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 vector<string> TrimSeqsCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "flip", "oligos", "maxambig", "maxhomop", "group","minlength", "maxlength", "qfile",
  17                                                                         "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "TrimSeqsCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 TrimSeqsCommand::TrimSeqsCommand(){
  28         try {
  29                 abort = true;
  30                 //initialize outputTypes
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["qual"] = tempOutNames;
  34                 outputTypes["group"] = tempOutNames;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> TrimSeqsCommand::getRequiredParameters(){
  43         try {
  44                 string Array[] =  {"fasta"};
  45                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "TrimSeqsCommand", "getRequiredParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> TrimSeqsCommand::getRequiredFiles(){
  55         try {
  56                 vector<string> myArray;
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "TrimSeqsCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 //***************************************************************************************************************
  65
  66 TrimSeqsCommand::TrimSeqsCommand(string option)  {
  67         try {
  68
  69                 abort = false;
  70                 comboStarts = 0;
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; }
  74
  75                 else {
  76                         //valid paramters for this command
  77                         string AlignArray[] =  {"fasta", "flip", "group","oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
  78                                                                         "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
  79
  80                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  81
  82                         OptionParser parser(option);
  83                         map<string,string> parameters = parser.getParameters();
  84
  85                         ValidParameters validParameter;
  86                         map<string,string>::iterator it;
  87
  88                         //check to make sure all parameters are valid for command
  89                         for (it = parameters.begin(); it != parameters.end(); it++) {
  90                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  91                         }
  92
  93                         //initialize outputTypes
  94                         vector<string> tempOutNames;
  95                         outputTypes["fasta"] = tempOutNames;
  96                         outputTypes["qual"] = tempOutNames;
  97                         outputTypes["group"] = tempOutNames;
  98
  99                         //if the user changes the input directory command factory will send this info to us in the output parameter
 100                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 101                         if (inputDir == "not found"){   inputDir = "";          }
 102                         else {
 103                                 string path;
 104                                 it = parameters.find("fasta");
 105                                 //user has given a template file
 106                                 if(it != parameters.end()){
 107                                         path = m->hasPath(it->second);
 108                                         //if the user has not given a path then, add inputdir. else leave path alone.
 109                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 110                                 }
 111
 112                                 it = parameters.find("oligos");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 118                                 }
 119
 120                                 it = parameters.find("qfile");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 126                                 }
 127
 128                                 it = parameters.find("group");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 134                                 }
 135                         }
 136
 137
 138                         //check for required parameters
 139                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 140                         if (fastaFile == "not found") { m->mothurOut("fasta is a required parameter for the trim.seqs command."); m->mothurOutEndLine(); abort = true; }
 141                         else if (fastaFile == "not open") { abort = true; }
 142
 143                         //if the user changes the output directory command factory will send this info to us in the output parameter
 144                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 145                                 outputDir = "";
 146                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 147                         }
 148
 149
 150                         //check for optional parameter and set defaults
 151                         // ...at some point should added some additional type checking...
 152                         string temp;
 153                         temp = validParameter.validFile(parameters, "flip", false);
 154                         if (temp == "not found"){       flip = 0;       }
 155                         else if(m->isTrue(temp))        {       flip = 1;       }
 156
 157                         temp = validParameter.validFile(parameters, "oligos", true);
 158                         if (temp == "not found"){       oligoFile = "";         }
 159                         else if(temp == "not open"){    abort = true;   }
 160                         else                                    {       oligoFile = temp;               }
 161
 162                         temp = validParameter.validFile(parameters, "group", true);
 163                         if (temp == "not found"){       groupfile = "";         }
 164                         else if(temp == "not open"){    abort = true;   }
 165                         else                                    {       groupfile = temp;               }
 166
 167                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
 168                         convert(temp, maxAmbig);
 169
 170                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "0"; }
 171                         convert(temp, maxHomoP);
 172
 173                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "0"; }
 174                         convert(temp, minLength);
 175
 176                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "0"; }
 177                         convert(temp, maxLength);
 178
 179                         temp = validParameter.validFile(parameters, "bdiffs", false);           if (temp == "not found") { temp = "0"; }
 180                         convert(temp, bdiffs);
 181
 182                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found") { temp = "0"; }
 183                         convert(temp, pdiffs);
 184
 185                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs;  temp = toString(tempTotal); }
 186                         convert(temp, tdiffs);
 187
 188                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs;       }
 189
 190                         temp = validParameter.validFile(parameters, "qfile", true);
