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", "processors", "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                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 124                         convert(temp, processors);
 125
 126                         if(allFiles && oligoFile == ""){
 127                                 m->mothurOut("You selected allfiles, but didn't enter an oligos file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 128                         }
 129                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 130                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 131                                 qAverage=0;
 132                                 qThreshold=0;
 133                         }
 134                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 135                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 136                                 abort = true;
 137                         }
 138                 }
 139
 140         }
 141         catch(exception& e) {
 142                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 143                 exit(1);
 144         }
 145 }
 146 //**********************************************************************************************************************
 147
 148 void TrimSeqsCommand::help(){
 149         try {
 150                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 151                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, qtrim and allfiles.\n");
 152                 m->mothurOut("The fasta parameter is required.\n");
 153                 m->mothurOut("The flip parameter .... The default is 0.\n");
 154                 m->mothurOut("The oligos parameter .... The default is "".\n");
 155                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 156                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 157                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 158                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 159                 m->mothurOut("The qfile parameter .....\n");
 160                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 161                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 162                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 163                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 164                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 165                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 166                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 167                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 168                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 169                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 170
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "TrimSeqsCommand", "help");
 174                 exit(1);
 175         }
 176 }
 177
 178
 179 //***************************************************************************************************************
 180
 181 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 182
 183 //***************************************************************************************************************
 184
 185 int TrimSeqsCommand::execute(){
 186         try{
 187
 188                 if (abort == true) { return 0; }
 189
 190                 numFPrimers = 0;  //this needs to be initialized
 191                 numRPrimers = 0;
 192
 193                 string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
 194                 outputNames.push_back(trimSeqFile);
 195                 string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
 196                 outputNames.push_back(scrapSeqFile);
 197                 string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
 198
 199                 vector<string> fastaFileNames;
 200                 if(oligoFile != ""){
 201                         outputNames.push_back(groupFile);
 202                         getOligos(fastaFileNames);
 203                 }
 204
 205                 if(qFileName != "")     {       setLines(qFileName, qLines);    }
 206
 207
 208                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 209                                 if(processors == 1){
 210                                         ifstream inFASTA;
 211                                         openInputFile(fastaFile, inFASTA);
 212                                         int numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 213                                         inFASTA.close();
 214
 215                                         lines.push_back(new linePair(0, numSeqs));
 216
 217                                         driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 218
 219                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 220                                                 rename((fastaFileNames[j] + toString(getpid()) + ".temp").c_str(), fastaFileNames[j].c_str());
 221                                         }
 222
 223                                 }else{
 224                                         setLines(fastaFile, lines);
 225                                         if(qFileName == "")     {       qLines = lines; }
 226
 227                                         createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames);
 228
 229                                         rename((trimSeqFile + toString(processIDS[0]) + ".temp").c_str(), trimSeqFile.c_str());
 230                                         rename((scrapSeqFile + toString(processIDS[0]) + ".temp").c_str(), scrapSeqFile.c_str());
 231                                         rename((groupFile + toString(processIDS[0]) + ".temp").c_str(), groupFile.c_str());
 232                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 233                                                 rename((fastaFileNames[j] + toString(processIDS[0]) + ".temp").c_str(), fastaFileNames[j].c_str());
 234                                         }
 235                                         //append files
 236                                         for(int i=1;i<processors;i++){
 237                                                 appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
 238                                                 remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
 239                                                 appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
 240                                                 remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
 241                                                 appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 242                                                 remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 243                                                 for (int j = 0; j < fastaFileNames.size(); j++) {
 244                                                         appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
 245                                                         remove((fastaFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
 246                                                 }
 247                                         }
 248                                 }
 249
