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 #include "needlemanoverlap.hpp"
  12
  13 //***************************************************************************************************************
  14
  15 TrimSeqsCommand::TrimSeqsCommand(string option)  {
  16         try {
  17
  18                 abort = false;
  19                 comboStarts = 0;
  20
  21                 //allow user to run help
  22                 if(option == "help") { help(); abort = true; }
  23
  24                 else {
  25                         //valid paramters for this command
  26                         string AlignArray[] =  {"fasta", "flip", "oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
  27                                                                         "qthreshold", "qaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
  28
  29                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  30
  31                         OptionParser parser(option);
  32                         map<string,string> parameters = parser.getParameters();
  33
  34                         ValidParameters validParameter;
  35                         map<string,string>::iterator it;
  36
  37                         //check to make sure all parameters are valid for command
  38                         for (it = parameters.begin(); it != parameters.end(); it++) {
  39                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  40                         }
  41
  42                         //if the user changes the input directory command factory will send this info to us in the output parameter
  43                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  44                         if (inputDir == "not found"){   inputDir = "";          }
  45                         else {
  46                                 string path;
  47                                 it = parameters.find("fasta");
  48                                 //user has given a template file
  49                                 if(it != parameters.end()){
  50                                         path = hasPath(it->second);
  51                                         //if the user has not given a path then, add inputdir. else leave path alone.
  52                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  53                                 }
  54
  55                                 it = parameters.find("oligos");
  56                                 //user has given a template file
  57                                 if(it != parameters.end()){
  58                                         path = hasPath(it->second);
  59                                         //if the user has not given a path then, add inputdir. else leave path alone.
  60                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
  61                                 }
  62
  63                                 it = parameters.find("qfile");
  64                                 //user has given a template file
  65                                 if(it != parameters.end()){
  66                                         path = hasPath(it->second);
  67                                         //if the user has not given a path then, add inputdir. else leave path alone.
  68                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
  69                                 }
  70                         }
  71
  72
  73                         //check for required parameters
  74                         fastaFile = validParameter.validFile(parameters, "fasta", true);
  75                         if (fastaFile == "not found") { m->mothurOut("fasta is a required parameter for the screen.seqs command."); m->mothurOutEndLine(); abort = true; }
  76                         else if (fastaFile == "not open") { abort = true; }
  77
  78                         //if the user changes the output directory command factory will send this info to us in the output parameter
  79                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  80                                 outputDir = "";
  81                                 outputDir += hasPath(fastaFile); //if user entered a file with a path then preserve it
  82                         }
  83
  84                         //check for optional parameter and set defaults
  85                         // ...at some point should added some additional type checking...
  86                         string temp;
  87                         temp = validParameter.validFile(parameters, "flip", false);
  88                         if (temp == "not found"){       flip = 0;       }
  89                         else if(isTrue(temp))   {       flip = 1;       }
  90
  91                         temp = validParameter.validFile(parameters, "oligos", true);
  92                         if (temp == "not found"){       oligoFile = "";         }
  93                         else if(temp == "not open"){    abort = true;   }
  94                         else                                    {       oligoFile = temp;               }
  95
  96                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
  97                         convert(temp, maxAmbig);
  98
  99                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "0"; }
 100                         convert(temp, maxHomoP);
 101
 102                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "0"; }
 103                         convert(temp, minLength);
 104
 105                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "0"; }
 106                         convert(temp, maxLength);
 107
 108
 109                         temp = validParameter.validFile(parameters, "bdiffs", false);           if (temp == "not found") { temp = "0"; }
 110                         convert(temp, bdiffs);
 111
 112                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found") { temp = "0"; }
 113                         convert(temp, pdiffs);
 114
 115                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs;  temp = toString(tempTotal); }
 116                         convert(temp, tdiffs);
 117
 118                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs;       }
 119
 120                         temp = validParameter.validFile(parameters, "qfile", true);
 121                         if (temp == "not found")        {       qFileName = "";         }
 122                         else if(temp == "not open")     {       abort = true;           }
 123                         else                                            {       qFileName = temp;       }
 124
 125                         temp = validParameter.validFile(parameters, "qthreshold", false);       if (temp == "not found") { temp = "0"; }
 126                         convert(temp, qThreshold);
 127
 128                         temp = validParameter.validFile(parameters, "qtrim", false);    if (temp == "not found") { temp = "F"; }
 129                         qtrim = isTrue(temp);
 130
 131                         temp = validParameter.validFile(parameters, "qaverage", false);         if (temp == "not found") { temp = "0"; }