 191                         if (temp == "not found")        {       qFileName = "";         }
 192                         else if(temp == "not open")     {       abort = true;           }
 193                         else                                            {       qFileName = temp;       }
 194
 195                         temp = validParameter.validFile(parameters, "qthreshold", false);       if (temp == "not found") { temp = "0"; }
 196                         convert(temp, qThreshold);
 197
 198                         temp = validParameter.validFile(parameters, "qtrim", false);            if (temp == "not found") { temp = "F"; }
 199                         qtrim = m->isTrue(temp);
 200
 201                         temp = validParameter.validFile(parameters, "rollaverage", false);      if (temp == "not found") { temp = "0"; }
 202                         convert(temp, qRollAverage);
 203
 204                         temp = validParameter.validFile(parameters, "qwindowaverage", false);if (temp == "not found") { temp = "0"; }
 205                         convert(temp, qWindowAverage);
 206
 207                         temp = validParameter.validFile(parameters, "qwindowsize", false);      if (temp == "not found") { temp = "100"; }
 208                         convert(temp, qWindowSize);
 209
 210                         temp = validParameter.validFile(parameters, "qstepsize", false);                if (temp == "not found") { temp = "10"; }
 211                         convert(temp, qWindowStep);
 212
 213                         temp = validParameter.validFile(parameters, "qaverage", false);         if (temp == "not found") { temp = "0"; }
 214                         convert(temp, qAverage);
 215
 216                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found") { temp = "F"; }
 217                         allFiles = m->isTrue(temp);
 218
 219                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found") { temp = "1"; }
 220                         convert(temp, processors);
 221
 222                         if ((oligoFile != "") && (groupfile != "")) {
 223                                 m->mothurOut("You given both a oligos file and a groupfile, only one is allowed."); m->mothurOutEndLine(); abort = true;
 224                         }
 225
 226
 227                         if(allFiles && (oligoFile == "") && (groupfile == "")){
 228                                 m->mothurOut("You selected allfiles, but didn't enter an oligos or group file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 229                         }
 230                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 231                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 232                                 qAverage=0;
 233                                 qThreshold=0;
 234                         }
 235                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 236                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 237                                 abort = true;
 238                         }
 239                 }
 240
 241         }
 242         catch(exception& e) {
 243                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 244                 exit(1);
 245         }
 246 }
 247 //**********************************************************************************************************************
 248
 249 void TrimSeqsCommand::help(){
 250         try {
 251                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 252                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, group, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim and allfiles.\n");
 253                 m->mothurOut("The fasta parameter is required.\n");
 254                 m->mothurOut("The group parameter allows you to enter a group file for your fasta file.\n");
 255                 m->mothurOut("The flip parameter will output the reverse compliment of your trimmed sequence. The default is false.\n");
 256                 m->mothurOut("The oligos parameter .... The default is "".\n");
 257                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 258                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 259                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 260                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 261                 m->mothurOut("The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs.\n");
 262                 m->mothurOut("The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n");
 263                 m->mothurOut("The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n");
 264                 m->mothurOut("The qfile parameter .....\n");
 265                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 266                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 267                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 268                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 269                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 270                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 271                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 272                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 273                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 274                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 275
 276         }
 277         catch(exception& e) {
 278                 m->errorOut(e, "TrimSeqsCommand", "help");
 279                 exit(1);
 280         }
 281 }
 282
 283
 284 //***************************************************************************************************************
 285
 286 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 287
 288 //***************************************************************************************************************
 289
 290 int TrimSeqsCommand::execute(){
 291         try{
 292
 293                 if (abort == true) { return 0; }
 294
 295                 numFPrimers = 0;  //this needs to be initialized
 296                 numRPrimers = 0;
 297                 vector<string> fastaFileNames;
 298                 vector<string> qualFileNames;
 299
 300                 string trimSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.fasta";
 301                 outputNames.push_back(trimSeqFile); outputTypes["fasta"].push_back(trimSeqFile);
 302                 string scrapSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.fasta";
 303                 outputNames.push_back(scrapSeqFile); outputTypes["fasta"].push_back(scrapSeqFile);
 304                 string trimQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.qual";
 305                 string scrapQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.qual";
 306                 if (qFileName != "") {  outputNames.push_back(trimQualFile); outputNames.push_back(scrapQualFile);  outputTypes["qual"].push_back(trimQualFile); outputTypes["qual"].push_back(scrapQualFile); }
 307                 string groupFile = "";
 308                 if (groupfile == "") { groupFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "groups"; }
 309                 else{
 310                         groupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "trim.groups";
 311                         outputNames.push_back(groupFile); outputTypes["group"].push_back(groupFile);
 312                         groupMap = new GroupMap(groupfile);
