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", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "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 = m->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 = m->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 = m->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 += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
  82                         }
  83
  84
  85                         //check for optional parameter and set defaults
  86                         // ...at some point should added some additional type checking...
  87                         string temp;
  88                         temp = validParameter.validFile(parameters, "flip", false);
  89                         if (temp == "not found"){       flip = 0;       }
  90                         else if(m->isTrue(temp))        {       flip = 1;       }
  91
  92                         temp = validParameter.validFile(parameters, "oligos", true);
  93                         if (temp == "not found"){       oligoFile = "";         }
  94                         else if(temp == "not open"){    abort = true;   }
  95                         else                                    {       oligoFile = temp;               }
  96
  97                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
  98                         convert(temp, maxAmbig);
  99
 100                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "0"; }
 101                         convert(temp, maxHomoP);
 102
 103                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "0"; }
 104                         convert(temp, minLength);
 105
 106                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "0"; }
 107                         convert(temp, maxLength);
 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 = m->isTrue(temp);
 130
 131                         temp = validParameter.validFile(parameters, "rollaverage", false);      if (temp == "not found") { temp = "0"; }
 132                         convert(temp, qRollAverage);
 133
 134                         temp = validParameter.validFile(parameters, "qwindowaverage", false);if (temp == "not found") { temp = "0"; }
 135                         convert(temp, qWindowAverage);
 136
 137                         temp = validParameter.validFile(parameters, "qwindowsize", false);      if (temp == "not found") { temp = "100"; }
 138                         convert(temp, qWindowSize);
 139
 140                         temp = validParameter.validFile(parameters, "qstepsize", false);                if (temp == "not found") { temp = "10"; }
 141                         convert(temp, qWindowStep);
 142
 143                         temp = validParameter.validFile(parameters, "qaverage", false);         if (temp == "not found") { temp = "0"; }
 144                         convert(temp, qAverage);
 145
 146                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found") { temp = "F"; }
 147                         allFiles = m->isTrue(temp);
 148
 149                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found") { temp = "1"; }
 150                         convert(temp, processors);
 151
 152                         if(allFiles && oligoFile == ""){
 153                                 m->mothurOut("You selected allfiles, but didn't enter an oligos file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 154                         }
 155                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 156                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 157                                 qAverage=0;
 158                                 qThreshold=0;
 159                         }
 160                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 161                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 162                                 abort = true;
 163                         }
 164                 }
 165
 166         }
 167         catch(exception& e) {
 168                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 169                 exit(1);
 170         }
 171 }
 172 //**********************************************************************************************************************
 173
 174 void TrimSeqsCommand::help(){
 175         try {
 176                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 177                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim and allfiles.\n");
 178                 m->mothurOut("The fasta parameter is required.\n");
 179                 m->mothurOut("The flip parameter will output the reverse compliment of your trimmed sequence. The default is false.\n");
 180                 m->mothurOut("The oligos parameter .... The default is "".\n");
 181                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 182                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 183                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 184                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 185                 m->mothurOut("The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs.\n");
 186                 m->mothurOut("The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n");
 187                 m->mothurOut("The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n");
 188                 m->mothurOut("The qfile parameter .....\n");
 189                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 190                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 191                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 192                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 193                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 194                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 195                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 196                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 197                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 198                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 199
 200         }
 201         catch(exception& e) {
 202                 m->errorOut(e, "TrimSeqsCommand", "help");
 203                 exit(1);
 204         }
 205 }
 206
 207
 208 //***************************************************************************************************************
 209
 210 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 211
 212 //***************************************************************************************************************
 213
 214 int TrimSeqsCommand::execute(){
 215         try{
 216
 217                 if (abort == true) { return 0; }
 218
 219                 numFPrimers = 0;  //this needs to be initialized
 220                 numRPrimers = 0;
 221
 222                 string trimSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.fasta";
 223                 outputNames.push_back(trimSeqFile);
 224                 string scrapSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.fasta";
 225                 outputNames.push_back(scrapSeqFile);
 226                 string trimQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.qual";
 227                 outputNames.push_back(trimQualFile);
 228                 string scrapQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.qual";
 229                 outputNames.push_back(scrapQualFile);
 230                 string groupFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "groups";
 231
 232                 vector<string> fastaFileNames;
 233                 vector<string> qualFileNames;
 234                 if(oligoFile != ""){
 235                         outputNames.push_back(groupFile);
