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
 326                 m->mothurOutEndLine();
 327                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 328                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 329                 m->mothurOutEndLine();
 330
 331                 return 0;
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "TrimSeqsCommand", "execute");
 335                 exit(1);
 336         }
 337 }
 338
 339 //***************************************************************************************************************
 340
 341 int TrimFlowsCommand::driverCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames, linePair* line){
 342
 343         try {
 344
 345                 ofstream trimFlowFile;
 346                 m->openOutputFile(trimFlowFileName, trimFlowFile);
 347                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 348
 349                 ofstream scrapFlowFile;
 350                 m->openOutputFile(scrapFlowFileName, scrapFlowFile);
 351                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 352
 353                 if(line->start == 4){
 354                         scrapFlowFile << maxFlows << endl;
 355                         trimFlowFile << maxFlows << endl;
 356                         if(allFiles){
 357                                 for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 358                                         for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 359                                                 //                              barcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 360                                                 ofstream temp;
 361                                                 m->openOutputFile(barcodePrimerComboFileNames[i][j], temp);
 362                                                 temp << maxFlows << endl;
 363                                                 temp.close();
 364                                         }
 365                                 }
 366                         }
 367                 }
 368
 369                 FlowData flowData(numFlows, signal, noise, maxHomoP);
 370
 371                 ofstream fastaFile;
 372                 if(fasta){      m->openOutputFile(fastaFileName, fastaFile);    }
 373
 374                 ifstream flowFile;
 375                 m->openInputFile(flowFileName, flowFile);
 376
 377                 flowFile.seekg(line->start);
 378
 379                 int count = 0;
 380                 bool moreSeqs = 1;
 381
 382                 while(moreSeqs) {
 383
 384                         int success = 1;
 385                         int currentSeqDiffs = 0;
 386                         string trashCode = "";
 387
 388                         flowData.getNext(flowFile);
 389                         flowData.capFlows(maxFlows);
 390
 391                         Sequence currSeq = flowData.getSequence();
 392                         if(!flowData.hasMinFlows(minFlows)){    //screen to see if sequence is of a minimum number of flows
 393                                 success = 0;
 394                                 trashCode += 'l';
 395                         }
 396
 397                         if(minLength > 0 || maxLength > 0){     //screen to see if sequence is above and below a specific number of bases
 398                                 int seqLength = currSeq.getNumBases();
 399                                 if(seqLength < minLength || seqLength > maxLength){
 400                                         success = 0;
 401                                         trashCode += 'l';
 402                                 }
 403                         }
 404
 405                         int primerIndex = 0;
 406                         int barcodeIndex = 0;
 407
 408                         if(barcodes.size() != 0){
 409                                 success = stripBarcode(currSeq, barcodeIndex);
 410                                 if(success > bdiffs)            {       trashCode += 'b';       }
 411                                 else{ currentSeqDiffs += success;  }
 412                         }
 413
 414                         if(numFPrimers != 0){
 415                                 success = stripForward(currSeq, primerIndex);
 416                                 if(success > pdiffs)            {       trashCode += 'f';       }
 417                                 else{ currentSeqDiffs += success;  }
 418                         }
 419
 420                         if (currentSeqDiffs > tdiffs)   {       trashCode += 't';   }
 421
 422                         if(numRPrimers != 0){
 423                                 success = stripReverse(currSeq);
 424                                 if(!success)                            {       trashCode += 'r';       }
 425                         }
 426
 427                         if(trashCode.length() == 0){
 428
 429                                 flowData.printFlows(trimFlowFile);
 430
 431                                 if(fasta)       {       currSeq.printSequence(fastaFile);       }
 432
 433                                 if(allFiles){
 434 //                                      string fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + barcodePrimerCombos[barcodeIndex][primerIndex] + ".flow";
 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                         }
 444                         else{
 445                                 flowData.printFlows(scrapFlowFile, trashCode);
 446                         }
 447
 448                         count++;
 449
 450                         //report progress
 451                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 452
 453 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 454                         unsigned long int pos = flowFile.tellg();
 455
 456                         if ((pos == -1) || (pos >= line->end)) { break; }
 457 #else
 458                         if (flowFile.eof()) { break; }
 459 #endif
 460
 461                 }
 462                 //report progress
 463                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 464
 465                 trimFlowFile.close();
 466                 scrapFlowFile.close();
 467                 flowFile.close();
 468                 if(fasta){      fastaFile.close();      }
 469
 470                 return count;
 471         }
 472         catch(exception& e) {
 473                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 474                 exit(1);
 475         }
 476 }
 477
 478 //***************************************************************************************************************
 479
 480 void TrimFlowsCommand::getOligos(vector<vector<string> >& outFlowFileNames){
 481         try {
 482                 ifstream oligosFile;
 483                 m->openInputFile(oligoFileName, oligosFile);
 484
 485                 string type, oligo, group;
 486
 487                 int indexPrimer = 0;
 488                 int indexBarcode = 0;
 489
 490                 while(!oligosFile.eof()){
 491
 492                         oligosFile >> type; m->gobble(oligosFile);      //get the first column value of the row - is it a comment or a feature we are interested in?