 313                         groupMap->readMap();
 314
 315                         if(allFiles){
 316                                 for (int i = 0; i < groupMap->namesOfGroups.size(); i++) {
 317                                         groupToIndex[groupMap->namesOfGroups[i]] = i;
 318                                         groupVector.push_back(groupMap->namesOfGroups[i]);
 319                                         fastaFileNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) +  groupMap->namesOfGroups[i] + ".fasta"));
 320
 321                                         //we append later, so we want to clear file
 322                                         ofstream outRemove;
 323                                         m->openOutputFile(fastaFileNames[i], outRemove);
 324                                         outRemove.close();
 325                                         if(qFileName != ""){
 326                                                 qualFileNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) +  groupMap->namesOfGroups[i] + ".qual"));
 327                                                 ofstream outRemove2;
 328                                                 m->openOutputFile(qualFileNames[i], outRemove2);
 329                                                 outRemove2.close();
 330                                         }
 331                                 }
 332                         }
 333                         comboStarts = fastaFileNames.size()-1;
 334                 }
 335
 336                 if(oligoFile != ""){
 337                         outputNames.push_back(groupFile); outputTypes["group"].push_back(groupFile);
 338                         getOligos(fastaFileNames, qualFileNames);
 339                 }
 340
 341                 vector<unsigned long int> fastaFilePos;
 342                 vector<unsigned long int> qFilePos;
 343
 344                 setLines(fastaFile, qFileName, fastaFilePos, qFilePos);
 345
 346                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 347                         lines.push_back(new linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 348                         if (qFileName != "") {  qLines.push_back(new linePair(qFilePos[i], qFilePos[(i+1)]));  }
 349                 }
 350                 if(qFileName == "")     {       qLines = lines; } //files with duds
 351
 352                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 353                                 if(processors == 1){
 354                                         driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qLines[0]);
 355                                 }else{
 356                                         createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames);
 357                                 }
 358                 #else
 359                                 driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qLines[0]);
 360                 #endif
 361
 362                 if (m->control_pressed) {  return 0; }
 363
 364                 for(int i=0;i<fastaFileNames.size();i++){
 365
 366                         if (m->isBlank(fastaFileNames[i])) {  remove(fastaFileNames[i].c_str());        }
 367                         else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
 368                         else {
 369                                 ifstream inFASTA;
 370                                 string seqName;
 371                                 m->openInputFile(fastaFileNames[i], inFASTA);
 372                                 ofstream outGroups;
 373                                 string outGroupFilename = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "groups";
 374                                 m->openOutputFile(outGroupFilename, outGroups);
 375                                 outputNames.push_back(outGroupFilename); outputTypes["group"].push_back(outGroupFilename);
 376
 377                                 string thisGroup = "";
 378                                 if (i > comboStarts) {
 379                                         map<string, int>::iterator itCombo;
 380                                         for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 381                                                 if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 382                                         }
 383                                 }else{ thisGroup = groupVector[i]; }
 384
 385                                 while(!inFASTA.eof()){
 386                                         if(inFASTA.get() == '>'){
 387                                                 inFASTA >> seqName;
 388                                                 outGroups << seqName << '\t' << thisGroup << endl;
 389                                         }
 390                                         while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 391                                 }
 392                                 outGroups.close();
 393                                 inFASTA.close();
 394                         }
 395                 }
 396
 397                 if(qFileName != ""){
 398                         for(int i=0;i<qualFileNames.size();i++){
 399                                 if (m->isBlank(qualFileNames[i])) {  remove(qualFileNames[i].c_str());  }
 400                                 else if (filesToRemove.count(qualFileNames[i]) > 0) {  remove(qualFileNames[i].c_str()); }
 401                                 else {
 402                                         ifstream inQual;
 403                                         string seqName;
 404                                         m->openInputFile(qualFileNames[i], inQual);
 405 //                                      ofstream outGroups;
 406 //
 407 //                                      string thisGroup = "";
 408 //                                      if (i > comboStarts) {
 409 //                                              map<string, int>::iterator itCombo;
 410 //                                              for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 411 //                                                      if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 412 //                                              }
 413 //                                      }
 414 //                                      else{ thisGroup = groupVector[i]; }
 415
 416                                         inQual.close();
 417                                 }
 418                         }
 419                 }
 420
 421
 422                 if (m->control_pressed) {
 423                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 424                         return 0;
 425                 }
 426
 427                 m->mothurOutEndLine();
 428                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 429                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 430                 m->mothurOutEndLine();
 431
 432                 return 0;
 433
 434         }
 435         catch(exception& e) {
 436                 m->errorOut(e, "TrimSeqsCommand", "execute");
 437                 exit(1);
 438         }
 439 }
 440
 441 /**************************************************************************************/
 442
 443 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) {
 444
 445         try {
 446
 447                 ofstream outFASTA;
 448                 int able = m->openOutputFile(trimFile, outFASTA);