 236                         getOligos(fastaFileNames, qualFileNames);
 237                 }
 238
 239                 vector<unsigned long int> fastaFilePos;
 240                 vector<unsigned long int> qFilePos;
 241
 242                 setLines(fastaFile, qFileName, fastaFilePos, qFilePos);
 243
 244                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 245                         lines.push_back(new linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 246                         if (qFileName != "") {  qLines.push_back(new linePair(qFilePos[i], qFilePos[(i+1)]));  }
 247                 }
 248                 if(qFileName == "")     {       qLines = lines; } //files with duds
 249
 250                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 251                                 if(processors == 1){
 252
 253                                         driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qLines[0]);
 254
 255                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 256                                                 rename((fastaFileNames[j] + toString(getpid()) + ".temp").c_str(), fastaFileNames[j].c_str());
 257                                         }
 258                                         if(qFileName != ""){
 259                                                 for (int j = 0; j < qualFileNames.size(); j++) {
 260                                                         rename((qualFileNames[j] + toString(getpid()) + ".temp").c_str(), qualFileNames[j].c_str());
 261                                                 }
 262                                         }
 263
 264                                 }else{
 265                                         createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames);
 266
 267                                         rename((trimSeqFile + toString(processIDS[0]) + ".temp").c_str(), trimSeqFile.c_str());
 268                                         rename((scrapSeqFile + toString(processIDS[0]) + ".temp").c_str(), scrapSeqFile.c_str());
 269                                         rename((groupFile + toString(processIDS[0]) + ".temp").c_str(), groupFile.c_str());
 270
 271                                         if(qFileName != ""){
 272                                                 rename((trimQualFile + toString(processIDS[0]) + ".temp").c_str(), trimQualFile.c_str());
 273                                                 rename((scrapQualFile + toString(processIDS[0]) + ".temp").c_str(), scrapQualFile.c_str());
 274                                         }
 275
 276
 277                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 278                                                 rename((fastaFileNames[j] + toString(processIDS[0]) + ".temp").c_str(), fastaFileNames[j].c_str());
 279                                         }
 280                                         if(qFileName != ""){
 281                                                 for (int j = 0; j < qualFileNames.size(); j++) {
 282                                                         rename((qualFileNames[j] + toString(getpid()) + ".temp").c_str(), qualFileNames[j].c_str());
 283                                                 }
 284                                         }
 285
 286                                         //append files
 287                                         for(int i=1;i<processors;i++){
 288                                                 m->appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
 289                                                 remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
 290                                                 m->appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
 291                                                 remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
 292
 293                                                 m->appendFiles((trimQualFile + toString(processIDS[i]) + ".temp"), trimQualFile);
 294                                                 remove((trimQualFile + toString(processIDS[i]) + ".temp").c_str());
 295                                                 m->appendFiles((scrapQualFile + toString(processIDS[i]) + ".temp"), scrapQualFile);
 296                                                 remove((scrapQualFile + toString(processIDS[i]) + ".temp").c_str());
 297
 298                                                 m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 299                                                 remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 300                                                 for (int j = 0; j < fastaFileNames.size(); j++) {
 301                                                         m->appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
 302                                                         remove((fastaFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
 303                                                 }
 304
 305                                                 if(qFileName != ""){
 306                                                         for (int j = 0; j < qualFileNames.size(); j++) {
 307                                                                 m->appendFiles((qualFileNames[j] + toString(processIDS[i]) + ".temp"), qualFileNames[j]);
 308                                                                 remove((qualFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
 309                                                         }
 310                                                 }
 311
 312
 313                                         }
 314                                 }
 315
 316                                 if (m->control_pressed) {  return 0; }
 317                 #else
 318                                 driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qlines[0]);
 319
 320                                 for (int j = 0; j < fastaFileNames.size(); j++) {
 321                                         rename((fastaFileNames[j] + toString(j) + ".temp").c_str(), fastaFileNames[j].c_str());
 322                                 }
 323                                 if(qFileName != ""){
 324                                         for (int j = 0; j < qualFileNames.size(); j++) {
 325                                                 rename((qualFileNames[j] + toString(j) + ".temp").c_str(), qualFileNames[j].c_str());
 326                                         }
 327                                 }
 328
 329                                 if (m->control_pressed) {  return 0; }
 330                 #endif
 331
 332
 333                 for(int i=0;i<fastaFileNames.size();i++){
 334                         if (m->isBlank(fastaFileNames[i])) { remove(fastaFileNames[i].c_str()); }
 335                         else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
 336                         else {
 337                                 ifstream inFASTA;
 338                                 string seqName;
 339                                 m->openInputFile(fastaFileNames[i], inFASTA);
 340                                 ofstream outGroups;
 341                                 string outGroupFilename = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "groups";
 342                                 m->openOutputFile(outGroupFilename, outGroups);
 343                                 outputNames.push_back(outGroupFilename);
 344
 345                                 string thisGroup = "";
 346                                 if (i > comboStarts) {
 347                                         map<string, int>::iterator itCombo;
 348                                         for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 349                                                 if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 350                                         }
