trimflowscommand.cpp

   1 /*
   2  *  trimflowscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 12/22/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "trimflowscommand.h"
  11 #include "needlemanoverlap.hpp"
  12
  13 //**********************************************************************************************************************
  14
  15 vector<string> TrimFlowsCommand::getValidParameters(){
  16         try {
  17                 string Array[] =  {"flow", "maxflows", "minflows",
  18                         "fasta", "minlength", "maxlength", "maxhomop", "signal", "noise"
  19                         "oligos", "pdiffs", "bdiffs", "tdiffs",
  20                         "allfiles", "processors",
  21                         "outputdir","inputdir"
  22
  23                 };
  24                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "TrimFlowsCommand", "getValidParameters");
  29                 exit(1);
  30         }
  31 }
  32
  33 //**********************************************************************************************************************
  34
  35 vector<string> TrimFlowsCommand::getRequiredParameters(){
  36         try {
  37                 string Array[] =  {"flow"};
  38                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  39                 return myArray;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "TrimFlowsCommand", "getRequiredParameters");
  43                 exit(1);
  44         }
  45 }
  46
  47 //**********************************************************************************************************************
  48
  49 vector<string> TrimFlowsCommand::getRequiredFiles(){
  50         try {
  51                 vector<string> myArray;
  52                 return myArray;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "TrimFlowsCommand", "getRequiredFiles");
  56                 exit(1);
  57         }
  58 }
  59
  60 //**********************************************************************************************************************
  61
  62 TrimFlowsCommand::TrimFlowsCommand(){
  63         try {
  64                 abort = true; calledHelp = true;
  65                 vector<string> tempOutNames;
  66                 outputTypes["flow"] = tempOutNames;
  67                 outputTypes["fasta"] = tempOutNames;
  68         }
  69         catch(exception& e) {
  70                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
  71                 exit(1);
  72         }
  73 }
  74
  75 //***************************************************************************************************************
  76
  77 TrimFlowsCommand::~TrimFlowsCommand(){  /*      do nothing      */      }
  78
  79 //***************************************************************************************************************
  80
  81 void TrimFlowsCommand::help(){
  82         try {
  83                 m->mothurOut("The trim.flows command reads a flowgram file and creates .....\n");
  84                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
  85                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.flows.\n\n");
  86
  87         }
  88         catch(exception& e) {
  89                 m->errorOut(e, "TrimFlowsCommand", "help");
  90                 exit(1);
  91         }
  92 }
  93
  94 //**********************************************************************************************************************
  95
  96 TrimFlowsCommand::TrimFlowsCommand(string option)  {
  97         try {
  98
  99                 abort = false; calledHelp = false;
 100                 comboStarts = 0;
 101
 102                 //allow user to run help
 103                 if(option == "help") { help(); abort = true; calledHelp = true; }
 104
 105                 else {
 106                         //valid paramters for this command
 107                         string AlignArray[] =  {"flow", "maxflows", "minflows",
 108                                 "fasta", "minlength", "maxlength", "maxhomop", "signal", "noise",
 109                                 "oligos", "pdiffs", "bdiffs", "tdiffs",
 110                                 "allfiles", "processors",
 111
 112                                 //                      "group",
 113                                 "outputdir","inputdir"
 114
 115                         };
 116
 117                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
 118
 119                         OptionParser parser(option);
 120                         map<string,string> parameters = parser.getParameters();
 121
 122                         ValidParameters validParameter;
 123                         map<string,string>::iterator it;
 124
 125                         //check to make sure all parameters are valid for command
 126                         for (it = parameters.begin(); it != parameters.end(); it++) {
 127                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 128                         }
 129
 130                         //initialize outputTypes
 131                         vector<string> tempOutNames;
 132                         outputTypes["flow"] = tempOutNames;
 133                         outputTypes["fasta"] = tempOutNames;
 134
 135                         //if the user changes the input directory command factory will send this info to us in the output parameter
 136                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 137                         if (inputDir == "not found"){   inputDir = "";          }
 138                         else {
 139                                 string path;
 140                                 it = parameters.find("flow");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["flow"] = inputDir + it->second;             }
 146                                 }
 147
 148                                 it = parameters.find("oligos");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 154                                 }
 155
 156
 157 //                              it = parameters.find("group");
 158 //                              //user has given a template file
 159 //                              if(it != parameters.end()){
 160 //                                      path = m->hasPath(it->second);
 161 //                                      //if the user has not given a path then, add inputdir. else leave path alone.