 449
 450                 ofstream scrapFASTA;
 451                 m->openOutputFile(scrapFile, scrapFASTA);
 452
 453                 ofstream outQual;
 454                 ofstream scrapQual;
 455                 if(qFileName != ""){
 456                         m->openOutputFile(trimQFile, outQual);
 457                         m->openOutputFile(scrapQFile, scrapQual);
 458                 }
 459
 460                 ofstream outGroups;
 461
 462                 if (oligoFile != "") {
 463                         m->openOutputFile(groupFile, outGroups);
 464                 }
 465
 466                 ifstream inFASTA;
 467                 m->openInputFile(filename, inFASTA);
 468                 inFASTA.seekg(line->start);
 469
 470                 ifstream qFile;
 471                 if(qFileName != "")     {       m->openInputFile(qFileName, qFile);     qFile.seekg(qline->start);  }
 472
 473                 bool done = false;
 474                 int count = 0;
 475
 476                 while (!done) {
 477
 478                         if (m->control_pressed) {
 479                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close();
 480                                 if (oligoFile != "") {   outGroups.close();   }
 481
 482                                 if(qFileName != ""){
 483                                         qFile.close();
 484                                 }
 485                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 486
 487                                 return 0;
 488                         }
 489
 490                         int success = 1;
 491
 492
 493                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 494
 495                         QualityScores currQual;
 496                         if(qFileName != ""){
 497                                 currQual = QualityScores(qFile, currSeq.getNumBases());  m->gobble(qFile);
 498                         }
 499
 500                         string origSeq = currSeq.getUnaligned();
 501                         if (origSeq != "") {
 502                                 int groupBar, groupPrime;
 503                                 string trashCode = "";
 504                                 int currentSeqsDiffs = 0;
 505
 506                                 if(qFileName != ""){
 507                                         if(qThreshold != 0)                     {       success = currQual.stripQualThreshold(currSeq, qThreshold);                     }
 508                                         else if(qAverage != 0)          {       success = currQual.cullQualAverage(currSeq, qAverage);                          }
 509                                         else if(qRollAverage != 0)      {       success = currQual.stripQualRollingAverage(currSeq, qRollAverage);      }
 510                                         else if(qWindowAverage != 0){   success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage);   }
 511
 512                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 513                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 514                                         }
 515
 516                                         if(!success)                            {       trashCode += 'q';       }
 517                                 }
 518
 519                                 if(barcodes.size() != 0){
 520                                         success = stripBarcode(currSeq, currQual, groupBar);
 521                                         if(success > bdiffs)            {       trashCode += 'b';       }
 522                                         else{ currentSeqsDiffs += success;  }
 523                                 }
 524
 525                                 if(numFPrimers != 0){
 526                                         success = stripForward(currSeq, currQual, groupPrime);
 527                                         if(success > pdiffs)            {       trashCode += 'f';       }
 528                                         else{ currentSeqsDiffs += success;  }
 529                                 }
 530
 531                                 if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
 532
 533                                 if(numRPrimers != 0){
 534                                         success = stripReverse(currSeq, currQual);
 535                                         if(!success)                            {       trashCode += 'r';       }
 536                                 }
 537
 538                                 if(minLength > 0 || maxLength > 0){
 539                                         success = cullLength(currSeq);
 540                                         if(!success)                            {       trashCode += 'l';       }
 541                                 }
 542                                 if(maxHomoP > 0){
 543                                         success = cullHomoP(currSeq);
 544                                         if(!success)                            {       trashCode += 'h';       }
 545                                 }
 546                                 if(maxAmbig != -1){
 547                                         success = cullAmbigs(currSeq);
 548                                         if(!success)                            {       trashCode += 'n';       }
 549                                 }
 550
 551                                 if(flip){       currSeq.reverseComplement();            currQual.flipQScores(); }               // should go last
 552
 553                                 if(trashCode.length() == 0){
 554                                         currSeq.setAligned(currSeq.getUnaligned());
 555                                         currSeq.printSequence(outFASTA);
 556                                         currQual.printQScores(outQual);
 557
 558                                         if(barcodes.size() != 0){
 559                                                 string thisGroup = groupVector[groupBar];
 560                                                 int indexToFastaFile = groupBar;
 561                                                 if (primers.size() != 0){
 562                                                         //does this primer have a group
 563                                                         if (groupVector[groupPrime] != "") {
 564                                                                 thisGroup += "." + groupVector[groupPrime];
 565                                                                 indexToFastaFile = combos[thisGroup];
 566                                                         }
 567                                                 }
 568                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 569                                                 if(allFiles){
 570                                                         ofstream outTemp;
 571                                                         m->openOutputFileAppend(fastaNames[indexToFastaFile], outTemp);
 572                                                         //currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 573                                                         currSeq.printSequence(outTemp);
 574                                                         outTemp.close();
 575
 576                                                         if(qFileName != ""){
 577                                                                 //currQual.printQScores(*qualFileNames[indexToFastaFile]);