 250                                 if (m->control_pressed) {  return 0; }
 251                 #else
 252                                 ifstream inFASTA;
 253                                 openInputFile(fastafileNames[s], inFASTA);
 254                                 numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 255                                 inFASTA.close();
 256
 257                                 lines.push_back(new linePair(0, numSeqs));
 258
 259                                 driverCreateSummary(fastafile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 260
 261                                 if (m->control_pressed) {  return 0; }
 262                 #endif
 263
 264
 265                 for(int i=0;i<fastaFileNames.size();i++){
 266                         ifstream inFASTA;
 267                         string seqName;
 268                         openInputFile(getRootName(fastaFile) + groupVector[i] + ".fasta", inFASTA);
 269                         ofstream outGroups;
 270                         openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
 271                         outputNames.push_back(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups");
 272
 273                         while(!inFASTA.eof()){
 274                                 if(inFASTA.get() == '>'){
 275                                         inFASTA >> seqName;
 276                                         outGroups << seqName << '\t' << groupVector[i] << endl;
 277                                 }
 278                                 while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 279                         }
 280                         outGroups.close();
 281                         inFASTA.close();
 282                 }
 283
 284                 if (m->control_pressed) {
 285                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 286                         return 0;
 287                 }
 288
 289                 m->mothurOutEndLine();
 290                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 291                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 292                 m->mothurOutEndLine();
 293
 294                 return 0;
 295
 296         }
 297         catch(exception& e) {
 298                 m->errorOut(e, "TrimSeqsCommand", "execute");
 299                 exit(1);
 300         }
 301 }
 302
 303 /**************************************************************************************/
 304 int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames, linePair* line, linePair* qline) {
 305         try {
 306
 307                 ofstream outFASTA;
 308                 int able = openOutputFile(trimFile, outFASTA);
 309
 310                 ofstream scrapFASTA;
 311                 openOutputFile(scrapFile, scrapFASTA);
 312
 313                 ofstream outGroups;
 314                 vector<ofstream*> fastaFileNames;
 315                 if (oligoFile != "") {
 316                         openOutputFile(groupFile, outGroups);
 317                         for (int i = 0; i < fastaNames.size(); i++) {
 318                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
 319                         }
 320                 }
 321
 322                 ifstream inFASTA;
 323                 openInputFile(filename, inFASTA);
 324
 325                 ifstream qFile;
 326                 if(qFileName != "")     {       openInputFile(qFileName, qFile);        }
 327
 328                 qFile.seekg(qline->start);
 329                 inFASTA.seekg(line->start);
 330
 331                 for(int i=0;i<line->num;i++){
 332
 333                         if (m->control_pressed) {
 334                                 inFASTA.close();
 335                                 outFASTA.close();
 336                                 scrapFASTA.close();
 337                                 if (oligoFile != "") {   outGroups.close();   }
 338                                 if(qFileName != "")     {       qFile.close();  }
 339                                 for(int i=0;i<fastaFileNames.size();i++){
 340                                         fastaFileNames[i]->close();
 341                                         delete fastaFileNames[i];
 342                                 }
 343                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 344                                 return 0;
 345                         }
 346
 347                         bool success = 1;
 348
 349                         Sequence currSeq(inFASTA);
 350
 351                         string origSeq = currSeq.getUnaligned();
 352                         if (origSeq != "") {
 353                                 int group;
 354                                 string trashCode = "";
 355
 356                                 if(qFileName != ""){
 357                                         if(qThreshold != 0)             {       success = stripQualThreshold(currSeq, qFile);   }
 358                                         else if(qAverage != 0)  {       success = cullQualAverage(currSeq, qFile);              }
 359                                         if ((!qtrim) && (origSeq.length() != currSeq.getUnaligned().length())) {
 360                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 361                                         }
 362                                         if(!success)                    {       trashCode += 'q';                                                               }
 363                                 }
 364
 365                                 if(barcodes.size() != 0){
 366                                         success = stripBarcode(currSeq, group);
 367                                         if(!success){   trashCode += 'b';       }
 368                                 }
 369
 370                                 if(numFPrimers != 0){
 371                                         success = stripForward(currSeq);
 372                                         if(!success){   trashCode += 'f';       }
 373                                 }
 374
 375                                 if(numRPrimers != 0){
 376                                         success = stripReverse(currSeq);
 377                                         if(!success){   trashCode += 'r';       }
 378                                 }
 379
 380                                 if(minLength > 0 || maxLength > 0){
 381                                         success = cullLength(currSeq);
 382                                         if(!success){   trashCode += 'l'; }
 383                                 }
 384                                 if(maxHomoP > 0){
 385                                         success = cullHomoP(currSeq);
 386                                         if(!success){   trashCode += 'h';       }
 387                                 }
 388                                 if(maxAmbig != -1){
 389                                         success = cullAmbigs(currSeq);
 390                                         if(!success){   trashCode += 'n';       }
 391                                 }
 392
 393                                 if(flip){       currSeq.reverseComplement();    }               // should go last
 394
 395                                 if(trashCode.length() == 0){
 396                                         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.