 132                         convert(temp, qAverage);
 133
 134                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found") { temp = "F"; }
 135                         allFiles = isTrue(temp);
 136
 137                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 138                         convert(temp, processors);
 139
 140                         if(allFiles && oligoFile == ""){
 141                                 m->mothurOut("You selected allfiles, but didn't enter an oligos file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 142                         }
 143                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 144                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 145                                 qAverage=0;
 146                                 qThreshold=0;
 147                         }
 148                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 149                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 150                                 abort = true;
 151                         }
 152                 }
 153
 154         }
 155         catch(exception& e) {
 156                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 157                 exit(1);
 158         }
 159 }
 160 //**********************************************************************************************************************
 161
 162 void TrimSeqsCommand::help(){
 163         try {
 164                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 165                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim and allfiles.\n");
 166                 m->mothurOut("The fasta parameter is required.\n");
 167                 m->mothurOut("The flip parameter will output the reverse compliment of your trimmed sequence. The default is false.\n");
 168                 m->mothurOut("The oligos parameter .... The default is "".\n");
 169                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 170                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 171                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 172                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 173                 m->mothurOut("The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs.\n");
 174                 m->mothurOut("The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n");
 175                 m->mothurOut("The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n");
 176                 m->mothurOut("The qfile parameter .....\n");
 177                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 178                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 179                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 180                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 181                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 182                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 183                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 184                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 185                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 186                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 187
 188         }
 189         catch(exception& e) {
 190                 m->errorOut(e, "TrimSeqsCommand", "help");
 191                 exit(1);
 192         }
 193 }
 194
 195
 196 //***************************************************************************************************************
 197
 198 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 199
 200 //***************************************************************************************************************
 201
 202 int TrimSeqsCommand::execute(){
 203         try{
 204
 205                 if (abort == true) { return 0; }
 206
 207                 numFPrimers = 0;  //this needs to be initialized
 208                 numRPrimers = 0;
 209
 210                 string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
 211                 outputNames.push_back(trimSeqFile);
 212                 string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
 213                 outputNames.push_back(scrapSeqFile);
 214                 string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
 215
 216                 vector<string> fastaFileNames;
 217                 if(oligoFile != ""){
 218                         outputNames.push_back(groupFile);
 219                         getOligos(fastaFileNames);
 220                 }
 221
 222                 if(qFileName != "")     {       setLines(qFileName, qLines);    }
 223
 224
 225                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 226                                 if(processors == 1){
 227                                         ifstream inFASTA;
 228                                         int numSeqs;
 229                                         openInputFile(fastaFile, inFASTA);
 230                                         getNumSeqs(inFASTA, numSeqs);
 231                                         inFASTA.close();
 232
 233                                         lines.push_back(new linePair(0, numSeqs));
 234
 235                                         driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 236
 237                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 238                                                 rename((fastaFileNames[j] + toString(getpid()) + ".temp").c_str(), fastaFileNames[j].c_str());
 239                                         }
 240
 241                                 }else{
 242                                         setLines(fastaFile, lines);
 243                                         if(qFileName == "")     {       qLines = lines; }
 244
 245                                         createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames);
 246
 247                                         rename((trimSeqFile + toString(processIDS[0]) + ".temp").c_str(), trimSeqFile.c_str());
 248                                         rename((scrapSeqFile + toString(processIDS[0]) + ".temp").c_str(), scrapSeqFile.c_str());
 249                                         rename((groupFile + toString(processIDS[0]) + ".temp").c_str(), groupFile.c_str());
 250                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 251                                                 rename((fastaFileNames[j] + toString(processIDS[0]) + ".temp").c_str(), fastaFileNames[j].c_str());
 252                                         }
 253                                         //append files
 254                                         for(int i=1;i<processors;i++){
 255                                                 appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
 256                                                 remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
 257                                                 appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