 578                                                                 ofstream outTemp2;
 579                                                                 m->openOutputFileAppend(qualNames[indexToFastaFile], outTemp2);
 580                                                                 currQual.printQScores(outTemp2);
 581                                                                 outTemp2.close();
 582                                                         }
 583                                                 }
 584                                         }
 585
 586                                         if (groupfile != "") {
 587                                                 string thisGroup = groupMap->getGroup(currSeq.getName());
 588
 589                                                 if (thisGroup != "not found") {
 590                                                         outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 591                                                         if (allFiles) {
 592                                                                 ofstream outTemp;
 593                                                                 m->openOutputFileAppend(fastaNames[groupToIndex[thisGroup]], outTemp);
 594                                                                 currSeq.printSequence(outTemp);
 595                                                                 outTemp.close();
 596                                                                 if(qFileName != ""){
 597                                                                         ofstream outTemp2;
 598                                                                         m->openOutputFileAppend(qualNames[groupToIndex[thisGroup]], outTemp2);
 599                                                                         currQual.printQScores(outTemp2);
 600                                                                         outTemp2.close();
 601                                                                 }
 602                                                         }
 603                                                 }else{
 604                                                         m->mothurOut(currSeq.getName() + " is not in your groupfile, adding to group XXX."); m->mothurOutEndLine();
 605                                                         outGroups << currSeq.getName() << '\t' << "XXX" << endl;
 606                                                         if (allFiles) {
 607                                                                 m->mothurOut("[ERROR]: " + currSeq.getName() + " will not be added to any .group.fasta or .group.qual file."); m->mothurOutEndLine();
 608                                                         }
 609                                                 }
 610                                         }
 611                                 }
 612                                 else{
 613                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 614                                         currSeq.setUnaligned(origSeq);
 615                                         currSeq.setAligned(origSeq);
 616                                         currSeq.printSequence(scrapFASTA);
 617                                         currQual.printQScores(scrapQual);
 618                                 }
 619                                 count++;
 620                         }
 621
 622                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 623                                 unsigned long int pos = inFASTA.tellg();
 624                                 if ((pos == -1) || (pos >= line->end)) { break; }
 625                         #else
 626                                 if (inFASTA.eof()) { break; }
 627                         #endif
 628
 629                         //report progress
 630                         if((count) % 1000 == 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 631
 632                 }
 633                 //report progress
 634                 if((count) % 1000 != 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 635
 636
 637                 inFASTA.close();
 638                 outFASTA.close();
 639                 scrapFASTA.close();
 640                 if (oligoFile != "") {   outGroups.close();   }
 641                 if(qFileName != "")     {       qFile.close();  scrapQual.close(); outQual.close();     }
 642
 643                 return count;
 644         }
 645         catch(exception& e) {
 646                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 647                 exit(1);
 648         }
 649 }
 650
 651 /**************************************************************************************************/
 652
 653 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
 654         try {
 655 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 656                 int process = 1;
 657                 int exitCommand = 1;
 658                 processIDS.clear();
 659
 660                 //loop through and create all the processes you want
 661                 while (process != processors) {
 662                         int pid = fork();
 663
 664                         if (pid > 0) {
 665                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 666                                 process++;
 667                         }else if (pid == 0){
 668                                 for (int i = 0; i < fastaNames.size(); i++) {
 669                                         fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
 670                                         //clear old file if it exists
 671                                         ofstream temp;
 672                                         m->openOutputFile(fastaNames[i], temp);
 673                                         temp.close();
 674                                         if(qFileName != ""){
 675                                                 qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
 676                                                 //clear old file if it exists
 677                                                 ofstream temp2;
 678                                                 m->openOutputFile(qualNames[i], temp2);
 679                                                 temp2.close();
 680                                         }
 681                                 }
 682
 683                                 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]);
 684                                 exit(0);
 685                         }else {
 686                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 687                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 688                                 exit(0);
 689                         }
 690                 }
 691
 692                 //parent do my part
 693                 for (int i = 0; i < fastaNames.size(); i++) {
 694                         //clear old file if it exists
 695                         ofstream temp;
 696                         m->openOutputFile(fastaNames[i], temp);
 697                         temp.close();
 698                         if(qFileName != ""){
 699                                 //clear old file if it exists
 700                                 ofstream temp2;
 701                                 m->openOutputFile(qualNames[i], temp2);
