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                 vector<string> outputNames;
 188
 189                 numFPrimers = 0;  //this needs to be initialized
 190                 numRPrimers = 0;
 191
 192                 ifstream inFASTA;
 193                 openInputFile(fastaFile, inFASTA);
 194
 195                 ofstream outFASTA;
 196                 string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
 197                 openOutputFile(trimSeqFile, outFASTA);
 198                 outputNames.push_back(trimSeqFile);
 199
 200                 ofstream outGroups;
 201                 vector<ofstream*> fastaFileNames;
 202                 if(oligoFile != ""){
 203                         string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
 204                         openOutputFile(groupFile, outGroups);
 205                         outputNames.push_back(groupFile);
 206                         getOligos(fastaFileNames);
 207                 }
 208
 209                 ofstream scrapFASTA;
 210                 string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
 211                 openOutputFile(scrapSeqFile, scrapFASTA);
 212                 outputNames.push_back(scrapSeqFile);
 213
 214                 ifstream qFile;
 215                 if(qFileName != "")     {       openInputFile(qFileName, qFile);        }
 216
 217                 bool success;
 218
 219                 while(!inFASTA.eof()){
 220                         Sequence currSeq(inFASTA);
 221
 222                         string origSeq = currSeq.getUnaligned();
 223                         if (origSeq != "") {
 224                                 int group;
 225                                 string trashCode = "";
 226
 227                                 if(qFileName != ""){
 228                                         if(qThreshold != 0)             {       success = stripQualThreshold(currSeq, qFile);   }
 229                                         else if(qAverage != 0)  {       success = cullQualAverage(currSeq, qFile);              }
 230                                         if ((!qtrim) && (origSeq.length() != currSeq.getUnaligned().length())) {
 231                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 232                                         }
 233                                         if(!success)                    {       trashCode += 'q';                                                               }
 234                                 }
 235
 236                                 if(barcodes.size() != 0){
 237                                         success = stripBarcode(currSeq, group);
 238                                         if(!success){   trashCode += 'b';       }
 239                                 }
 240
 241                                 if(numFPrimers != 0){
 242                                         success = stripForward(currSeq);
 243                                         if(!success){   trashCode += 'f';       }
 244                                 }
 245
 246                                 if(numRPrimers != 0){
 247                                         success = stripReverse(currSeq);
 248                                         if(!success){   trashCode += 'r';       }
 249                                 }
 250
 251                                 if(minLength > 0 || maxLength > 0){
 252                                         success = cullLength(currSeq);
 253                                         if(!success){   trashCode += 'l'; }
 254                                 }
 255                                 if(maxHomoP > 0){
 256                                         success = cullHomoP(currSeq);
 257                                         if(!success){   trashCode += 'h';       }
 258                                 }
 259                                 if(maxAmbig != -1){
 260                                         success = cullAmbigs(currSeq);
 261                                         if(!success){   trashCode += 'n';       }
 262                                 }
 263
 264                                 if(flip){       currSeq.reverseComplement();    }               // should go last
 265
 266                                 if(trashCode.length() == 0){
 267                                         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.
