trimseqscommand.cpp

   1 /*
   2  *  trimseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 6/6/09.
   6  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "trimseqscommand.h"
  11
  12 //***************************************************************************************************************
  13
  14 TrimSeqsCommand::TrimSeqsCommand(string option)  {
  15         try {
  16
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string AlignArray[] =  {"fasta", "flip", "oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
  25                                                                         "qthreshold", "qaverage", "allfiles", "qtrim", "outputdir","inputdir"};
  26
  27                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  28
  29                         OptionParser parser(option);
  30                         map<string,string> parameters = parser.getParameters();
  31
  32                         ValidParameters validParameter;
  33                         map<string,string>::iterator it;
  34
  35                         //check to make sure all parameters are valid for command
  36                         for (it = parameters.begin(); it != parameters.end(); it++) {
  37                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  38                         }
  39
  40                         //if the user changes the input directory command factory will send this info to us in the output parameter
  41                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  42                         if (inputDir == "not found"){   inputDir = "";          }
  43                         else {
  44                                 string path;
  45                                 it = parameters.find("fasta");
  46                                 //user has given a template file
  47                                 if(it != parameters.end()){
  48                                         path = hasPath(it->second);
  49                                         //if the user has not given a path then, add inputdir. else leave path alone.
  50                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  51                                 }
  52
  53                                 it = parameters.find("oligos");
  54                                 //user has given a template file
  55                                 if(it != parameters.end()){
  56                                         path = hasPath(it->second);
  57                                         //if the user has not given a path then, add inputdir. else leave path alone.
  58                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
  59                                 }
  60
  61                                 it = parameters.find("qfile");
  62                                 //user has given a template file
  63                                 if(it != parameters.end()){
  64                                         path = hasPath(it->second);
  65                                         //if the user has not given a path then, add inputdir. else leave path alone.
  66                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
  67                                 }
  68                         }
  69
  70
  71                         //check for required parameters
  72                         fastaFile = validParameter.validFile(parameters, "fasta", true);
  73                         if (fastaFile == "not found") { m->mothurOut("fasta is a required parameter for the screen.seqs command."); m->mothurOutEndLine(); abort = true; }
  74                         else if (fastaFile == "not open") { abort = true; }
  75
  76                         //if the user changes the output directory command factory will send this info to us in the output parameter
  77                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  78                                 outputDir = "";
  79                                 outputDir += hasPath(fastaFile); //if user entered a file with a path then preserve it
  80                         }
  81
  82                         //check for optional parameter and set defaults
  83                         // ...at some point should added some additional type checking...
  84                         string temp;
  85                         temp = validParameter.validFile(parameters, "flip", false);
  86                         if (temp == "not found"){       flip = 0;       }
  87                         else if(isTrue(temp))   {       flip = 1;       }
  88
  89                         temp = validParameter.validFile(parameters, "oligos", true);
  90                         if (temp == "not found"){       oligoFile = "";         }
  91                         else if(temp == "not open"){    abort = true;   }
  92                         else                                    {       oligoFile = temp;               }
  93
  94                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
  95                         convert(temp, maxAmbig);
  96
  97                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "0"; }
  98                         convert(temp, maxHomoP);
  99
 100                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "0"; }
 101                         convert(temp, minLength);
 102
 103                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "0"; }
 104                         convert(temp, maxLength);
 105
 106                         temp = validParameter.validFile(parameters, "qfile", true);
 107                         if (temp == "not found")        {       qFileName = "";         }
 108                         else if(temp == "not open")     {       abort = true;           }
 109                         else                                            {       qFileName = temp;       }
 110
 111                         temp = validParameter.validFile(parameters, "qthreshold", false);       if (temp == "not found") { temp = "0"; }
 112                         convert(temp, qThreshold);
 113
 114                         temp = validParameter.validFile(parameters, "qtrim", false);    if (temp == "not found") { temp = "F"; }
 115                         qtrim = isTrue(temp);
 116
 117                         temp = validParameter.validFile(parameters, "qaverage", false);         if (temp == "not found") { temp = "0"; }
 118                         convert(temp, qAverage);
 119
 120                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found") { temp = "F"; }
 121                         allFiles = isTrue(temp);
 122
 123                         if(allFiles && oligoFile == ""){
 124                                 m->mothurOut("You selected allfiles, but didn't enter an oligos file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 125                         }
 126                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 127                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 128                                 qAverage=0;
 129                                 qThreshold=0;
 130                         }
 131                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 132                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 133                                 abort = true;
 134                         }
 135                 }
 136
 137         }
 138         catch(exception& e) {
 139                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 140                 exit(1);