 258                                                 remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
 259                                                 appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 260                                                 remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 261                                                 for (int j = 0; j < fastaFileNames.size(); j++) {
 262                                                         appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
 263                                                         remove((fastaFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
 264                                                 }
 265                                         }
 266                                 }
 267
 268                                 if (m->control_pressed) {  return 0; }
 269                 #else
 270                                 ifstream inFASTA;
 271                                 int numSeqs;
 272                                 openInputFile(fastaFile, inFASTA);
 273                                 getNumSeqs(inFASTA, numSeqs);
 274                                 inFASTA.close();
 275
 276                                 lines.push_back(new linePair(0, numSeqs));
 277
 278                                 driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 279
 280                                 if (m->control_pressed) {  return 0; }
 281                 #endif
 282
 283
 284                 for(int i=0;i<fastaFileNames.size();i++){
 285                         if (isBlank(fastaFileNames[i])) { remove(fastaFileNames[i].c_str()); }
 286                         else {
 287                                 ifstream inFASTA;
 288                                 string seqName;
 289                                 //openInputFile(getRootName(fastaFile) +  groupVector[i] + ".fasta", inFASTA);
 290                                 openInputFile(fastaFileNames[i], inFASTA);
 291                                 ofstream outGroups;
 292                                 string outGroupFilename = outputDir + getRootName(getSimpleName(fastaFileNames[i])) + "groups";
 293                                 openOutputFile(outGroupFilename, outGroups);
 294                                 //openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
 295                                 outputNames.push_back(outGroupFilename);
 296
 297                                 string thisGroup = "";
 298                                 if (i > comboStarts) {
 299                                         map<string, int>::iterator itCombo;
 300                                         for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 301                                                 if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 302                                         }
 303                                 }else{ thisGroup = groupVector[i]; }
 304
 305                                 while(!inFASTA.eof()){
 306                                         if(inFASTA.get() == '>'){
 307                                                 inFASTA >> seqName;
 308                                                 outGroups << seqName << '\t' << thisGroup << endl;
 309                                         }
 310                                         while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 311                                 }
 312                                 outGroups.close();
 313                                 inFASTA.close();
 314                         }
 315                 }
 316
 317                 if (m->control_pressed) {
 318                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 319                         return 0;
 320                 }
 321
 322                 m->mothurOutEndLine();
 323                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 324                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 325                 m->mothurOutEndLine();
 326
 327                 return 0;
 328
 329         }
 330         catch(exception& e) {
 331                 m->errorOut(e, "TrimSeqsCommand", "execute");
 332                 exit(1);
 333         }
 334 }
 335
 336 /**************************************************************************************/
 337 int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames, linePair* line, linePair* qline) {
 338         try {
 339
 340                 ofstream outFASTA;
 341                 int able = openOutputFile(trimFile, outFASTA);
 342
 343                 ofstream scrapFASTA;
 344                 openOutputFile(scrapFile, scrapFASTA);
 345
 346                 ofstream outGroups;
 347                 vector<ofstream*> fastaFileNames;
 348                 if (oligoFile != "") {
 349                         openOutputFile(groupFile, outGroups);
 350                         for (int i = 0; i < fastaNames.size(); i++) {
 351                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 352                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
 353                         #else
 354                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(i) + ".temp").c_str(), ios::ate));
 355                         #endif
 356                         }
 357                 }
 358
 359                 ifstream inFASTA;
 360                 openInputFile(filename, inFASTA);
 361
 362                 ifstream qFile;
 363                 if(qFileName != "")     {       openInputFile(qFileName, qFile);        }
 364
 365                 qFile.seekg(qline->start);
 366                 inFASTA.seekg(line->start);
 367
 368                 for(int i=0;i<line->num;i++){
 369
 370                         if (m->control_pressed) {
 371                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close(); if (oligoFile != "") {    outGroups.close();   } if(qFileName != "")     {       qFile.close();  }
 372                                 for(int i=0;i<fastaFileNames.size();i++){  fastaFileNames[i]->close(); delete fastaFileNames[i];  }
 373                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 374                                 return 0;
 375                         }
 376
 377                         int success = 1;
 378
 379                         Sequence currSeq(inFASTA);
 380
 381                         string origSeq = currSeq.getUnaligned();
 382                         if (origSeq != "") {
 383                                 int groupBar, groupPrime;
 384                                 string trashCode = "";
 385                                 int currentSeqsDiffs = 0;
 386
 387                                 if(qFileName != ""){
 388                                         if(qThreshold != 0)             {       success = stripQualThreshold(currSeq, qFile);   }
 389                                         else if(qAverage != 0)  {       success = cullQualAverage(currSeq, qFile);              }
 390
 391                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 392                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 393                                         }