 493
 494                         if(type[0] == '#'){     //igore the line because there's a comment
 495                                 while (!oligosFile.eof())       {       char c = oligosFile.get(); if (c == 10 || c == 13){     break;  }       } // get rest of line if there's any crap there
 496                         }
 497                         else{                           //there's a feature we're interested in
 498
 499                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }                                  //make type case insensitive
 500
 501                                 oligosFile >> oligo;    //get the DNA sequence for the feature
 502
 503                                 for(int i=0;i<oligo.length();i++){      //make type case insensitive and change any U's to T's
 504                                         oligo[i] = toupper(oligo[i]);
 505                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 506                                 }
 507
 508                                 if(type == "FORWARD"){  //if the feature is a forward primer...
 509                                         group = "";
 510
 511                                         while (!oligosFile.eof())       {       // get rest of line in case there is a primer name = will have the name of the primer
 512                                                 char c = oligosFile.get();
 513                                                 if (c == 10 || c == 13){        break;  }
 514                                                 else if (c == 32 || c == 9){;} //space or tab
 515                                                 else {  group += c;  }
 516                                         }
 517
 518                                         //have we seen this primer already?
 519                                         map<string, int>::iterator itPrimer = primers.find(oligo);
 520                                         if (itPrimer != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 521
 522                                         primers[oligo]=indexPrimer; indexPrimer++;
 523                                         primerNameVector.push_back(group);
 524
 525                                 }
 526                                 else if(type == "REVERSE"){
 527                                         Sequence oligoRC("reverse", oligo);
 528                                         oligoRC.reverseComplement();
 529                                         revPrimer.push_back(oligoRC.getUnaligned());
 530                                 }
 531                                 else if(type == "BARCODE"){
 532                                         oligosFile >> group;
 533
 534                                         //check for repeat barcodes
 535                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 536                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 537
 538                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 539                                         barcodeNameVector.push_back(group);
 540                                 }
 541                                 else{
 542                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 543                                 }
 544                         }
 545
 546                         m->gobble(oligosFile);
 547                 }
 548                 oligosFile.close();
 549
 550                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 551
 552                 //add in potential combos
 553                 if(barcodeNameVector.size() == 0){
 554                         barcodes[""] = 0;
 555                         barcodeNameVector.push_back("");
 556                 }
 557
 558                 if(primerNameVector.size() == 0){
 559                         primers[""] = 0;
 560                         primerNameVector.push_back("");
 561                 }
 562
 563
 564                 outFlowFileNames.resize(barcodeNameVector.size());
 565                 for(int i=0;i<outFlowFileNames.size();i++){
 566                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 567                 }
 568
 569                 if(allFiles){
 570
 571                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 572                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 573
 574                                         string primerName = primerNameVector[itPrimer->second];
 575                                         string barcodeName = barcodeNameVector[itBar->second];
 576
 577                                         string comboGroupName = "";
 578                                         string fileName = "";
 579
 580                                         if(primerName == ""){
 581                                                 comboGroupName = barcodeNameVector[itBar->second];
 582                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 583                                         }
 584                                         else{
 585                                                 if(barcodeName == ""){
 586                                                         comboGroupName = primerNameVector[itPrimer->second];
 587                                                 }
 588                                                 else{
 589                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 590                                                 }
 591                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 592                                         }
 593
 594                                         outputNames.push_back(fileName);
 595                                         outputTypes["flow"].push_back(fileName);
 596                                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 597
 598                                         ofstream temp;
 599                                         m->openOutputFile(fileName, temp);
 600                                         temp.close();
 601                                 }
 602                         }
 603                 }
 604
 605                 numFPrimers = primers.size();
 606                 numRPrimers = revPrimer.size();
 607
 608         }
 609         catch(exception& e) {
 610                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 611                 exit(1);
 612         }
 613 }
 614
 615 //***************************************************************************************************************
 616
 617 int TrimFlowsCommand::stripBarcode(Sequence& seq, int& group){
 618         try {
 619
 620                 string rawSequence = seq.getUnaligned();
 621                 int success = bdiffs + 1;       //guilty until proven innocent
 622
 623                 //can you find the barcode
 624                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 625                         string oligo = it->first;
 626                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 627                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 628                                 break;
 629                         }
 630
 631                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 632                                 group = it->second;