 351                                 }else{ thisGroup = groupVector[i]; }
 352
 353                                 while(!inFASTA.eof()){
 354                                         if(inFASTA.get() == '>'){
 355                                                 inFASTA >> seqName;
 356                                                 outGroups << seqName << '\t' << thisGroup << endl;
 357                                         }
 358                                         while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 359                                 }
 360                                 outGroups.close();
 361                                 inFASTA.close();
 362                         }
 363                 }
 364
 365                 if(qFileName != ""){
 366                         for(int i=0;i<qualFileNames.size();i++){
 367                                 if (m->isBlank(qualFileNames[i])) { remove(qualFileNames[i].c_str());   }
 368                                 else if (filesToRemove.count(qualFileNames[i]) > 0) { remove(qualFileNames[i].c_str()); }
 369                                 else {
 370                                         ifstream inQual;
 371                                         string seqName;
 372                                         m->openInputFile(qualFileNames[i], inQual);
 373 //                                      ofstream outGroups;
 374 //
 375 //                                      string thisGroup = "";
 376 //                                      if (i > comboStarts) {
 377 //                                              map<string, int>::iterator itCombo;
 378 //                                              for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 379 //                                                      if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 380 //                                              }
 381 //                                      }
 382 //                                      else{ thisGroup = groupVector[i]; }
 383
 384                                         inQual.close();
 385                                 }
 386                         }
 387                 }
 388
 389
 390                 if (m->control_pressed) {
 391                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 392                         return 0;
 393                 }
 394
 395                 m->mothurOutEndLine();
 396                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 397                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 398                 m->mothurOutEndLine();
 399
 400                 return 0;
 401
 402         }
 403         catch(exception& e) {
 404                 m->errorOut(e, "TrimSeqsCommand", "execute");
 405                 exit(1);
 406         }
 407 }
 408
 409 /**************************************************************************************/
 410
 411 int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames, linePair* line, linePair* qline) {
 412
 413         try {
 414
 415                 ofstream outFASTA;
 416                 int able = m->openOutputFile(trimFile, outFASTA);
 417
 418                 ofstream scrapFASTA;
 419                 m->openOutputFile(scrapFile, scrapFASTA);
 420
 421                 ofstream outQual;
 422                 ofstream scrapQual;
 423                 if(qFileName != ""){
 424                         m->openOutputFile(trimQFile, outQual);
 425                         m->openOutputFile(scrapQFile, scrapQual);
 426                 }
 427
 428                 ofstream outGroups;
 429                 vector<ofstream*> fastaFileNames;
 430                 vector<ofstream*> qualFileNames;
 431
 432
 433                 if (oligoFile != "") {
 434                         m->openOutputFile(groupFile, outGroups);
 435                         for (int i = 0; i < fastaNames.size(); i++) {
 436                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 437                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
 438                                 if(qFileName != ""){
 439                                         qualFileNames.push_back(new ofstream((qualNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
 440                                 }
 441                         #else
 442                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(i) + ".temp").c_str(), ios::ate));
 443                                 if(qFileName != ""){
 444                                         qualFileNames.push_back(new ofstream((qualNames[i] + toString(i) + ".temp").c_str(), ios::ate));
 445                                 }
 446                         #endif
 447                         }
 448                 }
 449
 450                 ifstream inFASTA;
 451                 m->openInputFile(filename, inFASTA);
 452                 inFASTA.seekg(line->start);
 453
 454                 ifstream qFile;
 455                 if(qFileName != "")     {       m->openInputFile(qFileName, qFile);     qFile.seekg(qline->start);  }
 456
 457                 bool done = false;
 458                 int count = 0;
 459
 460                 while (!done) {
 461
 462                         if (m->control_pressed) {
 463                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close();
 464                                 if (oligoFile != "") {   outGroups.close();   }
 465
 466                                 for(int i=0;i<fastaFileNames.size();i++){  fastaFileNames[i]->close(); delete fastaFileNames[i];  }
 467
 468                                 if(qFileName != ""){
 469                                         qFile.close();
 470                                         for(int i=0;i<qualFileNames.size();i++){  qualFileNames[i]->close(); delete qualFileNames[i];  }
 471                                 }
 472                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 473
 474                                 return 0;
 475                         }
 476
 477                         int success = 1;
 478
 479
 480                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 481
 482                         QualityScores currQual;
 483                         if(qFileName != ""){
 484                                 currQual = QualityScores(qFile, currSeq.getNumBases());  m->gobble(qFile);
 485                         }
 486
 487                         string origSeq = currSeq.getUnaligned();
 488                         if (origSeq != "") {
 489                                 int groupBar, groupPrime;
 490                                 string trashCode = "";
 491                                 int currentSeqsDiffs = 0;
 492
 493                                 if(qFileName != ""){
 494                                         if(qThreshold != 0)                     {       success = currQual.stripQualThreshold(currSeq, qThreshold);                     }
 495                                         else if(qAverage != 0)          {       success = currQual.cullQualAverage(currSeq, qAverage);                          }
 496                                         else if(qRollAverage != 0)      {       success = currQual.stripQualRollingAverage(currSeq, qRollAverage);      }
 497                                         else if(qWindowAverage != 0){   success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage);   }
 498
 499                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 500                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 501                                         }
 502
 503                                         if(!success)                            {       trashCode += 'q';       }
 504                                 }