 702                                 temp2.close();
 703                         }
 704                 }
 705
 706                 driverCreateTrim(filename, qFileName, trimFile, scrapFile, trimQFile, scrapQFile, groupFile, fastaNames, qualNames, lines[0], qLines[0]);
 707
 708
 709                 //force parent to wait until all the processes are done
 710                 for (int i=0;i<processIDS.size();i++) {
 711                         int temp = processIDS[i];
 712                         wait(&temp);
 713                 }
 714
 715                 //append files
 716                 for(int i=0;i<processIDS.size();i++){
 717                         m->mothurOut("Appending files from process " + processIDS[i]); m->mothurOutEndLine();
 718
 719                         m->appendFiles((trimFile + toString(processIDS[i]) + ".temp"), trimFile);
 720                         remove((trimFile + toString(processIDS[i]) + ".temp").c_str());
 721                         m->appendFiles((scrapFile + toString(processIDS[i]) + ".temp"), scrapFile);
 722                         remove((scrapFile + toString(processIDS[i]) + ".temp").c_str());
 723
 724                         m->mothurOut("Done with fasta files"); m->mothurOutEndLine();
 725
 726                         if(qFileName != ""){
 727                                 m->appendFiles((trimQFile + toString(processIDS[i]) + ".temp"), trimQFile);
 728                                 remove((trimQFile + toString(processIDS[i]) + ".temp").c_str());
 729                                 m->appendFiles((scrapQFile + toString(processIDS[i]) + ".temp"), scrapQFile);
 730                                 remove((scrapQFile + toString(processIDS[i]) + ".temp").c_str());
 731
 732                                 m->mothurOut("Done with quality files"); m->mothurOutEndLine();
 733                         }
 734
 735                         m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 736                         remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 737
 738                         m->mothurOut("Done with group file"); m->mothurOutEndLine();
 739
 740                         for (int j = 0; j < fastaNames.size(); j++) {
 741                                 m->appendFiles((fastaNames[j] + toString(processIDS[i]) + ".temp"), fastaNames[j]);
 742                                 remove((fastaNames[j] + toString(processIDS[i]) + ".temp").c_str());
 743                         }
 744
 745                         if(qFileName != ""){
 746                                 for (int j = 0; j < qualNames.size(); j++) {
 747                                         m->appendFiles((qualNames[j] + toString(processIDS[i]) + ".temp"), qualNames[j]);
 748                                         remove((qualNames[j] + toString(processIDS[i]) + ".temp").c_str());
 749                                 }
 750                         }
 751
 752                         if (allFiles) { m->mothurOut("Done with allfiles"); m->mothurOutEndLine(); }
 753                 }
 754
 755                 return exitCommand;
 756 #endif
 757         }
 758         catch(exception& e) {
 759                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 760                 exit(1);
 761         }
 762 }
 763
 764 /**************************************************************************************************/
 765
 766 int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
 767         try {
 768
 769                 //set file positions for fasta file
 770                 fastaFilePos = m->divideFile(filename, processors);
 771
 772                 if (qfilename == "") { return processors; }
 773
 774                 //get name of first sequence in each chunk
 775                 map<string, int> firstSeqNames;
 776                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 777                         ifstream in;
 778                         m->openInputFile(filename, in);
 779                         in.seekg(fastaFilePos[i]);
 780
 781                         Sequence temp(in);
 782                         firstSeqNames[temp.getName()] = i;
 783
 784                         in.close();
 785                 }
 786
 787                 //seach for filePos of each first name in the qfile and save in qfileFilePos
 788                 ifstream inQual;
 789                 m->openInputFile(qfilename, inQual);
 790
 791                 string input;
 792                 while(!inQual.eof()){
 793                         input = m->getline(inQual);
 794
 795                         if (input.length() != 0) {
 796                                 if(input[0] == '>'){ //this is a sequence name line
 797                                         istringstream nameStream(input);
 798
 799                                         string sname = "";  nameStream >> sname;
 800                                         sname = sname.substr(1);
 801
 802                                         map<string, int>::iterator it = firstSeqNames.find(sname);
 803
 804                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
 805                                                 unsigned long int pos = inQual.tellg();
 806                                                 qfileFilePos.push_back(pos - input.length() - 1);
 807                                                 firstSeqNames.erase(it);
 808                                         }
 809                                 }
 810                         }
 811
 812                         if (firstSeqNames.size() == 0) { break; }
 813                 }
 814                 inQual.close();
 815
 816
 817                 if (firstSeqNames.size() != 0) {
 818                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
 819                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
 820                         }
 821                         qFileName = "";
 822                         return processors;
 823                 }
 824
 825                 //get last file position of qfile
 826                 FILE * pFile;
 827                 unsigned long int size;
 828
 829                 //get num bytes in file
 830                 pFile = fopen (qfilename.c_str(),"rb");
 831                 if (pFile==NULL) perror ("Error opening file");
 832                 else{
 833                         fseek (pFile, 0, SEEK_END);
 834                         size=ftell (pFile);
 835                         fclose (pFile);
 836                 }
 837
 838                 qfileFilePos.push_back(size);
 839
 840                 return processors;
 841         }
 842         catch(exception& e) {
 843                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 844                 exit(1);
 845         }
 846 }
 847 //***************************************************************************************************************
 848
 849 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
 850         try {
 851                 ifstream inOligos;
 852                 m->openInputFile(oligoFile, inOligos);
 853
 854                 ofstream test;
 855
 856                 string type, oligo, group;
 857                 int index=0;
 858                 //int indexPrimer = 0;
 859
 860                 while(!inOligos.eof()){
 861                         inOligos >> type; m->gobble(inOligos);
 862
 863                         if(type[0] == '#'){
 864                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 865                         }