 268                                         currSeq.printSequence(outFASTA);
 269                                         if(barcodes.size() != 0){
 270                                                 outGroups << currSeq.getName() << '\t' << groupVector[group] << endl;
 271
 272                                                 if(allFiles){
 273                                                         currSeq.printSequence(*fastaFileNames[group]);
 274                                                 }
 275                                         }
 276                                 }
 277                                 else{
 278                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 279                                         currSeq.setUnaligned(origSeq);
 280                                         currSeq.printSequence(scrapFASTA);
 281                                 }
 282                         }
 283                         gobble(inFASTA);
 284                 }
 285                 inFASTA.close();
 286                 outFASTA.close();
 287                 scrapFASTA.close();
 288                 outGroups.close();
 289                 if(qFileName != "")     {       qFile.close();  }
 290
 291                 for(int i=0;i<fastaFileNames.size();i++){
 292                         fastaFileNames[i]->close();
 293                         delete fastaFileNames[i];
 294                 }
 295
 296                 for(int i=0;i<fastaFileNames.size();i++){
 297                         string seqName;
 298                         openInputFile(getRootName(fastaFile) + groupVector[i] + ".fasta", inFASTA);
 299                         ofstream outGroups;
 300                         openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
 301                         outputNames.push_back(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups");
 302
 303                         while(!inFASTA.eof()){
 304                                 if(inFASTA.get() == '>'){
 305                                         inFASTA >> seqName;
 306                                         outGroups << seqName << '\t' << groupVector[i] << endl;
 307                                 }
 308                                 while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 309                         }
 310                         outGroups.close();
 311                         inFASTA.close();
 312                 }
 313
 314                 m->mothurOutEndLine();
 315                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 316                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 317                 m->mothurOutEndLine();
 318
 319                 return 0;
 320         }
 321         catch(exception& e) {
 322                 m->errorOut(e, "TrimSeqsCommand", "execute");
 323                 exit(1);
 324         }
 325 }
 326
 327 //***************************************************************************************************************
 328
 329 void TrimSeqsCommand::getOligos(vector<ofstream*>& outFASTAVec){
 330         try {
 331                 ifstream inOligos;
 332                 openInputFile(oligoFile, inOligos);
 333
 334                 ofstream test;
 335
 336                 string type, oligo, group;
 337                 int index=0;
 338
 339                 while(!inOligos.eof()){
 340                         inOligos >> type;
 341
 342                         if(type[0] == '#'){
 343                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 344                         }
 345                         else{
 346                                 inOligos >> oligo;
 347
 348                                 for(int i=0;i<oligo.length();i++){
 349                                         oligo[i] = toupper(oligo[i]);
 350                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 351                                 }
 352
 353                                 if(type == "forward"){
 354                                         forPrimer.push_back(oligo);
 355                                 }
 356                                 else if(type == "reverse"){
 357                                         Sequence oligoRC("reverse", oligo);
 358                                         oligoRC.reverseComplement();
 359                                         revPrimer.push_back(oligoRC.getUnaligned());
 360                                 }
 361                                 else if(type == "barcode"){
 362                                         inOligos >> group;
 363                                         barcodes[oligo]=index++;
 364                                         groupVector.push_back(group);
 365
 366                                         if(allFiles){
 367                                                 outFASTAVec.push_back(new ofstream((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta").c_str(), ios::ate));
 368                                         }
 369                                 }
 370                         }
 371                 }
 372
 373                 inOligos.close();
 374
 375                 numFPrimers = forPrimer.size();
 376                 numRPrimers = revPrimer.size();
 377
 378         }
 379         catch(exception& e) {
 380                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 381                 exit(1);
 382         }
 383
 384 }
 385
 386 //***************************************************************************************************************
 387
 388 bool TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
 389         try {
 390                 string rawSequence = seq.getUnaligned();
 391                 bool success = 0;       //guilty until proven innocent
 392
 393                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 394                         string oligo = it->first;
 395                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 396                                 success = 0;
 397                                 break;
 398                         }
 399
 400                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 401                                 group = it->second;
 402                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 403                                 success = 1;
 404                                 break;
 405                         }
 406                 }
 407                 return success;
 408
 409         }
 410         catch(exception& e) {
 411                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 412                 exit(1);
 413         }
 414
 415 }
 416
 417 //***************************************************************************************************************
 418
 419 bool TrimSeqsCommand::stripForward(Sequence& seq){
 420         try {
 421                 string rawSequence = seq.getUnaligned();
 422                 bool success = 0;       //guilty until proven innocent
 423
 424                 for(int i=0;i<numFPrimers;i++){
 425                         string oligo = forPrimer[i];
 426
 427                         if(rawSequence.length() < oligo.length()){
 428                                 success = 0;
 429                                 break;
 430                         }
 431
 432                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 433                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 434                                 success = 1;
 435                                 break;
 436                         }
 437                 }
 438
 439                 return success;
 440
 441         }
 442         catch(exception& e) {
 443                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
 444                 exit(1);
 445         }