 505
 506                                 if(barcodes.size() != 0){
 507                                         success = stripBarcode(currSeq, currQual, groupBar);
 508                                         if(success > bdiffs)            {       trashCode += 'b';       }
 509                                         else{ currentSeqsDiffs += success;  }
 510                                 }
 511
 512                                 if(numFPrimers != 0){
 513                                         success = stripForward(currSeq, currQual, groupPrime);
 514                                         if(success > pdiffs)            {       trashCode += 'f';       }
 515                                         else{ currentSeqsDiffs += success;  }
 516                                 }
 517
 518                                 if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
 519
 520                                 if(numRPrimers != 0){
 521                                         success = stripReverse(currSeq, currQual);
 522                                         if(!success)                            {       trashCode += 'r';       }
 523                                 }
 524
 525                                 if(minLength > 0 || maxLength > 0){
 526                                         success = cullLength(currSeq);
 527                                         if(!success)                            {       trashCode += 'l';       }
 528                                 }
 529                                 if(maxHomoP > 0){
 530                                         success = cullHomoP(currSeq);
 531                                         if(!success)                            {       trashCode += 'h';       }
 532                                 }
 533                                 if(maxAmbig != -1){
 534                                         success = cullAmbigs(currSeq);
 535                                         if(!success)                            {       trashCode += 'n';       }
 536                                 }
 537
 538                                 if(flip){       currSeq.reverseComplement();            currQual.flipQScores(); }               // should go last
 539
 540                                 if(trashCode.length() == 0){
 541                                         currSeq.setAligned(currSeq.getUnaligned());
 542                                         currSeq.printSequence(outFASTA);
 543                                         currQual.printQScores(outQual);
 544
 545                                         if(barcodes.size() != 0){
 546                                                 string thisGroup = groupVector[groupBar];
 547                                                 int indexToFastaFile = groupBar;
 548                                                 if (primers.size() != 0){
 549                                                         //does this primer have a group
 550                                                         if (groupVector[groupPrime] != "") {
 551                                                                 thisGroup += "." + groupVector[groupPrime];
 552                                                                 indexToFastaFile = combos[thisGroup];
 553                                                         }
 554                                                 }
 555                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 556                                                 if(allFiles){
 557                                                         currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 558
 559                                                         if(qFileName != ""){
 560                                                                 currQual.printQScores(*qualFileNames[indexToFastaFile]);
 561                                                         }
 562                                                 }
 563                                         }
 564                                 }
 565                                 else{
 566                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 567                                         currSeq.setUnaligned(origSeq);
 568                                         currSeq.setAligned(origSeq);
 569                                         currSeq.printSequence(scrapFASTA);
 570                                         currQual.printQScores(scrapQual);
 571                                 }
 572                                 count++;
 573                         }
 574
 575                         unsigned long int pos = inFASTA.tellg();
 576                         if ((pos == -1) || (pos >= line->end)) { break; }
 577
 578                         //report progress
 579                         if((count) % 1000 == 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 580
 581                 }
 582                 //report progress
 583                 if((count) % 1000 != 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 584
 585
 586                 inFASTA.close();
 587                 outFASTA.close();
 588                 scrapFASTA.close();
 589                 if (oligoFile != "") {   outGroups.close();   }
 590                 if(qFileName != "")     {       qFile.close();  scrapQual.close(); outQual.close();     }
 591
 592                 for(int i=0;i<fastaFileNames.size();i++){
 593                         fastaFileNames[i]->close();
 594                         delete fastaFileNames[i];
 595                 }
 596
 597                 if(qFileName != ""){
 598                         for(int i=0;i<qualFileNames.size();i++){
 599                                 qualFileNames[i]->close();
 600                                 delete qualFileNames[i];
 601                         }
 602                 }
 603
 604                 return count;
 605         }
 606         catch(exception& e) {
 607                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 608                 exit(1);
 609         }
 610 }
 611
 612 /**************************************************************************************************/
 613
 614 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
 615         try {
 616 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 617                 int process = 0;
 618                 int exitCommand = 1;
 619                 processIDS.clear();
 620
 621                 //loop through and create all the processes you want
 622                 while (process != processors) {
 623                         int pid = fork();
 624
 625                         if (pid > 0) {
 626                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 627                                 process++;
 628                         }else if (pid == 0){
 629                                 driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (trimQFile + toString(getpid()) + ".temp"), (scrapQFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, qualNames, lines[process], qLines[process]);
 630                                 exit(0);
 631                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 632                 }
 633
 634                 //force parent to wait until all the processes are done
 635                 for (int i=0;i<processors;i++) {
 636                         int temp = processIDS[i];
 637                         wait(&temp);
 638                 }
 639
 640                 return exitCommand;
 641 #endif
 642         }
 643         catch(exception& e) {
 644                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 645                 exit(1);
 646         }
 647 }
 648
 649 /**************************************************************************************************/
 650