 394
 395                                         if(!success)                    {       trashCode += 'q';                                                               }
 396                                 }
 397
 398                                 if(barcodes.size() != 0){
 399                                         success = stripBarcode(currSeq, groupBar);
 400                                         if(success > bdiffs){   trashCode += 'b';       }
 401                                         else{ currentSeqsDiffs += success;  }
 402                                 }
 403
 404                                 if(numFPrimers != 0){
 405                                         success = stripForward(currSeq, groupPrime);
 406                                         if(success > pdiffs){   trashCode += 'f';       }
 407                                         else{ currentSeqsDiffs += success;  }
 408                                 }
 409
 410                                 if (currentSeqsDiffs > tdiffs) { trashCode += 't';   }
 411
 412                                 if(numRPrimers != 0){
 413                                         success = stripReverse(currSeq);
 414                                         if(!success){   trashCode += 'r';       }
 415                                 }
 416
 417                                 if(minLength > 0 || maxLength > 0){
 418                                         success = cullLength(currSeq);
 419                                         if(!success){   trashCode += 'l'; }
 420                                 }
 421                                 if(maxHomoP > 0){
 422                                         success = cullHomoP(currSeq);
 423                                         if(!success){   trashCode += 'h';       }
 424                                 }
 425                                 if(maxAmbig != -1){
 426                                         success = cullAmbigs(currSeq);
 427                                         if(!success){   trashCode += 'n';       }
 428                                 }
 429
 430                                 if(flip){       currSeq.reverseComplement();    }               // should go last
 431
 432                                 if(trashCode.length() == 0){
 433                                         currSeq.setAligned(currSeq.getUnaligned());
 434                                         currSeq.printSequence(outFASTA);
 435                                         if(barcodes.size() != 0){
 436                                                 string thisGroup = groupVector[groupBar];
 437                                                 int indexToFastaFile = groupBar;
 438                                                 if (primers.size() != 0){
 439                                                         //does this primer have a group
 440                                                         if (groupVector[groupPrime] != "") {
 441                                                                 thisGroup += "." + groupVector[groupPrime];
 442                                                                 indexToFastaFile = combos[thisGroup];
 443                                                         }
 444                                                 }
 445                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 446
 447                                                 if(allFiles){
 448                                                         currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 449                                                 }
 450                                         }
 451                                 }
 452                                 else{
 453                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 454                                         currSeq.setUnaligned(origSeq);
 455                                         currSeq.setAligned(origSeq);
 456                                         currSeq.printSequence(scrapFASTA);
 457                                 }
 458                         }
 459                         gobble(inFASTA);
 460                 }
 461
 462                 inFASTA.close();
 463                 outFASTA.close();
 464                 scrapFASTA.close();
 465                 if (oligoFile != "") {   outGroups.close();   }
 466                 if(qFileName != "")     {       qFile.close();  }
 467
 468                 for(int i=0;i<fastaFileNames.size();i++){
 469                         fastaFileNames[i]->close();
 470                         delete fastaFileNames[i];
 471                 }
 472
 473                 return 0;
 474         }
 475         catch(exception& e) {
 476                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 477                 exit(1);
 478         }
 479 }
 480 /**************************************************************************************************/
 481 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames) {
 482         try {
 483 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 484                 int process = 0;
 485                 int exitCommand = 1;
 486                 processIDS.clear();
 487
 488                 //loop through and create all the processes you want
 489                 while (process != processors) {
 490                         int pid = fork();
 491
 492                         if (pid > 0) {
 493                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 494                                 process++;
 495                         }else if (pid == 0){
 496                                 driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, lines[process], qLines[process]);
 497                                 exit(0);
 498                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 499                 }
 500
 501                 //force parent to wait until all the processes are done
 502                 for (int i=0;i<processors;i++) {
 503                         int temp = processIDS[i];
 504                         wait(&temp);
 505                 }
 506
 507                 return exitCommand;
 508 #endif
 509         }
 510         catch(exception& e) {
 511                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 512                 exit(1);
 513         }
 514 }
 515 /**************************************************************************************************/
 516
 517 int TrimSeqsCommand::setLines(string filename, vector<linePair*>& lines) {
 518         try {
 519
 520                 lines.clear();
 521
 522                 vector<long int> positions;
 523
 524                 ifstream inFASTA;
 525                 openInputFile(filename, inFASTA);
 526
 527                 string input;
 528                 while(!inFASTA.eof()){
 529                         input = getline(inFASTA);
 530
 531                         if (input.length() != 0) {
 532                                 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);     }
 533                         }
 534                 }