 162 //                                      if (path == "") {       parameters["group"] = inputDir + it->second;            }
 163 //                              }
 164                         }
 165
 166
 167                         //check for required parameters
 168                         flowFileName = validParameter.validFile(parameters, "flow", true);
 169                         if (flowFileName == "not found") { m->mothurOut("flow is a required parameter for the trim.flows command."); m->mothurOutEndLine(); abort = true; }
 170                         else if (flowFileName == "not open") { abort = true; }
 171
 172                         //if the user changes the output directory command factory will send this info to us in the output parameter
 173                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 174                                 outputDir = "";
 175                                 outputDir += m->hasPath(flowFileName); //if user entered a file with a path then preserve it
 176                         }
 177
 178
 179                         //check for optional parameter and set defaults
 180                         // ...at some point should added some additional type checking...
 181
 182                         string temp;
 183                         temp = validParameter.validFile(parameters, "minflows", false); if (temp == "not found") { temp = "360"; }
 184                         convert(temp, minFlows);
 185
 186                         temp = validParameter.validFile(parameters, "maxflows", false); if (temp == "not found") { temp = "720"; }
 187                         convert(temp, maxFlows);
 188
 189
 190                         temp = validParameter.validFile(parameters, "oligos", true);
 191                         if (temp == "not found")        {       oligoFileName = "";             }
 192                         else if(temp == "not open")     {       abort = true;                   }
 193                         else                                            {       oligoFileName = temp;   }
 194
 195                         temp = validParameter.validFile(parameters, "fasta", false);            if (temp == "not found"){       fasta = 0;              }
 196                         else if(m->isTrue(temp))        {       fasta = 1;      }
 197
 198                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found"){       temp = "9";             }
 199                         convert(temp, maxHomoP);
 200
 201                         temp = validParameter.validFile(parameters, "signal", false);           if (temp == "not found"){       temp = "0.50";  }
 202                         convert(temp, signal);
 203
 204                         temp = validParameter.validFile(parameters, "noise", false);            if (temp == "not found"){       temp = "0.70";  }
 205                         convert(temp, noise);
 206
 207                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found"){       temp = "0";             }
 208                         convert(temp, minLength);
 209
 210                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found"){       temp = "0";             }
 211                         convert(temp, maxLength);
 212
 213                         temp = validParameter.validFile(parameters, "bdiffs", false);           if (temp == "not found"){       temp = "0";             }
 214                         convert(temp, bdiffs);
 215
 216                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found"){       temp = "0";             }
 217                         convert(temp, pdiffs);
 218
 219                         temp = validParameter.validFile(parameters, "tdiffs", false);
 220                         if (temp == "not found"){ int tempTotal = pdiffs + bdiffs;  temp = toString(tempTotal); }
 221                         convert(temp, tdiffs);
 222                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs;       }
 223
 224                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found"){ temp = "T";           }
 225                         allFiles = m->isTrue(temp);
 226
 227                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){ temp = "1";           }
 228                         convert(temp, processors);
 229
 230                         if(oligoFileName == ""){        allFiles = 0;           }
 231
 232                         numFPrimers = 0;
 233                         numRPrimers = 0;
 234                 }
 235
 236         }
 237         catch(exception& e) {
 238                 m->errorOut(e, "TrimFlowsCommand", "TrimSeqsCommand");
 239                 exit(1);
 240         }
 241 }
 242
 243 //***************************************************************************************************************
 244
 245 int TrimFlowsCommand::execute(){
 246         try{
 247
 248                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 249
 250                 string trimFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "trim.flow";
 251                 outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
 252
 253                 string scrapFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "scrap.flow";
 254                 outputNames.push_back(scrapFlowFileName); outputTypes["flow"].push_back(scrapFlowFileName);
 255
 256                 string fastaFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.fasta";
 257                 if(fasta){
 258                         outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName);
 259                 }
 260
 261                 vector<unsigned long int> flowFilePos = getFlowFileBreaks();
 262                 for (int i = 0; i < (flowFilePos.size()-1); i++) {