 446 }
 447
 448 //***************************************************************************************************************
 449
 450 bool TrimSeqsCommand::stripReverse(Sequence& seq){
 451         try {
 452                 string rawSequence = seq.getUnaligned();
 453                 bool success = 0;       //guilty until proven innocent
 454
 455                 for(int i=0;i<numRPrimers;i++){
 456                         string oligo = revPrimer[i];
 457
 458                         if(rawSequence.length() < oligo.length()){
 459                                 success = 0;
 460                                 break;
 461                         }
 462
 463                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 464                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 465                                 success = 1;
 466                                 break;
 467                         }
 468                 }
 469                 return success;
 470
 471         }
 472         catch(exception& e) {
 473                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
 474                 exit(1);
 475         }
 476 }
 477
 478 //***************************************************************************************************************
 479
 480 bool TrimSeqsCommand::cullLength(Sequence& seq){
 481         try {
 482
 483                 int length = seq.getNumBases();
 484                 bool success = 0;       //guilty until proven innocent
 485
 486                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
 487                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
 488                 else                                                                                            {       success = 0;    }
 489
 490                 return success;
 491
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
 495                 exit(1);
 496         }
 497
 498 }
 499
 500 //***************************************************************************************************************
 501
 502 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
 503         try {
 504                 int longHomoP = seq.getLongHomoPolymer();
 505                 bool success = 0;       //guilty until proven innocent
 506
 507                 if(longHomoP <= maxHomoP){      success = 1;    }
 508                 else                                    {       success = 0;    }
 509
 510                 return success;
 511         }
 512         catch(exception& e) {
 513                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
 514                 exit(1);
 515         }
 516
 517 }
 518
 519 //***************************************************************************************************************
 520
 521 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
 522         try {
 523                 int numNs = seq.getAmbigBases();
 524                 bool success = 0;       //guilty until proven innocent
 525
 526                 if(numNs <= maxAmbig)   {       success = 1;    }
 527                 else                                    {       success = 0;    }
 528
 529                 return success;
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
 533                 exit(1);
 534         }
 535
 536 }
 537
 538 //***************************************************************************************************************
 539
 540 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
 541         try {
 542                 bool success = 1;
 543                 int length = oligo.length();
 544
 545                 for(int i=0;i<length;i++){
 546
 547                         if(oligo[i] != seq[i]){
 548                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 549                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 550                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 551                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 552                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 553                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 554                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 555                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 556                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 557                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 558                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 559                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 560
 561                                 if(success == 0)        {       break;  }
 562                         }
 563                         else{
 564                                 success = 1;
 565                         }
 566                 }
 567
 568                 return success;
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
 572                 exit(1);
 573         }
 574
 575 }
 576
 577 //***************************************************************************************************************
 578
 579 bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
 580         try {
 581                 string rawSequence = seq.getUnaligned();
 582                 int seqLength = rawSequence.length();
 583                 string name;
 584
 585                 qFile >> name;
 586                 if (name.length() != 0) {  if(name.substr(1) != seq.getName())  {       m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); }  }
 587                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 588
 589                 int score;
 590                 int end = seqLength;
 591
 592                 for(int i=0;i<seqLength;i++){
 593                         qFile >> score;
 594
 595                         if(score <= qThreshold){
 596                                 end = i;
 597                                 break;
 598                         }
 599                 }
 600                 for(int i=end+1;i<seqLength;i++){
 601                         qFile >> score;
 602                 }
 603
 604                 seq.setUnaligned(rawSequence.substr(0,end));
 605
 606                 return 1;
 607         }
 608         catch(exception& e) {
 609                 m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
 610                 exit(1);
 611         }
 612 }
 613
 614 //***************************************************************************************************************
 615
 616 bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
 617         try {
 618                 string rawSequence = seq.getUnaligned();
 619                 int seqLength = seq.getNumBases();
 620                 bool success = 0;       //guilty until proven innocent
 621                 string name;
 622
 623                 qFile >> name;
 624                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
 625
 626                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 627
 628                 float score;
 629                 float average = 0;
 630
 631                 for(int i=0;i<seqLength;i++){
 632                         qFile >> score;
 633                         average += score;
 634                 }
 635                 average /= seqLength;
 636
 637                 if(average >= qAverage) {       success = 1;    }
 638                 else                                    {       success = 0;    }
 639
 640                 return success;
 641         }
 642         catch(exception& e) {
 643                 m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
 644                 exit(1);
 645         }
 646 }
 647
 648 //***************************************************************************************************************