 866                         else{
 867                                 //make type case insensitive
 868                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 869
 870                                 inOligos >> oligo;
 871
 872                                 for(int i=0;i<oligo.length();i++){
 873                                         oligo[i] = toupper(oligo[i]);
 874                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 875                                 }
 876
 877                                 if(type == "FORWARD"){
 878                                         group = "";
 879
 880                                         // get rest of line in case there is a primer name
 881                                         while (!inOligos.eof()) {
 882                                                 char c = inOligos.get();
 883                                                 if (c == 10 || c == 13){        break;  }
 884                                                 else if (c == 32 || c == 9){;} //space or tab
 885                                                 else {  group += c;  }
 886                                         }
 887
 888                                         //check for repeat barcodes
 889                                         map<string, int>::iterator itPrime = primers.find(oligo);
 890                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 891
 892                                                 primers[oligo]=index; index++;
 893                                                 groupVector.push_back(group);
 894
 895                                                 if(allFiles){
 896                                                         outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 897                                                         if(qFileName != ""){
 898                                                                 outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 899                                                         }
 900                                                         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
 901                                                                 filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 902                                                                 if(qFileName != ""){
 903                                                                         filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 904                                                                 }
 905                                                         }else {
 906                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 907                                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 908                                                                 if(qFileName != ""){
 909                                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 910                                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 911                                                                 }
 912                                                         }
 913                                                 }
 914
 915                                 }
 916                                 else if(type == "REVERSE"){
 917                                         Sequence oligoRC("reverse", oligo);
 918                                         oligoRC.reverseComplement();
 919                                         revPrimer.push_back(oligoRC.getUnaligned());
 920                                 }
 921                                 else if(type == "BARCODE"){
 922                                         inOligos >> group;
 923
 924                                         //check for repeat barcodes
 925                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 926                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 927
 928                                                 barcodes[oligo]=index; index++;
 929                                                 groupVector.push_back(group);
 930
 931                                                 if(allFiles){
 932                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 933                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 934                                                         outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 935                                                         if(qFileName != ""){
 936                                                                 outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 937                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 938                                                                 outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 939                                                         }
 940                                                 }
 941
 942                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 943                         }
 944                         m->gobble(inOligos);
 945                 }
 946
 947                 inOligos.close();
 948
 949                 //add in potential combos
 950                 if(allFiles){
 951                         comboStarts = outFASTAVec.size()-1;
 952                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 953                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 954                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 955                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 956                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 957                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 958                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 959
 960                                                 if(qFileName != ""){
 961                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 962                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 963                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 964                                                 }
 965                                         }
 966                                 }
 967                         }
 968                 }
 969
 970                 numFPrimers = primers.size();
 971                 numRPrimers = revPrimer.size();
 972
 973         }
 974         catch(exception& e) {
 975                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 976                 exit(1);
 977         }
 978 }
 979 //***************************************************************************************************************
 980
 981 int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
 982         try {
 983
 984                 string rawSequence = seq.getUnaligned();
 985                 int success = bdiffs + 1;       //guilty until proven innocent
 986
 987                 //can you find the barcode
 988                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 989                         string oligo = it->first;
 990                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 991                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 992                                 break;
 993                         }
 994
 995                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 996                                 group = it->second;
