good, if not, then disregard them
 229                                 for (int i = 0; i < nameFileNames.size(); i++) {
 230
 231                                         bool ignore = false;
 232                                         if (nameFileNames[i] == "current") {
 233                                                 nameFileNames[i] = m->getNameFile();
 234                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 235                                                 else {
 236                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 237                                                         //erase from file list
 238                                                         nameFileNames.erase(nameFileNames.begin()+i);
 239                                                         i--;
 240                                                 }
 241                                         }
 242
 243                                         if (!ignore) {
 244
 245                                                 if (inputDir != "") {
 246                                                         string path = m->hasPath(nameFileNames[i]);
 247                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 248                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 249                                                 }
 250
 251                                                 int ableToOpen;
 252                                                 ifstream in;
 253
 254                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 255
 256                                                 //if you can't open it, try default location
 257                                                 if (ableToOpen == 1) {
 258                                                         if (m->getDefaultPath() != "") { //default path is set
 259                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 260                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 261                                                                 ifstream in2;
 262                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 263                                                                 in2.close();
 264                                                                 nameFileNames[i] = tryPath;
 265                                                         }
 266                                                 }
 267
 268                                                 //if you can't open it, try default location
 269                                                 if (ableToOpen == 1) {
 270                                                         if (m->getOutputDir() != "") { //default path is set
 271                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 272                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 273                                                                 ifstream in2;
 274                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 275                                                                 in2.close();
 276                                                                 nameFileNames[i] = tryPath;
 277                                                         }
 278                                                 }
 279
 280                                                 in.close();
 281
 282                                                 if (ableToOpen == 1) {
 283                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 284                                                         //erase from file list
 285                                                         nameFileNames.erase(nameFileNames.begin()+i);
 286                                                         i--;
 287                                                 }else {
 288                                                         m->setNameFile(nameFileNames[i]);
 289                                                 }
 290                                         }
 291                                 }
 292
 293                                 //make sure there is at least one valid file left
 294                                 if (nameFileNames.size() != 0) {
 295                                         if (nameFileNames.size() != fastaFileNames.size()) {
 296                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 297                                                  abort = true;
 298                                         }
 299                                 }
 300                         }
 301
 302                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 303                         m->setProcessors(temp);
 304                         m->mothurConvert(temp, processors);
 305
 306                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 307                         save = m->isTrue(temp);
 308                         rdb->save = save;
 309                         if (save) { //clear out old references
 310                                 rdb->clearMemory();
 311                         }
 312
 313                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 314                         templatefile = validParameter.validFile(parameters, "reference", true);
 315                         if (templatefile == "not found") {
 316                                 //check for saved reference sequences
 317                                 if (rdb->referenceSeqs.size() != 0) {
 318                                         templatefile = "saved";
 319                                 }else {
 320                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 321                                         m->mothurOutEndLine();
 322                                         abort = true;
 323                                 }
 324                         }else if (templatefile == "not open") { abort = true; }
 325                         else {  if (save) {     rdb->setSavedReference(templatefile);   }       }
 326
 327
 328                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 329                         m->mothurConvert(temp, ksize);
 330
 331                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 332                         svg = m->isTrue(temp);
 333                         if (nameFileNames.size() != 0) { svg = true; }
 334
 335                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 336                         m->mothurConvert(temp, increment);
 337                 }
 338         }
 339         catch(exception& e) {
 340                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 341                 exit(1);
 342         }
 343 }
 344 //***************************************************************************************************************
 345
 346 int ChimeraCheckCommand::execute(){
 347         try{
 348
 349                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 350
 351                 for (int i = 0; i < fastaFileNames.size(); i++) {
 352
 353                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 354
 355                         int start = time(NULL);
 356
 357                         string thisNameFile = "";
 358                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 359
 360                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 361
 362                         if (m->control_pressed) { delete chimera;       return 0;       }
 363
 364                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 365                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + getOutputFileNameTag("chimera");
 366                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 367
 368                 #ifdef USE_MPI
 369
 370                                 int pid, numSeqsPerProcessor;
 371                                 int tag = 2001;
 372                                 vector<unsigned long long> MPIPos;
 373
 374                                 MPI_Status status;
 375                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 376                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 377
 378                                 MPI_File inMPI;
 379                                 MPI_File outMPI;
 380
 381                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 382                                 int inMode=MPI_MODE_RDONLY;
 383
 384                                 char outFilename[1024];
 385                                 strcpy(outFilename, outputFileName.c_str());
 386
 387                                 char inFileName[1024];
 388                                 strcpy(inFileName, fastaFileNames[i].c_str());
 389
 390                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 391                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 392
 393                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        } outputTypes.clear(); delete chimera; return 0;  }
 394
 395                                 if (pid == 0) { //you are the root process
 396                                         MPIPos = m->setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 397
 398                                         //send file positions to all processes
 399                                         for(int j = 1; j < processors; j++) {
 400                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 401                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 402                                         }
 403
 404                                         //figure out how many sequences you have to align
 405                                         numSeqsPerProcessor = numSeqs / processors;
 406                                         int startIndex =  pid * numSeqsPerProcessor;
 407                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 408
 409
 410                                         //align your part
 411                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 412
 413                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        }   outputTypes.clear(); delete chimera; return 0;  }
 414
 415                                         //wait on chidren
 416                                         for(int j = 1; j < processors; j++) {
 417                                                 char buf[5];
 418                                                 MPI_Recv(buf, 5, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 419                                         }
 420                                 }else{ //you are a child process
 421                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 422                                         MPIPos.resize(numSeqs+1);
 423                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 424
 425                                         //figure out how many sequences you have to align
 426                                         numSeqsPerProcessor = numSeqs / processors;
 427                                         int startIndex =  pid * numSeqsPerProcessor;
 428                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 429
 430                                         //align your part
 431                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 432
 433                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  outputTypes.clear(); delete chimera; return 0;  }
 434
 435                                         //tell parent you are done.