 397                                         currSeq.printSequence(outFASTA);
 398                                         if(barcodes.size() != 0){
 399                                                 outGroups << currSeq.getName() << '\t' << groupVector[group] << endl;
 400
 401                                                 if(allFiles){
 402                                                         currSeq.printSequence(*fastaFileNames[group]);
 403                                                 }
 404                                         }
 405                                 }
 406                                 else{
 407                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 408                                         currSeq.setUnaligned(origSeq);
 409                                         currSeq.printSequence(scrapFASTA);
 410                                 }
 411                         }
 412                         gobble(inFASTA);
 413                 }
 414
 415                 inFASTA.close();
 416                 outFASTA.close();
 417                 scrapFASTA.close();
 418                 if (oligoFile != "") {   outGroups.close();   }
 419                 if(qFileName != "")     {       qFile.close();  }
 420
 421                 for(int i=0;i<fastaFileNames.size();i++){
 422                         fastaFileNames[i]->close();
 423                         delete fastaFileNames[i];
 424                 }
 425
 426                 return 0;
 427         }
 428         catch(exception& e) {
 429                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 430                 exit(1);
 431         }
 432 }
 433 /**************************************************************************************************/
 434 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames) {
 435         try {
 436 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 437                 int process = 0;
 438                 int exitCommand = 1;
 439                 processIDS.clear();
 440
 441                 //loop through and create all the processes you want
 442                 while (process != processors) {
 443                         int pid = fork();
 444
 445                         if (pid > 0) {
 446                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 447                                 process++;
 448                         }else if (pid == 0){
 449                                 driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, lines[process], qLines[process]);
 450                                 exit(0);
 451                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 452                 }
 453
 454                 //force parent to wait until all the processes are done
 455                 for (int i=0;i<processors;i++) {
 456                         int temp = processIDS[i];
 457                         wait(&temp);
 458                 }
 459
 460                 return exitCommand;
 461 #endif
 462         }
 463         catch(exception& e) {
 464                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 465                 exit(1);
 466         }
 467 }
 468 /**************************************************************************************************/
 469
 470 int TrimSeqsCommand::setLines(string filename, vector<linePair*>& lines) {
 471         try {
 472
 473                 lines.clear();
 474
 475                 vector<long int> positions;
 476
 477                 ifstream inFASTA;
 478                 openInputFile(filename, inFASTA);
 479
 480                 string input;
 481                 while(!inFASTA.eof()){
 482                         input = getline(inFASTA);
 483
 484                         if (input.length() != 0) {
 485                                 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);     }
 486                         }
 487                 }
 488                 inFASTA.close();
 489
 490                 int numFastaSeqs = positions.size();
 491
 492                 FILE * pFile;
 493                 long size;
 494
 495                 //get num bytes in file
 496                 pFile = fopen (filename.c_str(),"rb");
 497                 if (pFile==NULL) perror ("Error opening file");
 498                 else{
 499                         fseek (pFile, 0, SEEK_END);
 500                         size=ftell (pFile);
 501                         fclose (pFile);
 502                 }
 503
 504                 int numSeqsPerProcessor = numFastaSeqs / processors;
 505
 506                 for (int i = 0; i < processors; i++) {
 507
 508                         long int startPos = positions[ i * numSeqsPerProcessor ];
 509                         if(i == processors - 1){
 510                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 511                         }else{
 512                                 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
 513                         }
 514                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 515                 }
 516
 517                 return numFastaSeqs;
 518         }
 519         catch(exception& e) {
 520                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 521                 exit(1);
 522         }
 523 }
 524 //***************************************************************************************************************
 525
 526 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec){ //vector<ofstream*>& outFASTAVec
 527         try {
 528                 ifstream inOligos;
 529                 openInputFile(oligoFile, inOligos);
 530
 531                 ofstream test;
 532
 533                 string type, oligo, group;
 534                 int index=0;
 535
 536                 while(!inOligos.eof()){
 537                         inOligos >> type;
 538
 539                         if(type[0] == '#'){
 540                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 541                         }
 542                         else{
 543                                 inOligos >> oligo;
 544
 545                                 for(int i=0;i<oligo.length();i++){