 263                         lines.push_back(new linePair(flowFilePos[i], flowFilePos[(i+1)]));
 264                 }
 265
 266                 vector<vector<string> > barcodePrimerComboFileNames;
 267                 if(oligoFileName != ""){
 268                         getOligos(barcodePrimerComboFileNames);
 269                 }
 270
 271 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 272                 if(processors == 1){
 273                         driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 274                 }else{
 275                         createProcessesCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames);
 276                 }
 277 #else
 278                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 279 #endif
 280
 281                 if (m->control_pressed) {  return 0; }
 282
 283                 string flowFilesFileName;
 284                 ofstream output;
 285
 286                 if(allFiles){
 287
 288                         flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
 289                         m->openOutputFile(flowFilesFileName, output);
 290
 291                         for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 292                                 for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 293
 294                                         FILE * pFile;
 295                                         unsigned long int size;
 296
 297                                         //get num bytes in file
 298                                         pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
 299                                         if (pFile==NULL) perror ("Error opening file");
 300                                         else{
 301                                                 fseek (pFile, 0, SEEK_END);
 302                                                 size=ftell (pFile);
 303                                                 fclose (pFile);
 304                                         }
 305
 306                                         if(size < 10){
 307                                                 remove(barcodePrimerComboFileNames[i][j].c_str());
 308                                         }
 309                                         else{
 310                                                 output << barcodePrimerComboFileNames[i][j] << endl;
 311                                         }
 312                                 }
 313                         }
 314                         output.close();
 315                 }
 316                 else{
 317                         flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
 318                         m->openOutputFile(flowFilesFileName, output);
 319
 320                         output << trimFlowFileName << endl;
 321
 322                         output.close();
 323                 }
 324                 outputTypes["flow.files"].push_back(flowFilesFileName);
 325                 outputNames.push_back(flowFileName);
 326
 327                 m->mothurOutEndLine();
 328                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 329                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 330                 m->mothurOutEndLine();
 331
 332                 return 0;
 333         }
 334         catch(exception& e) {
 335                 m->errorOut(e, "TrimSeqsCommand", "execute");
 336                 exit(1);
 337         }
 338 }
 339
 340 //***************************************************************************************************************
 341
 342 int TrimFlowsCommand::driverCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames, linePair* line){
 343
 344         try {
 345
 346                 ofstream trimFlowFile;
 347                 m->openOutputFile(trimFlowFileName, trimFlowFile);
 348                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 349
 350                 ofstream scrapFlowFile;
 351                 m->openOutputFile(scrapFlowFileName, scrapFlowFile);
 352                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 353
 354                 if(line->start == 4){
 355                         scrapFlowFile << maxFlows << endl;
 356                         trimFlowFile << maxFlows << endl;
 357                         if(allFiles){
 358                                 for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 359                                         for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 360                                                 //                              barcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 361                                                 ofstream temp;
 362                                                 m->openOutputFile(barcodePrimerComboFileNames[i][j], temp);
 363                                                 temp << maxFlows << endl;
 364                                                 temp.close();
 365                                         }
 366                                 }
 367                         }
 368                 }
 369
 370                 FlowData flowData(numFlows, signal, noise, maxHomoP);
 371
 372                 ofstream fastaFile;
 373                 if(fasta){      m->openOutputFile(fastaFileName, fastaFile);    }
 374
 375                 ifstream flowFile;
 376                 m->openInputFile(flowFileName, flowFile);
 377
 378                 flowFile.seekg(line->start);
 379
 380                 int count = 0;
 381                 bool moreSeqs = 1;
 382
 383                 while(moreSeqs) {
 384
 385                         int success = 1;
 386                         int currentSeqDiffs = 0;
 387                         string trashCode = "";
 388
 389                         flowData.getNext(flowFile);