 141         }
 142 }
 143 //**********************************************************************************************************************
 144
 145 void TrimSeqsCommand::help(){
 146         try {
 147                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 148                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, qtrim and allfiles.\n");
 149                 m->mothurOut("The fasta parameter is required.\n");
 150                 m->mothurOut("The flip parameter .... The default is 0.\n");
 151                 m->mothurOut("The oligos parameter .... The default is "".\n");
 152                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 153                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 154                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 155                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 156                 m->mothurOut("The qfile parameter .....\n");
 157                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 158                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 159                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 160                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 161                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 162                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 163                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 164                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 165                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 166                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 167
 168         }
 169         catch(exception& e) {
 170                 m->errorOut(e, "TrimSeqsCommand", "help");
 171                 exit(1);
 172         }
 173 }
 174
 175
 176 //***************************************************************************************************************
 177
 178 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 179
 180 //***************************************************************************************************************
 181
 182 int TrimSeqsCommand::execute(){
 183         try{
 184
 185                 if (abort == true) { return 0; }
 186
 187                 numFPrimers = 0;  //this needs to be initialized
 188                 numRPrimers = 0;
 189
 190                 ifstream inFASTA;
 191                 openInputFile(fastaFile, inFASTA);
 192
 193                 ofstream outFASTA;
 194                 string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
 195                 openOutputFile(trimSeqFile, outFASTA);
 196                 outputNames.push_back(trimSeqFile);
 197
 198                 ofstream outGroups;
 199                 vector<ofstream*> fastaFileNames;
 200                 if(oligoFile != ""){
 201                         string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
 202                         openOutputFile(groupFile, outGroups);
 203                         outputNames.push_back(groupFile);
 204                         getOligos(fastaFileNames);
 205                 }
 206
 207                 ofstream scrapFASTA;
 208                 string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
 209                 openOutputFile(scrapSeqFile, scrapFASTA);
 210                 outputNames.push_back(scrapSeqFile);
 211
 212                 ifstream qFile;
 213                 if(qFileName != "")     {       openInputFile(qFileName, qFile);        }
 214
 215                 bool success;
 216
 217                 while(!inFASTA.eof()){
 218
 219                         if (m->control_pressed) {
 220                                 inFASTA.close();
 221                                 outFASTA.close();
 222                                 scrapFASTA.close();
 223                                 outGroups.close();
 224                                 if(qFileName != "")     {       qFile.close();  }
 225                                 for(int i=0;i<fastaFileNames.size();i++){
 226                                         fastaFileNames[i]->close();
 227                                         delete fastaFileNames[i];
 228                                 }
 229                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 230                                 return 0;
 231                         }
 232
 233
 234                         Sequence currSeq(inFASTA);
 235
 236                         string origSeq = currSeq.getUnaligned();
 237                         if (origSeq != "") {
 238                                 int group;
 239                                 string trashCode = "";
 240
 241                                 if(qFileName != ""){
 242                                         if(qThreshold != 0)             {       success = stripQualThreshold(currSeq, qFile);   }
 243                                         else if(qAverage != 0)  {       success = cullQualAverage(currSeq, qFile);              }
 244                                         if ((!qtrim) && (origSeq.length() != currSeq.getUnaligned().length())) {
 245                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 246                                         }
 247                                         if(!success)                    {       trashCode += 'q';                                                               }
 248                                 }
 249
 250                                 if(barcodes.size() != 0){
 251                                         success = stripBarcode(currSeq, group);
 252                                         if(!success){   trashCode += 'b';       }
 253                                 }
 254
 255                                 if(numFPrimers != 0){
 256                                         success = stripForward(currSeq);
 257                                         if(!success){   trashCode += 'f';       }
 258                                 }
 259
 260                                 if(numRPrimers != 0){
 261                                         success = stripReverse(currSeq);
 262                                         if(!success){   trashCode += 'r';       }
 263                                 }
 264
 265                                 if(minLength > 0 || maxLength > 0){
 266                                         success = cullLength(currSeq);
 267                                         if(!success){   trashCode += 'l'; }
 268                                 }
 269                                 if(maxHomoP > 0){
 270                                         success = cullHomoP(currSeq);
 271                                         if(!success){   trashCode += 'h';       }
 272                                 }
 273                                 if(maxAmbig != -1){
 274                                         success = cullAmbigs(currSeq);
 275                                         if(!success){   trashCode += 'n';       }
 276                                 }
 277
 278                                 if(flip){       currSeq.reverseComplement();    }               // should go last
 279
 280                                 if(trashCode.length() == 0){
 281                                         currSeq.setAligned(currSeq.getUnaligned());  //this is because of a modification we made to the sequence class to fix a bug.  all seqs have an aligned version, which is the version that gets printed.