 546                                         oligo[i] = toupper(oligo[i]);
 547                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 548                                 }
 549
 550                                 if(type == "forward"){
 551                                         forPrimer.push_back(oligo);
 552                                 }
 553                                 else if(type == "reverse"){
 554                                         Sequence oligoRC("reverse", oligo);
 555                                         oligoRC.reverseComplement();
 556                                         revPrimer.push_back(oligoRC.getUnaligned());
 557                                 }
 558                                 else if(type == "barcode"){
 559                                         inOligos >> group;
 560                                         barcodes[oligo]=index++;
 561                                         groupVector.push_back(group);
 562
 563                                         if(allFiles){
 564                                                 //outFASTAVec.push_back(new ofstream((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta").c_str(), ios::ate));
 565                                                 outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta"));
 566                                                 outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta"));
 567                                         }
 568                                 }
 569                         }
 570                 }
 571
 572                 inOligos.close();
 573
 574                 numFPrimers = forPrimer.size();
 575                 numRPrimers = revPrimer.size();
 576
 577         }
 578         catch(exception& e) {
 579                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 580                 exit(1);
 581         }
 582
 583 }
 584
 585 //***************************************************************************************************************
 586
 587 bool TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
 588         try {
 589                 string rawSequence = seq.getUnaligned();
 590                 bool success = 0;       //guilty until proven innocent
 591
 592                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 593                         string oligo = it->first;
 594                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 595                                 success = 0;
 596                                 break;
 597                         }
 598
 599                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 600                                 group = it->second;
 601                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 602                                 success = 1;
 603                                 break;
 604                         }
 605                 }
 606                 return success;
 607
 608         }
 609         catch(exception& e) {
 610                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 611                 exit(1);
 612         }
 613
 614 }
 615
 616 //***************************************************************************************************************
 617
 618 bool TrimSeqsCommand::stripForward(Sequence& seq){
 619         try {
 620                 string rawSequence = seq.getUnaligned();
 621                 bool success = 0;       //guilty until proven innocent
 622
 623                 for(int i=0;i<numFPrimers;i++){
 624                         string oligo = forPrimer[i];
 625
 626                         if(rawSequence.length() < oligo.length()){
 627                                 success = 0;
 628                                 break;
 629                         }
 630
 631                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 632                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 633                                 success = 1;
 634                                 break;
 635                         }
 636                 }
 637
 638                 return success;
 639
 640         }
 641         catch(exception& e) {
 642                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
 643                 exit(1);
 644         }
 645 }
 646
 647 //***************************************************************************************************************
 648
 649 bool TrimSeqsCommand::stripReverse(Sequence& seq){
 650         try {
 651                 string rawSequence = seq.getUnaligned();
 652                 bool success = 0;       //guilty until proven innocent
 653
 654                 for(int i=0;i<numRPrimers;i++){
 655                         string oligo = revPrimer[i];
 656
 657                         if(rawSequence.length() < oligo.length()){
 658                                 success = 0;
 659                                 break;
 660                         }
 661
 662                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 663                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 664                                 success = 1;
 665                                 break;
 666                         }
 667                 }
 668                 return success;
 669
 670         }
 671         catch(exception& e) {
 672                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
 673                 exit(1);
 674         }
 675 }
 676
 677 //***************************************************************************************************************
 678
 679 bool TrimSeqsCommand::cullLength(Sequence& seq){
 680         try {
 681
 682                 int length = seq.getNumBases();
 683                 bool success = 0;       //guilty until proven innocent
 684
 685                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
 686                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
 687                 else                                                                                            {       success = 0;    }
 688