 633                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 634
 635                                 success = 0;
 636                                 break;
 637                         }
 638                 }
 639
 640                 //if you found the barcode or if you don't want to allow for diffs
 641                 if ((bdiffs == 0) || (success == 0)) { return success;  }
 642
 643                 else { //try aligning and see if you can find it
 644
 645                         int maxLength = 0;
 646
 647                         Alignment* alignment;
 648                         if (barcodes.size() > 0) {
 649                                 map<string,int>::iterator it=barcodes.begin();
 650
 651                                 for(it;it!=barcodes.end();it++){
 652                                         if(it->first.length() > maxLength){
 653                                                 maxLength = it->first.length();
 654                                         }
 655                                 }
 656                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
 657
 658                         }else{ alignment = NULL; }
 659
 660                         //can you find the barcode
 661                         int minDiff = 1e6;
 662                         int minCount = 1;
 663                         int minGroup = -1;
 664                         int minPos = 0;
 665
 666                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 667                                 string oligo = it->first;
 668                                 //                              int length = oligo.length();
 669
 670                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
 671                                         success = bdiffs + 10;
 672                                         break;
 673                                 }
 674
 675                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 676                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
 677                                 oligo = alignment->getSeqAAln();
 678                                 string temp = alignment->getSeqBAln();
 679
 680                                 int alnLength = oligo.length();
 681
 682                                 for(int i=oligo.length()-1;i>=0;i--){
 683                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 684                                 }
 685                                 oligo = oligo.substr(0,alnLength);
 686                                 temp = temp.substr(0,alnLength);
 687
 688                                 int numDiff = countDiffs(oligo, temp);
 689
 690                                 if(numDiff < minDiff){
 691                                         minDiff = numDiff;
 692                                         minCount = 1;
 693                                         minGroup = it->second;
 694                                         minPos = 0;
 695                                         for(int i=0;i<alnLength;i++){
 696                                                 if(temp[i] != '-'){
 697                                                         minPos++;
 698                                                 }
 699                                         }
 700                                 }
 701                                 else if(numDiff == minDiff){
 702                                         minCount++;
 703                                 }
 704
 705                         }
 706
 707                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
 708                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
 709                         else{                                                                                                   //use the best match
 710                                 group = minGroup;
 711                                 seq.setUnaligned(rawSequence.substr(minPos));
 712                                 success = minDiff;
 713                         }
 714
 715                         if (alignment != NULL) {  delete alignment;  }
 716
 717                 }
 718
 719                 return success;
 720
 721         }
 722         catch(exception& e) {
 723                 m->errorOut(e, "TrimFlowsCommand", "stripBarcode");
 724                 exit(1);
 725         }
 726
 727 }
 728
 729 //***************************************************************************************************************
 730
 731 int TrimFlowsCommand::stripForward(Sequence& seq, int& group){
 732         try {
 733
 734                 string rawSequence = seq.getUnaligned();
 735                 int success = pdiffs + 1;       //guilty until proven innocent
 736
 737                 //can you find the primer
 738                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 739                         string oligo = it->first;
 740                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
 741                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 742                                 break;
 743                         }
 744
 745                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 746                                 group = it->second;
 747                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 748                                 success = 0;
 749                                 break;
 750                         }
 751                 }
 752
 753                 //if you found the barcode or if you don't want to allow for diffs
 754                 if ((pdiffs == 0) || (success == 0)) {  return success;  }
 755
 756                 else { //try aligning and see if you can find it
 757
 758                         int maxLength = 0;
 759
 760                         Alignment* alignment;
 761                         if (primers.size() > 0) {
 762                                 map<string,int>::iterator it=primers.begin();
 763
 764                                 for(it;it!=primers.end();it++){
 765                                         if(it->first.length() > maxLength){
 766                                                 maxLength = it->first.length();
 767                                         }
 768                                 }
 769                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
 770
 771                         }else{ alignment = NULL; }
 772
 773                         //can you find the barcode
 774                         int minDiff = 1e6;
 775                         int minCount = 1;
 776                         int minGroup = -1;
 777                         int minPos = 0;
 778
 779                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 780                                 string oligo = it->first;
 781                                 //                              int length = oligo.length();
 782
 783                                 if(rawSequence.length() < maxLength){
 784                                         success = pdiffs + 100;
 785                                         break;
 786                                 }
 787
 788                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 789                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
 790                                 oligo = alignment->getSeqAAln();