 651 int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
 652         try {
 653
 654                 //set file positions for fasta file
 655                 fastaFilePos = m->divideFile(filename, processors);
 656
 657                 if (qfilename == "") { return processors; }
 658
 659                 //get name of first sequence in each chunk
 660                 map<string, int> firstSeqNames;
 661                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 662                         ifstream in;
 663                         m->openInputFile(filename, in);
 664                         in.seekg(fastaFilePos[i]);
 665
 666                         Sequence temp(in);
 667                         firstSeqNames[temp.getName()] = i;
 668
 669                         in.close();
 670                 }
 671
 672                 //seach for filePos of each first name in the qfile and save in qfileFilePos
 673                 ifstream inQual;
 674                 m->openInputFile(qfilename, inQual);
 675
 676                 string input;
 677                 while(!inQual.eof()){
 678                         input = m->getline(inQual);
 679
 680                         if (input.length() != 0) {
 681                                 if(input[0] == '>'){ //this is a sequence name line
 682                                         istringstream nameStream(input);
 683
 684                                         string sname = "";  nameStream >> sname;
 685                                         sname = sname.substr(1);
 686
 687                                         map<string, int>::iterator it = firstSeqNames.find(sname);
 688
 689                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
 690                                                 unsigned long int pos = inQual.tellg();
 691                                                 qfileFilePos.push_back(pos - input.length() - 1);
 692                                                 firstSeqNames.erase(it);
 693                                         }
 694                                 }
 695                         }
 696
 697                         if (firstSeqNames.size() == 0) { break; }
 698                 }
 699                 inQual.close();
 700
 701                 if (firstSeqNames.size() != 0) {
 702                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
 703                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
 704                         }
 705                         qFileName = "";
 706                         return processors;
 707                 }
 708
 709                 //get last file position of qfile
 710                 FILE * pFile;
 711                 unsigned long int size;
 712
 713                 //get num bytes in file
 714                 pFile = fopen (qfilename.c_str(),"rb");
 715                 if (pFile==NULL) perror ("Error opening file");
 716                 else{
 717                         fseek (pFile, 0, SEEK_END);
 718                         size=ftell (pFile);
 719                         fclose (pFile);
 720                 }
 721
 722                 qfileFilePos.push_back(size);
 723
 724                 return processors;
 725         }
 726         catch(exception& e) {
 727                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 728                 exit(1);
 729         }
 730 }
 731 //***************************************************************************************************************
 732
 733 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
 734         try {
 735                 ifstream inOligos;
 736                 m->openInputFile(oligoFile, inOligos);
 737
 738                 ofstream test;
 739
 740                 string type, oligo, group;
 741                 int index=0;
 742                 //int indexPrimer = 0;
 743
 744                 while(!inOligos.eof()){
 745                         inOligos >> type;
 746
 747                         if(type[0] == '#'){
 748                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 749                         }
 750                         else{
 751                                 //make type case insensitive
 752                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 753
 754                                 inOligos >> oligo;
 755
 756                                 for(int i=0;i<oligo.length();i++){
 757                                         oligo[i] = toupper(oligo[i]);
 758                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 759                                 }
 760
 761                                 if(type == "FORWARD"){
 762                                         group = "";
 763
 764                                         // get rest of line in case there is a primer name
 765                                         while (!inOligos.eof()) {
 766                                                 char c = inOligos.get();
 767                                                 if (c == 10 || c == 13){        break;  }
 768                                                 else if (c == 32 || c == 9){;} //space or tab
 769                                                 else {  group += c;  }
 770                                         }
 771
 772                                         //check for repeat barcodes
 773                                         map<string, int>::iterator itPrime = primers.find(oligo);
 774                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 775
 776                                         primers[oligo]=index; index++;
 777                                         groupVector.push_back(group);
 778
 779                                         if(allFiles){
 780                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 781                                                 if(qFileName != ""){
 782                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 783                                                 }
 784                                                 if (group == "") { //if there is not a group for this primer, then this file will not get written to, but we add it to keep the indexes correct
 785                                                         filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 786                                                         if(qFileName != ""){
 787                                                                 filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 788                                                         }
 789                                                 }else {
 790                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 791                                                         if(qFileName != ""){
 792                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 793                                                         }
 794                                                 }
 795                                         }
 796
 797                                 }
 798                                 else if(type == "REVERSE"){
 799                                         Sequence oligoRC("reverse", oligo);
 800                                         oligoRC.reverseComplement();
 801                                         revPrimer.push_back(oligoRC.getUnaligned());
 802                                 }
 803                                 else if(type == "BARCODE"){
 804                                         inOligos >> group;
 805
 806                                         //check for repeat barcodes
 807                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 808                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 809
 810                                         barcodes[oligo]=index; index++;