 390                         flowData.capFlows(maxFlows);
 391
 392                         Sequence currSeq = flowData.getSequence();
 393                         if(!flowData.hasMinFlows(minFlows)){    //screen to see if sequence is of a minimum number of flows
 394                                 success = 0;
 395                                 trashCode += 'l';
 396                         }
 397
 398                         if(minLength > 0 || maxLength > 0){     //screen to see if sequence is above and below a specific number of bases
 399                                 int seqLength = currSeq.getNumBases();
 400                                 if(seqLength < minLength || seqLength > maxLength){
 401                                         success = 0;
 402                                         trashCode += 'l';
 403                                 }
 404                         }
 405
 406                         int primerIndex = 0;
 407                         int barcodeIndex = 0;
 408
 409                         if(barcodes.size() != 0){
 410                                 success = stripBarcode(currSeq, barcodeIndex);
 411                                 if(success > bdiffs)            {       trashCode += 'b';       }
 412                                 else{ currentSeqDiffs += success;  }
 413                         }
 414
 415                         if(numFPrimers != 0){
 416                                 success = stripForward(currSeq, primerIndex);
 417                                 if(success > pdiffs)            {       trashCode += 'f';       }
 418                                 else{ currentSeqDiffs += success;  }
 419                         }
 420
 421                         if (currentSeqDiffs > tdiffs)   {       trashCode += 't';   }
 422
 423                         if(numRPrimers != 0){
 424                                 success = stripReverse(currSeq);
 425                                 if(!success)                            {       trashCode += 'r';       }
 426                         }
 427
 428                         if(trashCode.length() == 0){
 429
 430                                 flowData.printFlows(trimFlowFile);
 431
 432                                 if(fasta)       {       currSeq.printSequence(fastaFile);       }
 433
 434                                 if(allFiles){
 435                                         ofstream output;
 436                                         m->openOutputFileAppend(barcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 437                                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 438
 439                                         flowData.printFlows(output);
 440                                         output.close();
 441                                 }
 442                         }
 443                         else{
 444                                 flowData.printFlows(scrapFlowFile, trashCode);
 445                         }
 446
 447                         count++;
 448
 449                         //report progress
 450                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 451
 452 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 453                         unsigned long int pos = flowFile.tellg();
 454
 455                         if ((pos == -1) || (pos >= line->end)) { break; }
 456 #else
 457                         if (flowFile.eof()) { break; }
 458 #endif
 459
 460                 }
 461                 //report progress
 462                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 463
 464                 trimFlowFile.close();
 465                 scrapFlowFile.close();
 466                 flowFile.close();
 467                 if(fasta){      fastaFile.close();      }
 468
 469                 return count;
 470         }
 471         catch(exception& e) {
 472                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 473                 exit(1);
 474         }
 475 }
 476
 477 //***************************************************************************************************************
 478
 479 void TrimFlowsCommand::getOligos(vector<vector<string> >& outFlowFileNames){
 480         try {
 481                 ifstream oligosFile;
 482                 m->openInputFile(oligoFileName, oligosFile);
 483
 484                 string type, oligo, group;
 485
 486                 int indexPrimer = 0;
 487                 int indexBarcode = 0;
 488
 489                 while(!oligosFile.eof()){
 490
 491                         oligosFile >> type; m->gobble(oligosFile);      //get the first column value of the row - is it a comment or a feature we are interested in?
 492
 493                         if(type[0] == '#'){     //igore the line because there's a comment
 494                                 while (!oligosFile.eof())       {       char c = oligosFile.get(); if (c == 10 || c == 13){     break;  }       } // get rest of line if there's any crap there
 495                         }
 496                         else{                           //there's a feature we're interested in
 497
 498                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }                                  //make type case insensitive
 499
 500                                 oligosFile >> oligo;    //get the DNA sequence for the feature
 501
 502                                 for(int i=0;i<oligo.length();i++){      //make type case insensitive and change any U's to T's
 503                                         oligo[i] = toupper(oligo[i]);
 504                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 505                                 }
 506
 507                                 if(type == "FORWARD"){  //if the feature is a forward primer...