 436                                         char buf[5];
 437                                         strcpy(buf, "done");
 438                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 439                                 }
 440
 441                                 //close files
 442                                 MPI_File_close(&inMPI);
 443                                 MPI_File_close(&outMPI);
 444                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 445                 #else
 446
 447
 448
 449                         //break up file
 450                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 451                                 vector<unsigned long long> positions = m->divideFile(fastaFileNames[i], processors);
 452
 453                                 for (int s = 0; s < (positions.size()-1); s++) {
 454                                         lines.push_back(new linePair(positions[s], positions[(s+1)]));
 455                                 }
 456
 457                                 if(processors == 1){
 458                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 459
 460                                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } for (int j = 0; j < lines.size(); j++) {  delete lines[j];  } outputTypes.clear();  lines.clear(); delete chimera; return 0; }
 461
 462                                 }else{
 463                                         processIDS.resize(0);
 464
 465                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 466
 467                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 468
 469                                         //append output files
 470                                         for(int j=1;j<processors;j++){
 471                                                 m->appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 472                                                 m->mothurRemove((outputFileName + toString(processIDS[j]) + ".temp"));
 473                                         }
 474
 475                                         if (m->control_pressed) {
 476                                                 for (int j = 0; j < outputNames.size(); j++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear();
 477                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 478                                                 delete chimera;
 479                                                 return 0;
 480                                         }
 481                                 }
 482
 483                         #else
 484                                 lines.push_back(new linePair(0, 1000));
 485                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 486
 487                                 if (m->control_pressed) { for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear(); for (int j = 0; j < outputNames.size(); j++) {  m->mothurRemove(outputNames[j]);        } outputTypes.clear(); delete chimera; return 0; }
 488                         #endif
 489                 #endif
 490                         delete chimera;
 491                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 492
 493                         m->mothurOutEndLine(); m->mothurOut("This method does not determine if a sequence is chimeric, but allows you to make that determination based on the IS values."); m->mothurOutEndLine();
 494                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 495
 496                 }
 497
 498                 m->mothurOutEndLine();
 499                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 500                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 501                 m->mothurOutEndLine();
 502
 503                 return 0;
 504
 505         }
 506         catch(exception& e) {
 507                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 508                 exit(1);
 509         }
 510 }
 511 //**********************************************************************************************************************
 512
 513 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 514         try {
 515                 ofstream out;
 516                 m->openOutputFile(outputFName, out);
 517
 518                 ofstream 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                                 //find chimeras
 536                                 chimera->getChimeras(candidateSeq);
 537
 538                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 539
 540                                 //print results
 541                                 chimera->print(out, out2);
 542                         }
 543                         delete candidateSeq;
 544
 545                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 546                                 unsigned long long pos = inFASTA.tellg();
 547                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 548                         #else
 549                                 if (inFASTA.eof()) { break; }
 550                         #endif
 551
 552                         //report progress
 553                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 554                 }
 555                 //report progress
 556                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 557
 558                 out.close();
 559                 inFASTA.close();
 560
 561                 return count;
 562         }
 563         catch(exception& e) {
 564                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 565                 exit(1);
 566         }
 567 }
 568 //**********************************************************************************************************************
 569 #ifdef USE_MPI
 570 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long long>& MPIPos){
 571         try {
 572                 MPI_File outAccMPI;
 573                 MPI_Status status;
 574                 int pid;
 575                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 576
 577                 for(int i=0;i<num;i++){
 578
 579                         if (m->control_pressed) { return 0; }
 580
 581                         //read next sequence
 582                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 583
 584                         char* buf4 = new char[length];
 585                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 586
 587                         string tempBuf = buf4;
 588                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 589                         istringstream iss (tempBuf,istringstream::in);
 590                         delete buf4;
 591
 592                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 593
 594                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 595                                 //find chimeras
 596                                 chimera->getChimeras(candidateSeq);
 597
 598                                 //print results
 599                                 chimera->print(outMPI, outAccMPI);
 600                         }
 601                         delete candidateSeq;
 602
 603                         //report progress
 604                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 605                 }
 606                 //report progress
 607                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 608
 609                 return 0;
 610         }
 611         catch(exception& e) {
 612                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 613                 exit(1);
 614         }
 615 }
 616 #endif
 617
 618 /**************************************************************************************************/
 619
 620 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 621         try {
 622 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 623                 int process = 0;
 624                 int num = 0;
 625
 626                 //loop through and create all the processes you want
 627                 while (process != processors) {
 628                         int pid = fork();
 629
 630                         if (pid > 0) {
 631                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 632                                 process++;
 633                         }else if (pid == 0){
 634                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 635
 636                                 //pass numSeqs to parent
 637                                 ofstream out;
 638                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 639                                 m->openOutputFile(tempFile, out);
 640                                 out << num << endl;
 641                                 out.close();
 642
 643                                 exit(0);
 644                         }else {
 645                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 646                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 647                                 exit(0);
 648                         }
 649                 }
 650
 651                 //force parent to wait until all the processes are done
 652                 for (int i=0;i<processors;i++) {
 653                         int temp = processIDS[i];
 654                         wait(&temp);
 655                 }
 656
 657                 for (int i = 0; i < processIDS.size(); i++) {
 658                         ifstream in;
 659                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 660                         m->openInputFile(tempFile, in);
 661                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 662                         in.close(); m->mothurRemove(tempFile);
 663                 }
 664
 665                 return num;
 666 #endif
 667         }
 668         catch(exception& e) {
 669                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 670                 exit(1);
 671         }
 672 }
 673 /**************************************************************************************************/
 674
 675