 811                                         groupVector.push_back(group);
 812
 813                                         if(allFiles){
 814                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 815                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 816                                                 if(qFileName != ""){
 817                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 818                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 819                                                 }
 820                                         }
 821                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 822                         }
 823                         m->gobble(inOligos);
 824                 }
 825
 826                 inOligos.close();
 827
 828                 //add in potential combos
 829                 if(allFiles){
 830                         comboStarts = outFASTAVec.size()-1;
 831                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 832                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 833                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 834                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 835                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 836                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 837
 838                                                 if(qFileName != ""){
 839                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 840                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 841                                                 }
 842                                         }
 843                                 }
 844                         }
 845                 }
 846
 847                 numFPrimers = primers.size();
 848                 numRPrimers = revPrimer.size();
 849
 850         }
 851         catch(exception& e) {
 852                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 853                 exit(1);
 854         }
 855 }
 856 //***************************************************************************************************************
 857
 858 int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
 859         try {
 860
 861                 string rawSequence = seq.getUnaligned();
 862                 int success = bdiffs + 1;       //guilty until proven innocent
 863
 864                 //can you find the barcode
 865                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 866                         string oligo = it->first;
 867                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 868                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 869                                 break;
 870                         }
 871
 872                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 873                                 group = it->second;
 874                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 875
 876                                 if(qual.getName() != ""){
 877                                         qual.trimQScores(oligo.length(), -1);
 878                                 }
 879
 880                                 success = 0;
 881                                 break;
 882                         }
 883                 }
 884
 885                 //if you found the barcode or if you don't want to allow for diffs
 886 //              cout << success;
 887                 if ((bdiffs == 0) || (success == 0)) { return success;  }
 888
 889                 else { //try aligning and see if you can find it
 890 //                      cout << endl;
 891
 892                         int maxLength = 0;
 893
 894                         Alignment* alignment;
 895                         if (barcodes.size() > 0) {
 896                                 map<string,int>::iterator it=barcodes.begin();
 897
 898                                 for(it;it!=barcodes.end();it++){
 899                                         if(it->first.length() > maxLength){
 900                                                 maxLength = it->first.length();
 901                                         }
 902                                 }
 903                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
 904
 905                         }else{ alignment = NULL; }
 906
 907                         //can you find the barcode
 908                         int minDiff = 1e6;
 909                         int minCount = 1;
 910                         int minGroup = -1;
 911                         int minPos = 0;
 912
 913                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 914                                 string oligo = it->first;
 915 //                              int length = oligo.length();
 916
 917                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
 918                                         success = bdiffs + 10;
 919                                         break;
 920                                 }
 921
 922                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 923                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
 924                                 oligo = alignment->getSeqAAln();
 925                                 string temp = alignment->getSeqBAln();
 926
 927                                 int alnLength = oligo.length();
 928
 929                                 for(int i=oligo.length()-1;i>=0;i--){
 930                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
 931                                 }
 932                                 oligo = oligo.substr(0,alnLength);
 933                                 temp = temp.substr(0,alnLength);
 934
 935                                 int newStart=0;
 936                                 int numDiff = countDiffs(oligo, temp);
 937
 938 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
 939
 940                                 if(numDiff < minDiff){
 941                                         minDiff = numDiff;
 942                                         minCount = 1;
 943                                         minGroup = it->second;
 944                                         minPos = 0;
 945                                         for(int i=0;i<alnLength;i++){
 946                                                 if(temp[i] != '-'){
 947                                                         minPos++;
 948                                                 }
 949                                         }
 950                                 }
 951                                 else if(numDiff == minDiff){
 952                                         minCount++;
 953                                 }
 954
 955                         }
 956
 957                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
 958                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
 959                         else{                                                                                                   //use the best match
 960                                 group = minGroup;
 961                                 seq.setUnaligned(rawSequence.substr(minPos));
 962
 963                                 if(qual.getName() != ""){
 964                                         qual.trimQScores(minPos, -1);
 965                                 }
 966                                 success = minDiff;
 967                         }
 968
 969                         if (alignment != NULL) {  delete alignment;  }
 970
 971                 }
 972 //              cout << success << endl;
 973
 974                 return success;
 975
 976         }
 977         catch(exception& e) {