 508                                         group = "";
 509
 510                                         while (!oligosFile.eof())       {       // get rest of line in case there is a primer name = will have the name of the primer
 511                                                 char c = oligosFile.get();
 512                                                 if (c == 10 || c == 13){        break;  }
 513                                                 else if (c == 32 || c == 9){;} //space or tab
 514                                                 else {  group += c;  }
 515                                         }
 516
 517                                         //have we seen this primer already?
 518                                         map<string, int>::iterator itPrimer = primers.find(oligo);
 519                                         if (itPrimer != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 520
 521                                         primers[oligo]=indexPrimer; indexPrimer++;
 522                                         primerNameVector.push_back(group);
 523
 524                                 }
 525                                 else if(type == "REVERSE"){
 526                                         Sequence oligoRC("reverse", oligo);
 527                                         oligoRC.reverseComplement();
 528                                         revPrimer.push_back(oligoRC.getUnaligned());
 529                                 }
 530                                 else if(type == "BARCODE"){
 531                                         oligosFile >> group;
 532
 533                                         //check for repeat barcodes
 534                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 535                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 536
 537                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 538                                         barcodeNameVector.push_back(group);
 539                                 }
 540                                 else{
 541                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 542                                 }
 543                         }
 544
 545                         m->gobble(oligosFile);
 546                 }
 547                 oligosFile.close();
 548
 549                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 550
 551                 //add in potential combos
 552                 if(barcodeNameVector.size() == 0){
 553                         barcodes[""] = 0;
 554                         barcodeNameVector.push_back("");
 555                 }
 556
 557                 if(primerNameVector.size() == 0){
 558                         primers[""] = 0;
 559                         primerNameVector.push_back("");
 560                 }
 561
 562
 563                 outFlowFileNames.resize(barcodeNameVector.size());
 564                 for(int i=0;i<outFlowFileNames.size();i++){
 565                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 566                 }
 567
 568                 if(allFiles){
 569
 570                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 571                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 572
 573                                         string primerName = primerNameVector[itPrimer->second];
 574                                         string barcodeName = barcodeNameVector[itBar->second];
 575
 576                                         string comboGroupName = "";
 577                                         string fileName = "";
 578
 579                                         if(primerName == ""){
 580                                                 comboGroupName = barcodeNameVector[itBar->second];
 581                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 582                                         }
 583                                         else{
 584                                                 if(barcodeName == ""){
 585                                                         comboGroupName = primerNameVector[itPrimer->second];
 586                                                 }
 587                                                 else{
 588                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 589                                                 }
 590                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 591                                         }
 592
 593                                         outputNames.push_back(fileName);
 594                                         outputTypes["flow"].push_back(fileName);
 595                                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 596
 597                                         ofstream temp;
 598                                         m->openOutputFile(fileName, temp);
 599                                         temp.close();
 600                                 }
 601                         }
 602                 }
 603
 604                 numFPrimers = primers.size();
 605                 numRPrimers = revPrimer.size();
 606
 607         }
 608         catch(exception& e) {
 609                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 610                 exit(1);
 611         }
 612 }
 613
 614 //***************************************************************************************************************
 615
 616 int TrimFlowsCommand::stripBarcode(Sequence& seq, int& group){
 617         try {
 618
 619                 string rawSequence = seq.getUnaligned();
 620                 int success = bdiffs + 1;       //guilty until proven innocent
 621
 622                 //can you find the barcode
 623                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 624                         string oligo = it->first;
 625                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 626                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 627                                 break;
 628                         }
 629
 630                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 631                                 group = it->second;
 632                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 633
 634                                 success = 0;
 635                                 break;
 636                         }
 637                 }
 638
 639                 //if you found the barcode or if you don't want to allow for diffs
 640                 if ((bdiffs == 0) || (success == 0)) { return success;  }
 641
 642                 else { //try aligning and see if you can find it
 643
 644                         int maxLength = 0;
 645
 646                         Alignment* alignment;
 647                         if (barcodes.size() > 0) {
 648                                 map<string,int>::iterator it=barcodes.begin();
 649
 650                                 for(it;it!=barcodes.end();it++){
 651                                         if(it->first.length() > maxLength){