 791                                 string temp = alignment->getSeqBAln();
 792
 793                                 int alnLength = oligo.length();
 794
 795                                 for(int i=oligo.length()-1;i>=0;i--){
 796                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 797                                 }
 798                                 oligo = oligo.substr(0,alnLength);
 799                                 temp = temp.substr(0,alnLength);
 800
 801                                 int numDiff = countDiffs(oligo, temp);
 802
 803                                 if(numDiff < minDiff){
 804                                         minDiff = numDiff;
 805                                         minCount = 1;
 806                                         minGroup = it->second;
 807                                         minPos = 0;
 808                                         for(int i=0;i<alnLength;i++){
 809                                                 if(temp[i] != '-'){
 810                                                         minPos++;
 811                                                 }
 812                                         }
 813                                 }
 814                                 else if(numDiff == minDiff){
 815                                         minCount++;
 816                                 }
 817
 818                         }
 819
 820                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
 821                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
 822                         else{                                                                                                   //use the best match
 823                                 group = minGroup;
 824                                 seq.setUnaligned(rawSequence.substr(minPos));
 825                                 success = minDiff;
 826                         }
 827
 828                         if (alignment != NULL) {  delete alignment;  }
 829
 830                 }
 831
 832                 return success;
 833
 834         }
 835         catch(exception& e) {
 836                 m->errorOut(e, "TrimFlowsCommand", "stripForward");
 837                 exit(1);
 838         }
 839 }
 840
 841 //***************************************************************************************************************
 842
 843 bool TrimFlowsCommand::stripReverse(Sequence& seq){
 844         try {
 845
 846                 string rawSequence = seq.getUnaligned();
 847                 bool success = 0;       //guilty until proven innocent
 848
 849                 for(int i=0;i<numRPrimers;i++){
 850                         string oligo = revPrimer[i];
 851
 852                         if(rawSequence.length() < oligo.length()){
 853                                 success = 0;
 854                                 break;
 855                         }
 856
 857                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 858                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 859                                 success = 1;
 860                                 break;
 861                         }
 862                 }
 863
 864                 return success;
 865
 866         }
 867         catch(exception& e) {
 868                 m->errorOut(e, "TrimFlowsCommand", "stripReverse");
 869                 exit(1);
 870         }
 871 }
 872
 873
 874 //***************************************************************************************************************
 875
 876 bool TrimFlowsCommand::compareDNASeq(string oligo, string seq){
 877         try {
 878                 bool success = 1;
 879                 int length = oligo.length();
 880
 881                 for(int i=0;i<length;i++){
 882
 883                         if(oligo[i] != seq[i]){
 884                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 885                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 886                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 887                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 888                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 889                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 890                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 891                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 892                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 893                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 894                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 895                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 896
 897                                 if(success == 0)        {       break;   }
 898                         }
 899                         else{
 900                                 success = 1;
 901                         }
 902                 }
 903
 904                 return success;
 905         }
 906         catch(exception& e) {
 907                 m->errorOut(e, "TrimFlowsCommand", "compareDNASeq");
 908                 exit(1);
 909         }
 910
 911 }
 912
 913 //***************************************************************************************************************
 914
 915 int TrimFlowsCommand::countDiffs(string oligo, string seq){
 916         try {
 917
 918                 int length = oligo.length();
 919                 int countDiffs = 0;
 920
 921                 for(int i=0;i<length;i++){
 922
 923                         if(oligo[i] != seq[i]){
 924                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
 925                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
 926                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
 927                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
 928                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
 929                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
 930                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
 931                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
 932                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
 933                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
 934                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
 935                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
 936                         }
 937
 938                 }
 939
 940                 return countDiffs;
 941         }
 942         catch(exception& e) {
 943                 m->errorOut(e, "TrimFlowsCommand", "countDiffs");
 944                 exit(1);
 945         }
 946
 947 }
 948
 949 /**************************************************************************************************/
 950
 951 vector<unsigned long int> TrimFlowsCommand::getFlowFileBreaks() {
 952
 953         try{
 954
 955                 vector<unsigned long int> filePos;
 956                 filePos.push_back(0);
 957
 958                 FILE * pFile;
 959                 unsigned long int size;
 960
 961                 //get num bytes in file
 962                 pFile = fopen (flowFileName.c_str(),"rb");
 963                 if (pFile==NULL) perror ("Error opening file");