 978                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 979                 exit(1);
 980         }
 981
 982 }
 983
 984 //***************************************************************************************************************
 985
 986 int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
 987         try {
 988                 string rawSequence = seq.getUnaligned();
 989                 int success = pdiffs + 1;       //guilty until proven innocent
 990
 991                 //can you find the primer
 992                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
 993                         string oligo = it->first;
 994                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
 995                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 996                                 break;
 997                         }
 998
 999                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1000                                 group = it->second;
1001                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1002                                 if(qual.getName() != ""){
1003                                         qual.trimQScores(oligo.length(), -1);
1004
1005                                 }
1006                                 success = 0;
1007                                 break;
1008                         }
1009                 }
1010
1011                 //if you found the barcode or if you don't want to allow for diffs
1012 //              cout << success;
1013                 if ((pdiffs == 0) || (success == 0)) { return success;  }
1014
1015                 else { //try aligning and see if you can find it
1016 //                      cout << endl;
1017
1018                         int maxLength = 0;
1019
1020                         Alignment* alignment;
1021                         if (primers.size() > 0) {
1022                                 map<string,int>::iterator it=primers.begin();
1023
1024                                 for(it;it!=primers.end();it++){
1025                                         if(it->first.length() > maxLength){
1026                                                 maxLength = it->first.length();
1027                                         }
1028                                 }
1029                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
1030
1031                         }else{ alignment = NULL; }
1032
1033                         //can you find the barcode
1034                         int minDiff = 1e6;
1035                         int minCount = 1;
1036                         int minGroup = -1;
1037                         int minPos = 0;
1038
1039                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1040                                 string oligo = it->first;
1041 //                              int length = oligo.length();
1042
1043                                 if(rawSequence.length() < maxLength){
1044                                         success = pdiffs + 100;
1045                                         break;
1046                                 }
1047
1048                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1049                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
1050                                 oligo = alignment->getSeqAAln();
1051                                 string temp = alignment->getSeqBAln();
1052
1053                                 int alnLength = oligo.length();
1054
1055                                 for(int i=oligo.length()-1;i>=0;i--){
1056                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1057                                 }
1058                                 oligo = oligo.substr(0,alnLength);
1059                                 temp = temp.substr(0,alnLength);
1060
1061                                 int newStart=0;
1062                                 int numDiff = countDiffs(oligo, temp);
1063
1064 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1065
1066                                 if(numDiff < minDiff){
1067                                         minDiff = numDiff;
1068                                         minCount = 1;
1069                                         minGroup = it->second;
1070                                         minPos = 0;
1071                                         for(int i=0;i<alnLength;i++){
1072                                                 if(temp[i] != '-'){
1073                                                         minPos++;
1074                                                 }
1075                                         }
1076                                 }
1077                                 else if(numDiff == minDiff){
1078                                         minCount++;
1079                                 }
1080
1081                         }
1082
1083                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
1084                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
1085                         else{                                                                                                   //use the best match
1086                                 group = minGroup;
1087                                 seq.setUnaligned(rawSequence.substr(minPos));
1088                                 if(qual.getName() != ""){
1089                                         qual.trimQScores(minPos, -1);
1090                                 }
1091                                 success = minDiff;
1092                         }
1093
1094                         if (alignment != NULL) {  delete alignment;  }
1095
1096                 }
1097
1098                 return success;
1099
1100         }
1101         catch(exception& e) {
1102                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
1103                 exit(1);
1104         }
1105 }
1106
1107 //***************************************************************************************************************
1108
1109 bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
1110         try {
1111                 string rawSequence = seq.getUnaligned();
1112                 bool success = 0;       //guilty until proven innocent
1113
1114                 for(int i=0;i<numRPrimers;i++){
1115                         string oligo = revPrimer[i];
1116
1117                         if(rawSequence.length() < oligo.length()){
1118                                 success = 0;
1119                                 break;
1120                         }
1121
1122                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
1123                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
1124                                 if(qual.getName() != ""){
1125                                         qual.trimQScores(-1, rawSequence.length()-oligo.length());
1126                                 }
1127                                 success = 1;
1128                                 break;
1129                         }
1130                 }
1131                 return success;
1132
1133         }
1134         catch(exception& e) {
1135                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
1136                 exit(1);
1137         }
1138 }
1139
1140 //***************************************************************************************************************
1141
1142 bool TrimSeqsCommand::cullLength(Sequence& seq){
1143         try {
1144
1145                 int length = seq.getNumBases();
1146                 bool success = 0;       //guilty until proven innocent
1147
1148                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
1149                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