 652                                                 maxLength = it->first.length();
 653                                         }
 654                                 }
 655                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
 656
 657                         }else{ alignment = NULL; }
 658
 659                         //can you find the barcode
 660                         int minDiff = 1e6;
 661                         int minCount = 1;
 662                         int minGroup = -1;
 663                         int minPos = 0;
 664
 665                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 666                                 string oligo = it->first;
 667                                 //                              int length = oligo.length();
 668
 669                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
 670                                         success = bdiffs + 10;
 671                                         break;
 672                                 }
 673
 674                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 675                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
 676                                 oligo = alignment->getSeqAAln();
 677                                 string temp = alignment->getSeqBAln();
 678
 679                                 int alnLength = oligo.length();
 680
 681                                 for(int i=oligo.length()-1;i>=0;i--){
 682                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 683                                 }
 684                                 oligo = oligo.substr(0,alnLength);
 685                                 temp = temp.substr(0,alnLength);
 686
 687                                 int numDiff = countDiffs(oligo, temp);
 688
 689                                 if(numDiff < minDiff){
 690                                         minDiff = numDiff;
 691                                         minCount = 1;
 692                                         minGroup = it->second;
 693                                         minPos = 0;
 694                                         for(int i=0;i<alnLength;i++){
 695                                                 if(temp[i] != '-'){
 696                                                         minPos++;
 697                                                 }
 698                                         }
 699                                 }
 700                                 else if(numDiff == minDiff){
 701                                         minCount++;
 702                                 }
 703
 704                         }
 705
 706                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
 707                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
 708                         else{                                                                                                   //use the best match
 709                                 group = minGroup;
 710                                 seq.setUnaligned(rawSequence.substr(minPos));
 711                                 success = minDiff;
 712                         }
 713
 714                         if (alignment != NULL) {  delete alignment;  }
 715
 716                 }
 717
 718                 return success;
 719
 720         }
 721         catch(exception& e) {
 722                 m->errorOut(e, "TrimFlowsCommand", "stripBarcode");
 723                 exit(1);
 724         }
 725
 726 }
 727
 728 //***************************************************************************************************************
 729
 730 int TrimFlowsCommand::stripForward(Sequence& seq, int& group){
 731         try {
 732
 733                 string rawSequence = seq.getUnaligned();
 734                 int success = pdiffs + 1;       //guilty until proven innocent
 735
 736                 //can you find the primer
 737                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 738                         string oligo = it->first;
 739                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
 740                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 741                                 break;
 742                         }
 743
 744                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 745                                 group = it->second;
 746                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 747                                 success = 0;
 748                                 break;
 749                         }
 750                 }
 751
 752                 //if you found the barcode or if you don't want to allow for diffs
 753                 if ((pdiffs == 0) || (success == 0)) {  return success;  }
 754
 755                 else { //try aligning and see if you can find it
 756
 757                         int maxLength = 0;
 758
 759                         Alignment* alignment;
 760                         if (primers.size() > 0) {
 761                                 map<string,int>::iterator it=primers.begin();
 762
 763                                 for(it;it!=primers.end();it++){
 764                                         if(it->first.length() > maxLength){
 765                                                 maxLength = it->first.length();
 766                                         }
 767                                 }
 768                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
 769
 770                         }else{ alignment = NULL; }
 771
 772                         //can you find the barcode
 773                         int minDiff = 1e6;
 774                         int minCount = 1;
 775                         int minGroup = -1;
 776                         int minPos = 0;
 777
 778                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 779                                 string oligo = it->first;
 780                                 //                              int length = oligo.length();
 781
 782                                 if(rawSequence.length() < maxLength){
 783                                         success = pdiffs + 100;
 784                                         break;
 785                                 }
 786
 787                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 788                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
 789                                 oligo = alignment->getSeqAAln();
 790                                 string temp = alignment->getSeqBAln();
 791
 792                                 int alnLength = oligo.length();
 793
 794                                 for(int i=oligo.length()-1;i>=0;i--){
 795                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 796                                 }
 797                                 oligo = oligo.substr(0,alnLength);
 798                                 temp = temp.substr(0,alnLength);
 799
 800                                 int numDiff = countDiffs(oligo, temp);