 964                 else{
 965                         fseek (pFile, 0, SEEK_END);
 966                         size=ftell (pFile);
 967                         fclose (pFile);
 968                 }
 969
 970                 //estimate file breaks
 971                 unsigned long int chunkSize = 0;
 972                 chunkSize = size / processors;
 973
 974                 //file too small to divide by processors
 975                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 976
 977                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 978                 for (int i = 0; i < processors; i++) {
 979                         unsigned long int spot = (i+1) * chunkSize;
 980
 981                         ifstream in;
 982                         m->openInputFile(flowFileName, in);
 983                         in.seekg(spot);
 984
 985                         string dummy = m->getline(in);
 986
 987                         //there was not another sequence before the end of the file
 988                         unsigned long int sanityPos = in.tellg();
 989
 990 //                      if (sanityPos == -1) {  break;  }
 991 //                      else {  filePos.push_back(newSpot);  }
 992                         if (sanityPos == -1) {  break;  }
 993                         else {  filePos.push_back(sanityPos);  }
 994
 995                         in.close();
 996                 }
 997
 998                 //save end pos
 999                 filePos.push_back(size);
1000
1001                 //sanity check filePos
1002                 for (int i = 0; i < (filePos.size()-1); i++) {
1003                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
1004                 }
1005
1006                 ifstream in;
1007                 m->openInputFile(flowFileName, in);
1008                 in >> numFlows;
1009                 m->gobble(in);
1010                 unsigned long int spot = in.tellg();
1011                 filePos[0] = spot;
1012                 in.close();
1013
1014                 processors = (filePos.size() - 1);
1015
1016                 return filePos;
1017         }
1018         catch(exception& e) {
1019                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
1020                 exit(1);
1021         }
1022 }
1023
1024 /**************************************************************************************************/
1025
1026 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
1027
1028         try {
1029 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1030                 int process = 1;
1031                 int exitCommand = 1;
1032                 processIDS.clear();
1033
1034                 //loop through and create all the processes you want
1035                 while (process != processors) {
1036                         int pid = fork();
1037
1038                         if (pid > 0) {
1039                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1040                                 process++;
1041                         }else if (pid == 0){
1042
1043                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
1044                                 if(allFiles){
1045                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
1046                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
1047                                                         tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
1048                                                         ofstream temp;
1049                                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
1050                                                         temp.close();
1051
1052                                                 }
1053                                         }
1054                                 }
1055                                 driverCreateTrim(flowFileName,
1056                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
1057                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
1058                                                                  (fastaFileName + toString(getpid()) + ".temp"),
1059                                                                  tempBarcodePrimerComboFileNames, lines[process]);
1060
1061                                 exit(0);
1062                         }else {
1063                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1064                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1065                                 exit(0);
1066                         }
1067                 }
1068
1069                 //parent do my part
1070                 ofstream temp;
1071                 m->openOutputFile(trimFlowFileName, temp);
1072                 temp.close();
1073
1074                 m->openOutputFile(scrapFlowFileName, temp);
1075                 temp.close();
1076
1077                 if(fasta){
1078                         m->openOutputFile(fastaFileName, temp);
1079                         temp.close();
1080                 }
1081
1082                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
1083
1084                 //force parent to wait until all the processes are done
1085                 for (int i=0;i<processIDS.size();i++) {
1086                         int temp = processIDS[i];
1087                         wait(&temp);
1088                 }
1089
1090                 //append files
1091                 m->mothurOutEndLine();
1092                 for(int i=0;i<processIDS.size();i++){
1093
1094                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
1095
1096                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
1097                         remove((trimFlowFileName + toString(processIDS[i]) + ".temp").c_str());
1098 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
1099
1100                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
1101                         remove((scrapFlowFileName + toString(processIDS[i]) + ".temp").c_str());
1102 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
1103
1104                         if(fasta){
1105                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
1106                                 remove((fastaFileName + toString(processIDS[i]) + ".temp").c_str());
1107 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
1108                         }
1109                         if(allFiles){
1110                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
1111                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
1112                                                 m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
1113                                                 remove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp").c_str());
1114                                         }
1115                                 }
1116                         }
1117                 }
1118
1119                 return exitCommand;
1120 #endif
1121         }
1122         catch(exception& e) {
1123                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
1124                 exit(1);
1125         }
1126 }
1127
1128 //***************************************************************************************************************