1150                 else                                                                                            {       success = 0;    }
1151
1152                 return success;
1153
1154         }
1155         catch(exception& e) {
1156                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
1157                 exit(1);
1158         }
1159
1160 }
1161
1162 //***************************************************************************************************************
1163
1164 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
1165         try {
1166                 int longHomoP = seq.getLongHomoPolymer();
1167                 bool success = 0;       //guilty until proven innocent
1168
1169                 if(longHomoP <= maxHomoP){      success = 1;    }
1170                 else                                    {       success = 0;    }
1171
1172                 return success;
1173         }
1174         catch(exception& e) {
1175                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
1176                 exit(1);
1177         }
1178
1179 }
1180
1181 //***************************************************************************************************************
1182
1183 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
1184         try {
1185                 int numNs = seq.getAmbigBases();
1186                 bool success = 0;       //guilty until proven innocent
1187
1188                 if(numNs <= maxAmbig)   {       success = 1;    }
1189                 else                                    {       success = 0;    }
1190
1191                 return success;
1192         }
1193         catch(exception& e) {
1194                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
1195                 exit(1);
1196         }
1197
1198 }
1199
1200 //***************************************************************************************************************
1201
1202 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1203         try {
1204                 bool success = 1;
1205                 int length = oligo.length();
1206
1207                 for(int i=0;i<length;i++){
1208
1209                         if(oligo[i] != seq[i]){
1210                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1211                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1212                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1213                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1214                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1215                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1216                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1217                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1218                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1219                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1220                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1221                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1222
1223                                 if(success == 0)        {       break;   }
1224                         }
1225                         else{
1226                                 success = 1;
1227                         }
1228                 }
1229
1230                 return success;
1231         }
1232         catch(exception& e) {
1233                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1234                 exit(1);
1235         }
1236
1237 }
1238 //***************************************************************************************************************
1239
1240 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1241         try {
1242
1243                 int length = oligo.length();
1244                 int countDiffs = 0;
1245
1246                 for(int i=0;i<length;i++){
1247
1248                         if(oligo[i] != seq[i]){
1249                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1250                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1251                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1252                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1253                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1254                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1255                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1256                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1257                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1258                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1259                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1260                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1261                         }
1262
1263                 }
1264
1265                 return countDiffs;
1266         }
1267         catch(exception& e) {
1268                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1269                 exit(1);
1270         }
1271
1272 }
1273 //***************************************************************************************************************
1274
1275 //bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1276 //      try {
1277 //
1278 //              string rawSequence = seq.getUnaligned();
1279 //              int seqLength = seq.getNumBases();
1280 //              bool success = 0;       //guilty until proven innocent
1281 //              string name;
1282 //
1283 //              qFile >> name;
1284 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1285 //
1286 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1287 //
1288 //              int score;
1289 //              int end = seqLength;
1290 //
1291 //              for(int i=0;i<seqLength;i++){
1292 //                      qFile >> score;
1293 //
1294 //                      if(score < qThreshold){
1295 //                              end = i;
1296 //                              break;
1297 //                      }
1298 //              }
1299 //              for(int i=end+1;i<seqLength;i++){
1300 //                      qFile >> score;
1301 //              }
1302 //
1303 //              seq.setUnaligned(rawSequence.substr(0,end));
1304 //
1305 //              return 1;
1306 //      }
1307 //      catch(exception& e) {
1308 //              m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1309 //              exit(1);
1310 //      }
1311 //}
1312
1313 //***************************************************************************************************************
1314
1315 //bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1316 //      try {
1317 //              string rawSequence = seq.getUnaligned();
1318 //              int seqLength = seq.getNumBases();
1319 //              bool success = 0;       //guilty until proven innocent
1320 //              string name;
1321 //
1322 //              qFile >> name;
1323 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1324 //
1325 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1326 //
1327 //              float score;
1328 //              float average = 0;
1329 //
1330 //              for(int i=0;i<seqLength;i++){
1331 //                      qFile >> score;
1332 //                      average += score;
1333 //              }
1334 //              average /= seqLength;
1335 //
1336 //              if(average >= qAverage) {       success = 1;    }
1337 //              else                                    {       success = 0;    }
1338 //
1339 //              return success;
1340 //      }
1341 //      catch(exception& e) {
1342 //              m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1343 //              exit(1);
1344 //      }
1345 //}
1346
1347 //***************************************************************************************************************