 801
 802                                 if(numDiff < minDiff){
 803                                         minDiff = numDiff;
 804                                         minCount = 1;
 805                                         minGroup = it->second;
 806                                         minPos = 0;
 807                                         for(int i=0;i<alnLength;i++){
 808                                                 if(temp[i] != '-'){
 809                                                         minPos++;
 810                                                 }
 811                                         }
 812                                 }
 813                                 else if(numDiff == minDiff){
 814                                         minCount++;
 815                                 }
 816
 817                         }
 818
 819                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
 820                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
 821                         else{                                                                                                   //use the best match
 822                                 group = minGroup;
 823                                 seq.setUnaligned(rawSequence.substr(minPos));
 824                                 success = minDiff;
 825                         }
 826
 827                         if (alignment != NULL) {  delete alignment;  }
 828
 829                 }
 830
 831                 return success;
 832
 833         }
 834         catch(exception& e) {
 835                 m->errorOut(e, "TrimFlowsCommand", "stripForward");
 836                 exit(1);
 837         }
 838 }
 839
 840 //***************************************************************************************************************
 841
 842 bool TrimFlowsCommand::stripReverse(Sequence& seq){
 843         try {
 844
 845                 string rawSequence = seq.getUnaligned();
 846                 bool success = 0;       //guilty until proven innocent
 847
 848                 for(int i=0;i<numRPrimers;i++){
 849                         string oligo = revPrimer[i];
 850
 851                         if(rawSequence.length() < oligo.length()){
 852                                 success = 0;
 853                                 break;
 854                         }
 855
 856                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 857                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 858                                 success = 1;
 859                                 break;
 860                         }
 861                 }
 862
 863                 return success;
 864
 865         }
 866         catch(exception& e) {
 867                 m->errorOut(e, "TrimFlowsCommand", "stripReverse");
 868                 exit(1);
 869         }
 870 }
 871
 872
 873 //***************************************************************************************************************
 874
 875 bool TrimFlowsCommand::compareDNASeq(string oligo, string seq){
 876         try {
 877                 bool success = 1;
 878                 int length = oligo.length();
 879
 880                 for(int i=0;i<length;i++){
 881
 882                         if(oligo[i] != seq[i]){
 883                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 884                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 885                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 886                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 887                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 888                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 889                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 890                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 891                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 892                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 893                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 894                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 895
 896                                 if(success == 0)        {       break;   }
 897                         }
 898                         else{
 899                                 success = 1;
 900                         }
 901                 }
 902
 903                 return success;
 904         }
 905         catch(exception& e) {
 906                 m->errorOut(e, "TrimFlowsCommand", "compareDNASeq");
 907                 exit(1);
 908         }
 909
 910 }
 911
 912 //***************************************************************************************************************
 913
 914 int TrimFlowsCommand::countDiffs(string oligo, string seq){
 915         try {
 916
 917                 int length = oligo.length();
 918                 int countDiffs = 0;
 919
 920                 for(int i=0;i<length;i++){
 921
 922                         if(oligo[i] != seq[i]){
 923                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
 924                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
 925                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
 926                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
 927                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
 928                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
 929                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
 930                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
 931                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
 932                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
 933                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
 934                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
 935                         }
 936
 937                 }
 938
 939                 return countDiffs;
 940         }
 941         catch(exception& e) {
 942                 m->errorOut(e, "TrimFlowsCommand", "countDiffs");
 943                 exit(1);
 944         }
 945
 946 }
 947
 948 /**************************************************************************************************/
 949
 950 vector<unsigned long int> TrimFlowsCommand::getFlowFileBreaks() {
 951
 952         try{
 953
 954                 vector<unsigned long int> filePos;
 955                 filePos.push_back(0);
 956
 957                 FILE * pFile;
 958                 unsigned long int size;
 959
 960                 //get num bytes in file
 961                 pFile = fopen (flowFileName.c_str(),"rb");
 962                 if (pFile==NULL) perror ("Error opening file");
 963                 else{
 964                         fseek (pFile, 0, SEEK_END);
 965                         size=ftell (pFile);
 966                         fclose (pFile);
 967                 }
 968
 969                 //estimate file breaks