 535                 inFASTA.close();
 536
 537                 int numFastaSeqs = positions.size();
 538
 539                 FILE * pFile;
 540                 long size;
 541
 542                 //get num bytes in file
 543                 pFile = fopen (filename.c_str(),"rb");
 544                 if (pFile==NULL) perror ("Error opening file");
 545                 else{
 546                         fseek (pFile, 0, SEEK_END);
 547                         size=ftell (pFile);
 548                         fclose (pFile);
 549                 }
 550
 551                 int numSeqsPerProcessor = numFastaSeqs / processors;
 552
 553                 for (int i = 0; i < processors; i++) {
 554
 555                         long int startPos = positions[ i * numSeqsPerProcessor ];
 556                         if(i == processors - 1){
 557                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 558                         }else{
 559                                 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
 560                         }
 561                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 562                 }
 563
 564                 return numFastaSeqs;
 565         }
 566         catch(exception& e) {
 567                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 568                 exit(1);
 569         }
 570 }
 571 //***************************************************************************************************************
 572
 573 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec){ //vector<ofstream*>& outFASTAVec
 574         try {
 575                 ifstream inOligos;
 576                 openInputFile(oligoFile, inOligos);
 577
 578                 ofstream test;
 579
 580                 string type, oligo, group;
 581                 int index=0;
 582                 //int indexPrimer = 0;
 583
 584                 while(!inOligos.eof()){
 585                         inOligos >> type;
 586
 587                         if(type[0] == '#'){
 588                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 589                         }
 590                         else{
 591                                 inOligos >> oligo;
 592
 593                                 for(int i=0;i<oligo.length();i++){
 594                                         oligo[i] = toupper(oligo[i]);
 595                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 596                                 }
 597
 598                                 if(type == "forward"){
 599                                         group = "";
 600
 601                                         // get rest of line in case there is a primer name
 602                                         while (!inOligos.eof()) {
 603                                                 char c = inOligos.get();
 604                                                 if (c == 10 || c == 13){        break;  }
 605                                                 else if (c == 32 || c == 9){;} //space or tab
 606                                                 else {  group += c;  }
 607                                         }
 608
 609                                         //check for repeat barcodes
 610                                         map<string, int>::iterator itPrime = primers.find(oligo);
 611                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 612
 613                                         primers[oligo]=index; index++;
 614                                         groupVector.push_back(group);
 615
 616                                         if(allFiles){
 617                                                 if (group != "") { //there is a group for this primer
 618                                                         outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
 619                                                         outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
 620                                                 }
 621                                         }
 622                                 }
 623                                 else if(type == "reverse"){
 624                                         Sequence oligoRC("reverse", oligo);
 625                                         oligoRC.reverseComplement();
 626                                         revPrimer.push_back(oligoRC.getUnaligned());
 627                                 }
 628                                 else if(type == "barcode"){
 629                                         inOligos >> group;
 630
 631                                         //check for repeat barcodes
 632                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 633                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 634
 635                                         barcodes[oligo]=index; index++;
 636                                         groupVector.push_back(group);
 637
 638                                         if(allFiles){
 639                                                 outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
 640                                                 outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
 641                                         }
 642                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 643                         }
 644                         gobble(inOligos);
 645                 }
 646
 647                 inOligos.close();
 648
 649                 //add in potential combos
 650                 if(allFiles){
 651                         comboStarts = outFASTAVec.size()-1;
 652                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 653                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 654                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 655                                                 outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(itBar->second) + "." + groupVector[itBar->second] + "." + toString(itPrime->second) + "." + groupVector[itPrime->second] + ".fasta"));
 656                                                 outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(itBar->second) + "." + groupVector[itBar->second] + "." + toString(itPrime->second) + "." + groupVector[itPrime->second] + ".fasta"));
 657                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 658                                         }
 659                                 }
 660                         }
 661                 }
 662
 663                 numFPrimers = primers.size();
 664                 numRPrimers = revPrimer.size();
 665
 666         }
 667         catch(exception& e) {
 668                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 669                 exit(1);
 670         }
 671 }
 672 //***************************************************************************************************************
 673
 674 int TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
 675         try {
 676
 677                 string rawSequence = seq.getUnaligned();
 678                 int success = bdiffs + 1;       //guilty until proven innocent