 282                                         currSeq.printSequence(outFASTA);
 283                                         if(barcodes.size() != 0){
 284                                                 outGroups << currSeq.getName() << '\t' << groupVector[group] << endl;
 285
 286                                                 if(allFiles){
 287                                                         currSeq.printSequence(*fastaFileNames[group]);
 288                                                 }
 289                                         }
 290                                 }
 291                                 else{
 292                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 293                                         currSeq.setUnaligned(origSeq);
 294                                         currSeq.printSequence(scrapFASTA);
 295                                 }
 296                         }
 297                         gobble(inFASTA);
 298                 }
 299                 inFASTA.close();
 300                 outFASTA.close();
 301                 scrapFASTA.close();
 302                 outGroups.close();
 303                 if(qFileName != "")     {       qFile.close();  }
 304
 305                 for(int i=0;i<fastaFileNames.size();i++){
 306                         fastaFileNames[i]->close();
 307                         delete fastaFileNames[i];
 308                 }
 309
 310                 for(int i=0;i<fastaFileNames.size();i++){
 311                         string seqName;
 312                         openInputFile(getRootName(fastaFile) + groupVector[i] + ".fasta", inFASTA);
 313                         ofstream outGroups;
 314                         openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
 315                         outputNames.push_back(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups");
 316
 317                         while(!inFASTA.eof()){
 318                                 if(inFASTA.get() == '>'){
 319                                         inFASTA >> seqName;
 320                                         outGroups << seqName << '\t' << groupVector[i] << endl;
 321                                 }
 322                                 while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 323                         }
 324                         outGroups.close();
 325                         inFASTA.close();
 326                 }
 327
 328                 if (m->control_pressed) {
 329                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 330                         return 0;
 331                 }
 332
 333                 m->mothurOutEndLine();
 334                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 335                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 336                 m->mothurOutEndLine();
 337
 338                 return 0;
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "TrimSeqsCommand", "execute");
 342                 exit(1);
 343         }
 344 }
 345
 346 //***************************************************************************************************************
 347
 348 void TrimSeqsCommand::getOligos(vector<ofstream*>& outFASTAVec){
 349         try {
 350                 ifstream inOligos;
 351                 openInputFile(oligoFile, inOligos);
 352
 353                 ofstream test;
 354
 355                 string type, oligo, group;
 356                 int index=0;
 357
 358                 while(!inOligos.eof()){
 359                         inOligos >> type;
 360
 361                         if(type[0] == '#'){
 362                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 363                         }
 364                         else{
 365                                 inOligos >> oligo;
 366
 367                                 for(int i=0;i<oligo.length();i++){
 368                                         oligo[i] = toupper(oligo[i]);
 369                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 370                                 }
 371
 372                                 if(type == "forward"){
 373                                         forPrimer.push_back(oligo);
 374                                 }
 375                                 else if(type == "reverse"){
 376                                         Sequence oligoRC("reverse", oligo);
 377                                         oligoRC.reverseComplement();
 378                                         revPrimer.push_back(oligoRC.getUnaligned());
 379                                 }
 380                                 else if(type == "barcode"){
 381                                         inOligos >> group;
 382                                         barcodes[oligo]=index++;
 383                                         groupVector.push_back(group);
 384
 385                                         if(allFiles){
 386                                                 outFASTAVec.push_back(new ofstream((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta").c_str(), ios::ate));
 387                                                 outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta"));
 388                                         }
 389                                 }
 390                         }
 391                 }
 392
 393                 inOligos.close();
 394
 395                 numFPrimers = forPrimer.size();
 396                 numRPrimers = revPrimer.size();
 397
 398         }
 399         catch(exception& e) {
 400                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 401                 exit(1);
 402         }
 403
 404 }
 405
 406 //***************************************************************************************************************
 407
 408 bool TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
 409         try {
 410                 string rawSequence = seq.getUnaligned();
 411                 bool success = 0;       //guilty until proven innocent
 412
 413                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 414                         string oligo = it->first;
 415                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 416                                 success = 0;
 417                                 break;
 418                         }
 419
 420                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 421                                 group = it->second;
 422                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 423                                 success = 1;
 424                                 break;
 425                         }
 426                 }
 427                 return success;
 428
 429         }
 430         catch(exception& e) {
 431                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 432                 exit(1);
 433         }
 434
 435 }
 436
 437 //***************************************************************************************************************
 438
 439 bool TrimSeqsCommand::stripForward(Sequence& seq){
 440         try {
 441                 string rawSequence = seq.getUnaligned();
 442                 bool success = 0;       //guilty until proven innocent
 443
 444                 for(int i=0;i<numFPrimers;i++){
 445                         string oligo = forPrimer[i];
 446
 447                         if(rawSequence.length() < oligo.length()){
 448                                 success = 0;
 449                                 break;
 450                         }
 451
 452                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 453                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 454                                 success = 1;
 455                                 break;
 456                         }
 457                 }
 458
 459                 return success;
 460
 461         }
 462         catch(exception& e) {
 463                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