 970                 unsigned long int chunkSize = 0;
 971                 chunkSize = size / processors;
 972
 973                 //file too small to divide by processors
 974                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 975
 976                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 977                 for (int i = 0; i < processors; i++) {
 978                         unsigned long int spot = (i+1) * chunkSize;
 979
 980                         ifstream in;
 981                         m->openInputFile(flowFileName, in);
 982                         in.seekg(spot);
 983
 984                         string dummy = m->getline(in);
 985
 986                         //there was not another sequence before the end of the file
 987                         unsigned long int sanityPos = in.tellg();
 988
 989 //                      if (sanityPos == -1) {  break;  }
 990 //                      else {  filePos.push_back(newSpot);  }
 991                         if (sanityPos == -1) {  break;  }
 992                         else {  filePos.push_back(sanityPos);  }
 993
 994                         in.close();
 995                 }
 996
 997                 //save end pos
 998                 filePos.push_back(size);
 999
1000                 //sanity check filePos
1001                 for (int i = 0; i < (filePos.size()-1); i++) {
1002                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
1003                 }
1004
1005                 ifstream in;
1006                 m->openInputFile(flowFileName, in);
1007                 in >> numFlows;
1008                 m->gobble(in);
1009                 unsigned long int spot = in.tellg();
1010                 filePos[0] = spot;
1011                 in.close();
1012
1013                 processors = (filePos.size() - 1);
1014
1015                 return filePos;
1016         }
1017         catch(exception& e) {
1018                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
1019                 exit(1);
1020         }
1021 }
1022
1023 /**************************************************************************************************/
1024
1025 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
1026
1027         try {
1028 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1029                 int process = 1;
1030                 int exitCommand = 1;
1031                 processIDS.clear();
1032
1033                 //loop through and create all the processes you want
1034                 while (process != processors) {
1035                         int pid = fork();
1036
1037                         if (pid > 0) {
1038                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1039                                 process++;
1040                         }else if (pid == 0){
1041
1042                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
1043                                 if(allFiles){
1044                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
1045                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
1046                                                         tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
1047                                                         ofstream temp;
1048                                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
1049                                                         temp.close();
1050
1051                                                 }
1052                                         }
1053                                 }
1054                                 driverCreateTrim(flowFileName,
1055                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
1056                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
1057                                                                  (fastaFileName + toString(getpid()) + ".temp"),
1058                                                                  tempBarcodePrimerComboFileNames, lines[process]);
1059
1060                                 exit(0);
1061                         }else {
1062                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1063                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1064                                 exit(0);
1065                         }
1066                 }
1067
1068                 //parent do my part
1069                 ofstream temp;
1070                 m->openOutputFile(trimFlowFileName, temp);
1071                 temp.close();
1072
1073                 m->openOutputFile(scrapFlowFileName, temp);
1074                 temp.close();
1075
1076                 if(fasta){
1077                         m->openOutputFile(fastaFileName, temp);
1078                         temp.close();
1079                 }
1080
1081                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
1082
1083                 //force parent to wait until all the processes are done
1084                 for (int i=0;i<processIDS.size();i++) {
1085                         int temp = processIDS[i];
1086                         wait(&temp);
1087                 }
1088
1089                 //append files
1090                 m->mothurOutEndLine();
1091                 for(int i=0;i<processIDS.size();i++){
1092
1093                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
1094
1095                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
1096                         remove((trimFlowFileName + toString(processIDS[i]) + ".temp").c_str());
1097 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
1098
1099                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
1100                         remove((scrapFlowFileName + toString(processIDS[i]) + ".temp").c_str());
1101 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
1102
1103                         if(fasta){
1104                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
1105                                 remove((fastaFileName + toString(processIDS[i]) + ".temp").c_str());
1106 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
1107                         }
1108                         if(allFiles){
1109                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
1110                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
1111                                                 m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
1112                                                 remove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp").c_str());
1113                                         }
1114                                 }
1115                         }
1116                 }
1117
1118                 return exitCommand;
1119 #endif
1120         }
1121         catch(exception& e) {
1122                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
1123                 exit(1);
1124         }
1125 }
1126
1127 //***************************************************************************************************************