 679
 680                 //can you find the barcode
 681                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 682                         string oligo = it->first;
 683                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 684                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 685                                 break;
 686                         }
 687
 688                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 689                                 group = it->second;
 690                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 691                                 success = 0;
 692                                 break;
 693                         }
 694                 }
 695
 696                 //if you found the barcode or if you don't want to allow for diffs
 697 //              cout << success;
 698                 if ((bdiffs == 0) || (success == 0)) { return success;  }
 699
 700                 else { //try aligning and see if you can find it
 701 //                      cout << endl;
 702
 703                         int maxLength = 0;
 704
 705                         Alignment* alignment;
 706                         if (barcodes.size() > 0) {
 707                                 map<string,int>::iterator it=barcodes.begin();
 708
 709                                 for(it;it!=barcodes.end();it++){
 710                                         if(it->first.length() > maxLength){
 711                                                 maxLength = it->first.length();
 712                                         }
 713                                 }
 714                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
 715
 716                         }else{ alignment = NULL; }
 717
 718                         //can you find the barcode
 719                         int minDiff = 1e6;
 720                         int minCount = 1;
 721                         int minGroup = -1;
 722                         int minPos = 0;
 723
 724                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 725                                 string oligo = it->first;
 726 //                              int length = oligo.length();
 727
 728                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
 729                                         success = bdiffs + 10;
 730                                         break;
 731                                 }
 732
 733                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 734                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
 735                                 oligo = alignment->getSeqAAln();
 736                                 string temp = alignment->getSeqBAln();
 737
 738                                 int alnLength = oligo.length();
 739
 740                                 for(int i=oligo.length()-1;i>=0;i--){
 741                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 742                                 }
 743                                 oligo = oligo.substr(0,alnLength);
 744                                 temp = temp.substr(0,alnLength);
 745
 746                                 int newStart=0;
 747                                 int numDiff = countDiffs(oligo, temp);
 748
 749 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
 750
 751                                 if(numDiff < minDiff){
 752                                         minDiff = numDiff;
 753                                         minCount = 1;
 754                                         minGroup = it->second;
 755                                         minPos = 0;
 756                                         for(int i=0;i<alnLength;i++){
 757                                                 if(temp[i] != '-'){
 758                                                         minPos++;
 759                                                 }
 760                                         }
 761                                 }
 762                                 else if(numDiff == minDiff){
 763                                         minCount++;
 764                                 }
 765
 766                         }
 767
 768                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
 769                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
 770                         else{                                                                                                   //use the best match
 771                                 group = minGroup;
 772                                 seq.setUnaligned(rawSequence.substr(minPos));
 773                                 success = minDiff;
 774                         }
 775
 776                         if (alignment != NULL) {  delete alignment;  }
 777
 778                 }
 779 //              cout << success << endl;
 780
 781                 return success;
 782
 783         }
 784         catch(exception& e) {
 785                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 786                 exit(1);
 787         }
 788
 789 }
 790
 791 //***************************************************************************************************************
 792
 793 int TrimSeqsCommand::stripForward(Sequence& seq, int& group){
 794         try {
 795                 string rawSequence = seq.getUnaligned();
 796                 int success = pdiffs + 1;       //guilty until proven innocent
 797
 798                 //can you find the primer
 799                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 800                         string oligo = it->first;
 801                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
 802                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 803                                 break;
 804                         }
 805
 806                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 807                                 group = it->second;
 808                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 809                                 success = 0;
 810                                 break;
 811                         }
 812                 }
 813
 814                 //if you found the barcode or if you don't want to allow for diffs
 815 //              cout << success;
 816                 if ((pdiffs == 0) || (success == 0)) { return success;  }
 817
 818                 else { //try aligning and see if you can find it
 819 //                      cout << endl;
 820
 821                         int maxLength = 0;
 822
 823                         Alignment* alignment;
 824                         if (primers.size() > 0) {
 825                                 map<string,int>::iterator it=primers.begin();
 826
 827                                 for(it;it!=primers.end();it++){
 828                                         if(it->first.length() > maxLength){
 829                                                 maxLength = it->first.length();
 830                                         }
 831                                 }