 464                 exit(1);
 465         }
 466 }
 467
 468 //***************************************************************************************************************
 469
 470 bool TrimSeqsCommand::stripReverse(Sequence& seq){
 471         try {
 472                 string rawSequence = seq.getUnaligned();
 473                 bool success = 0;       //guilty until proven innocent
 474
 475                 for(int i=0;i<numRPrimers;i++){
 476                         string oligo = revPrimer[i];
 477
 478                         if(rawSequence.length() < oligo.length()){
 479                                 success = 0;
 480                                 break;
 481                         }
 482
 483                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 484                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 485                                 success = 1;
 486                                 break;
 487                         }
 488                 }
 489                 return success;
 490
 491         }
 492         catch(exception& e) {
 493                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
 494                 exit(1);
 495         }
 496 }
 497
 498 //***************************************************************************************************************
 499
 500 bool TrimSeqsCommand::cullLength(Sequence& seq){
 501         try {
 502
 503                 int length = seq.getNumBases();
 504                 bool success = 0;       //guilty until proven innocent
 505
 506                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
 507                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
 508                 else                                                                                            {       success = 0;    }
 509
 510                 return success;
 511
 512         }
 513         catch(exception& e) {
 514                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
 515                 exit(1);
 516         }
 517
 518 }
 519
 520 //***************************************************************************************************************
 521
 522 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
 523         try {
 524                 int longHomoP = seq.getLongHomoPolymer();
 525                 bool success = 0;       //guilty until proven innocent
 526
 527                 if(longHomoP <= maxHomoP){      success = 1;    }
 528                 else                                    {       success = 0;    }
 529
 530                 return success;
 531         }
 532         catch(exception& e) {
 533                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
 534                 exit(1);
 535         }
 536
 537 }
 538
 539 //***************************************************************************************************************
 540
 541 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
 542         try {
 543                 int numNs = seq.getAmbigBases();
 544                 bool success = 0;       //guilty until proven innocent
 545
 546                 if(numNs <= maxAmbig)   {       success = 1;    }
 547                 else                                    {       success = 0;    }
 548
 549                 return success;
 550         }
 551         catch(exception& e) {
 552                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
 553                 exit(1);
 554         }
 555
 556 }
 557
 558 //***************************************************************************************************************
 559
 560 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
 561         try {
 562                 bool success = 1;
 563                 int length = oligo.length();
 564
 565                 for(int i=0;i<length;i++){
 566
 567                         if(oligo[i] != seq[i]){
 568                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 569                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 570                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 571                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 572                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 573                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 574                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 575                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 576                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 577                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 578                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 579                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 580
 581                                 if(success == 0)        {       break;  }
 582                         }
 583                         else{
 584                                 success = 1;
 585                         }
 586                 }
 587
 588                 return success;
 589         }
 590         catch(exception& e) {
 591                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
 592                 exit(1);
 593         }
 594
 595 }
 596
 597 //***************************************************************************************************************
 598
 599 bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
 600         try {
 601                 string rawSequence = seq.getUnaligned();
 602                 int seqLength = rawSequence.length();
 603                 string name;
 604
 605                 qFile >> name;
 606                 if (name.length() != 0) {  if(name.substr(1) != seq.getName())  {       m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); }  }
 607                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 608
 609                 int score;
 610                 int end = seqLength;
 611
 612                 for(int i=0;i<seqLength;i++){
 613                         qFile >> score;
 614
 615                         if(score <= qThreshold){
 616                                 end = i;
 617                                 break;
 618                         }
 619                 }
 620                 for(int i=end+1;i<seqLength;i++){
 621                         qFile >> score;
 622                 }
 623
 624                 seq.setUnaligned(rawSequence.substr(0,end));
 625
 626                 return 1;
 627         }
 628         catch(exception& e) {
 629                 m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
 630                 exit(1);
 631         }
 632 }
 633
 634 //***************************************************************************************************************
 635
 636 bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
 637         try {
 638                 string rawSequence = seq.getUnaligned();
 639                 int seqLength = seq.getNumBases();
 640                 bool success = 0;       //guilty until proven innocent
 641                 string name;
 642
 643                 qFile >> name;
 644                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
 645
 646                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 647
 648                 float score;
 649                 float average = 0;
 650
 651                 for(int i=0;i<seqLength;i++){
 652                         qFile >> score;
 653                         average += score;
 654                 }
 655                 average /= seqLength;
 656
 657                 if(average >= qAverage) {       success = 1;    }
 658                 else                                    {       success = 0;    }
 659
 660                 return success;
 661         }
 662         catch(exception& e) {
 663                 m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
 664                 exit(1);
 665         }
 666 }
 667
 668 //***************************************************************************************************************