 689                 return success;
 690
 691         }
 692         catch(exception& e) {
 693                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
 694                 exit(1);
 695         }
 696
 697 }
 698
 699 //***************************************************************************************************************
 700
 701 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
 702         try {
 703                 int longHomoP = seq.getLongHomoPolymer();
 704                 bool success = 0;       //guilty until proven innocent
 705
 706                 if(longHomoP <= maxHomoP){      success = 1;    }
 707                 else                                    {       success = 0;    }
 708
 709                 return success;
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
 713                 exit(1);
 714         }
 715
 716 }
 717
 718 //***************************************************************************************************************
 719
 720 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
 721         try {
 722                 int numNs = seq.getAmbigBases();
 723                 bool success = 0;       //guilty until proven innocent
 724
 725                 if(numNs <= maxAmbig)   {       success = 1;    }
 726                 else                                    {       success = 0;    }
 727
 728                 return success;
 729         }
 730         catch(exception& e) {
 731                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
 732                 exit(1);
 733         }
 734
 735 }
 736
 737 //***************************************************************************************************************
 738
 739 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
 740         try {
 741                 bool success = 1;
 742                 int length = oligo.length();
 743
 744                 for(int i=0;i<length;i++){
 745
 746                         if(oligo[i] != seq[i]){
 747                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 748                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 749                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 750                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 751                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 752                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 753                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 754                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 755                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 756                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 757                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 758                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 759
 760                                 if(success == 0)        {       break;  }
 761                         }
 762                         else{
 763                                 success = 1;
 764                         }
 765                 }
 766
 767                 return success;
 768         }
 769         catch(exception& e) {
 770                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
 771                 exit(1);
 772         }
 773
 774 }
 775
 776 //***************************************************************************************************************
 777
 778 bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
 779         try {
 780                 string rawSequence = seq.getUnaligned();
 781                 int seqLength = rawSequence.length();
 782                 string name;
 783
 784                 qFile >> name;
 785                 if (name.length() != 0) {  if(name.substr(1) != seq.getName())  {       m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); }  }
 786                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 787
 788                 int score;
 789                 int end = seqLength;
 790
 791                 for(int i=0;i<seqLength;i++){
 792                         qFile >> score;
 793
 794                         if(score <= qThreshold){
 795                                 end = i;
 796                                 break;
 797                         }
 798                 }
 799                 for(int i=end+1;i<seqLength;i++){
 800                         qFile >> score;
 801                 }
 802
 803                 seq.setUnaligned(rawSequence.substr(0,end));
 804
 805                 return 1;
 806         }
 807         catch(exception& e) {
 808                 m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
 809                 exit(1);
 810         }
 811 }
 812
 813 //***************************************************************************************************************
 814
 815 bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
 816         try {
 817                 string rawSequence = seq.getUnaligned();
 818                 int seqLength = seq.getNumBases();
 819                 bool success = 0;       //guilty until proven innocent
 820                 string name;
 821
 822                 qFile >> name;
 823                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
 824
 825                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 826
 827                 float score;
 828                 float average = 0;
 829
 830                 for(int i=0;i<seqLength;i++){
 831                         qFile >> score;
 832                         average += score;
 833                 }
 834                 average /= seqLength;
 835
 836                 if(average >= qAverage) {       success = 1;    }
 837                 else                                    {       success = 0;    }
 838
 839                 return success;
 840         }
 841         catch(exception& e) {
 842                 m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
 843                 exit(1);
 844         }
 845 }
 846
 847 //***************************************************************************************************************