 832                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
 833
 834                         }else{ alignment = NULL; }
 835
 836                         //can you find the barcode
 837                         int minDiff = 1e6;
 838                         int minCount = 1;
 839                         int minGroup = -1;
 840                         int minPos = 0;
 841
 842                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 843                                 string oligo = it->first;
 844 //                              int length = oligo.length();
 845
 846                                 if(rawSequence.length() < maxLength){
 847                                         success = pdiffs + 100;
 848                                         break;
 849                                 }
 850
 851                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 852                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
 853                                 oligo = alignment->getSeqAAln();
 854                                 string temp = alignment->getSeqBAln();
 855
 856                                 int alnLength = oligo.length();
 857
 858                                 for(int i=oligo.length()-1;i>=0;i--){
 859                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 860                                 }
 861                                 oligo = oligo.substr(0,alnLength);
 862                                 temp = temp.substr(0,alnLength);
 863
 864                                 int newStart=0;
 865                                 int numDiff = countDiffs(oligo, temp);
 866
 867 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
 868
 869                                 if(numDiff < minDiff){
 870                                         minDiff = numDiff;
 871                                         minCount = 1;
 872                                         minGroup = it->second;
 873                                         minPos = 0;
 874                                         for(int i=0;i<alnLength;i++){
 875                                                 if(temp[i] != '-'){
 876                                                         minPos++;
 877                                                 }
 878                                         }
 879                                 }
 880                                 else if(numDiff == minDiff){
 881                                         minCount++;
 882                                 }
 883
 884                         }
 885
 886                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
 887                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
 888                         else{                                                                                                   //use the best match
 889                                 group = minGroup;
 890                                 seq.setUnaligned(rawSequence.substr(minPos));
 891                                 success = minDiff;
 892                         }
 893
 894                         if (alignment != NULL) {  delete alignment;  }
 895
 896                 }
 897
 898                 return success;
 899
 900         }
 901         catch(exception& e) {
 902                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
 903                 exit(1);
 904         }
 905 }
 906
 907 //***************************************************************************************************************
 908
 909 bool TrimSeqsCommand::stripReverse(Sequence& seq){
 910         try {
 911                 string rawSequence = seq.getUnaligned();
 912                 bool success = 0;       //guilty until proven innocent
 913
 914                 for(int i=0;i<numRPrimers;i++){
 915                         string oligo = revPrimer[i];
 916
 917                         if(rawSequence.length() < oligo.length()){
 918                                 success = 0;
 919                                 break;
 920                         }
 921
 922                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 923                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 924                                 success = 1;
 925                                 break;
 926                         }
 927                 }
 928                 return success;
 929
 930         }
 931         catch(exception& e) {
 932                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
 933                 exit(1);
 934         }
 935 }
 936
 937 //***************************************************************************************************************
 938
 939 bool TrimSeqsCommand::cullLength(Sequence& seq){
 940         try {
 941
 942                 int length = seq.getNumBases();
 943                 bool success = 0;       //guilty until proven innocent
 944
 945                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
 946                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
 947                 else                                                                                            {       success = 0;    }
 948
 949                 return success;
 950
 951         }
 952         catch(exception& e) {
 953                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
 954                 exit(1);
 955         }
 956
 957 }
 958
 959 //***************************************************************************************************************
 960
 961 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
 962         try {
 963                 int longHomoP = seq.getLongHomoPolymer();
 964                 bool success = 0;       //guilty until proven innocent
 965
 966                 if(longHomoP <= maxHomoP){      success = 1;    }
 967                 else                                    {       success = 0;    }
 968
 969                 return success;
 970         }
 971         catch(exception& e) {
 972                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
 973                 exit(1);
 974         }
 975
 976 }
 977
 978 //***************************************************************************************************************
 979
 980 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
 981         try {
 982                 int numNs = seq.getAmbigBases();
 983                 bool success = 0;       //guilty until proven innocent
 984
 985                 if(numNs <= maxAmbig)   {       success = 1;    }
 986                 else                                    {       success = 0;    }
 987
 988                 return success;
 989         }
 990         catch(exception& e) {
 991                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
 992                 exit(1);
 993         }
 994
 995 }
 996
 997 //***************************************************************************************************************
 998
 999 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1000         try {
1001                 bool success = 1;
1002                 int length = oligo.length();
1003
1004                 for(int i=0;i<length;i++){
1005
1006                         if(oligo[i] != seq[i]){
1007                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1008                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1009                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1010                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1011                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1012                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1013                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1014                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1015                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1016                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1017                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1018                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1019
1020                                 if(success == 0)        {       break;   }
1021                         }
1022                         else{
1023                                 success = 1;
1024                         }
1025                 }
1026
1027                 return success;
1028         }
1029         catch(exception& e) {
1030                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1031                 exit(1);
1032         }
1033
1034 }
1035 //***************************************************************************************************************
1036
1037 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1038         try {
1039
1040                 int length = oligo.length();
1041                 int countDiffs = 0;
1042
1043                 for(int i=0;i<length;i++){
1044
1045                         if(oligo[i] != seq[i]){
1046                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1047                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1048                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1049                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1050                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1051                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1052                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1053                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1054                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1055                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1056                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1057                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1058
1059                         }
1060                 }
1061
1062                 return countDiffs;
1063         }
1064         catch(exception& e) {
1065                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1066                 exit(1);
1067         }
1068
1069 }
1070 //***************************************************************************************************************
1071
1072 bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1073         try {
1074 //              string rawSequence = seq.getUnaligned();
1075 //              int seqLength;  // = rawSequence.length();
1076 //              string name, temp, temp2;
1077 //
1078 //              qFile >> name;
1079 //
1080 //              //get rest of line
1081 //              temp = "";
1082 //              while (!qFile.eof())    {
1083 //                      char c = qFile.get();
1084 //                      if (c == 10 || c == 13){        break;  }
1085 //                      else { temp += c; }
1086 //              }
1087 //
1088 //              int pos = temp.find("length");
1089 //              if (pos == temp.npos) { m->mothurOut("Cannot find length in qfile for " + seq.getName()); m->mothurOutEndLine();  seqLength = 0;  }
1090 //              else {
1091 //                      string tempLength = temp.substr(pos);
1092 //                      istringstream iss (tempLength,istringstream::in);
1093 //                      iss >> temp;
1094 //              }
1095 //
1096 //              splitAtEquals(temp2, temp); //separates length=242, temp=length, temp2=242
1097 //              convert(temp, seqLength); //converts string to int
1098 //
1099 //              if (name.length() != 0) {  if(name.substr(1) != seq.getName())  {       m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); }  }
1100
1101                 string rawSequence = seq.getUnaligned();
1102                 int seqLength = seq.getNumBases();
1103                 bool success = 0;       //guilty until proven innocent
1104                 string name;
1105
1106                 qFile >> name;
1107                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1108
1109                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1110
1111                 int score;
1112                 int end = seqLength;
1113
1114                 for(int i=0;i<seqLength;i++){
1115                         qFile >> score;
1116
1117                         if(score < qThreshold){
1118                                 end = i;
1119                                 break;
1120                         }
1121                 }
1122                 for(int i=end+1;i<seqLength;i++){
1123                         qFile >> score;
1124                 }
1125
1126                 seq.setUnaligned(rawSequence.substr(0,end));
1127
1128                 return 1;
1129         }
1130         catch(exception& e) {
1131                 m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1132                 exit(1);
1133         }
1134 }
1135
1136 //***************************************************************************************************************
1137
1138 bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1139         try {
1140                 string rawSequence = seq.getUnaligned();
1141                 int seqLength = seq.getNumBases();
1142                 bool success = 0;       //guilty until proven innocent
1143                 string name;
1144
1145                 qFile >> name;
1146                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1147
1148                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1149
1150                 float score;
1151                 float average = 0;
1152
1153                 for(int i=0;i<seqLength;i++){
1154                         qFile >> score;
1155                         average += score;
1156                 }
1157                 average /= seqLength;
1158
1159                 if(average >= qAverage) {       success = 1;    }
1160                 else                                    {       success = 0;    }
1161
1162                 return success;
1163         }
1164         catch(exception& e) {
1165                 m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1166                 exit(1);
1167         }
1168 }
1169
1170 //***************************************************************************************************************