 997                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 998
 999                                 if(qual.getName() != ""){
1000                                         qual.trimQScores(oligo.length(), -1);
1001                                 }
1002
1003                                 success = 0;
1004                                 break;
1005                         }
1006                 }
1007
1008                 //if you found the barcode or if you don't want to allow for diffs
1009 //              cout << success;
1010                 if ((bdiffs == 0) || (success == 0)) { return success;  }
1011
1012                 else { //try aligning and see if you can find it
1013 //                      cout << endl;
1014
1015                         int maxLength = 0;
1016
1017                         Alignment* alignment;
1018                         if (barcodes.size() > 0) {
1019                                 map<string,int>::iterator it=barcodes.begin();
1020
1021                                 for(it;it!=barcodes.end();it++){
1022                                         if(it->first.length() > maxLength){
1023                                                 maxLength = it->first.length();
1024                                         }
1025                                 }
1026                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
1027
1028                         }else{ alignment = NULL; }
1029
1030                         //can you find the barcode
1031                         int minDiff = 1e6;
1032                         int minCount = 1;
1033                         int minGroup = -1;
1034                         int minPos = 0;
1035
1036                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
1037                                 string oligo = it->first;
1038 //                              int length = oligo.length();
1039
1040                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
1041                                         success = bdiffs + 10;
1042                                         break;
1043                                 }
1044
1045                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1046                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
1047                                 oligo = alignment->getSeqAAln();
1048                                 string temp = alignment->getSeqBAln();
1049
1050                                 int alnLength = oligo.length();
1051
1052                                 for(int i=oligo.length()-1;i>=0;i--){
1053                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1054                                 }
1055                                 oligo = oligo.substr(0,alnLength);
1056                                 temp = temp.substr(0,alnLength);
1057
1058                                 int newStart=0;
1059                                 int numDiff = countDiffs(oligo, temp);
1060
1061 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1062
1063                                 if(numDiff < minDiff){
1064                                         minDiff = numDiff;
1065                                         minCount = 1;
1066                                         minGroup = it->second;
1067                                         minPos = 0;
1068                                         for(int i=0;i<alnLength;i++){
1069                                                 if(temp[i] != '-'){
1070                                                         minPos++;
1071                                                 }
1072                                         }
1073                                 }
1074                                 else if(numDiff == minDiff){
1075                                         minCount++;
1076                                 }
1077
1078                         }
1079
1080                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
1081                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
1082                         else{                                                                                                   //use the best match
1083                                 group = minGroup;
1084                                 seq.setUnaligned(rawSequence.substr(minPos));
1085
1086                                 if(qual.getName() != ""){
1087                                         qual.trimQScores(minPos, -1);
1088                                 }
1089                                 success = minDiff;
1090                         }
1091
1092                         if (alignment != NULL) {  delete alignment;  }
1093
1094                 }
1095 //              cout << success << endl;
1096
1097                 return success;
1098
1099         }
1100         catch(exception& e) {
1101                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
1102                 exit(1);
1103         }
1104
1105 }
1106
1107 //***************************************************************************************************************
1108
1109 int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
1110         try {
1111                 string rawSequence = seq.getUnaligned();
1112                 int success = pdiffs + 1;       //guilty until proven innocent
1113
1114                 //can you find the primer
1115                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1116                         string oligo = it->first;
1117                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
1118                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
1119                                 break;
1120                         }
1121
1122                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1123                                 group = it->second;
1124                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1125                                 if(qual.getName() != ""){
1126                                         qual.trimQScores(oligo.length(), -1);
1127
1128                                 }
1129                                 success = 0;
1130                                 break;
1131                         }
1132                 }
1133
1134                 //if you found the barcode or if you don't want to allow for diffs
1135 //              cout << success;
1136                 if ((pdiffs == 0) || (success == 0)) { return success;  }
1137
1138                 else { //try aligning and see if you can find it
1139 //                      cout << endl;
1140
1141                         int maxLength = 0;
1142
1143                         Alignment* alignment;
1144                         if (primers.size() > 0) {
1145                                 map<string,int>::iterator it=primers.begin();
1146
1147                                 for(it;it!=primers.end();it++){
1148                                         if(it->first.length() > maxLength){
1149                                                 maxLength = it->first.length();
1150                                         }
1151                                 }
1152                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
1153
1154                         }else{ alignment = NULL; }
1155
1156                         //can you find the barcode
1157                         int minDiff = 1e6;
1158                         int minCount = 1;
1159                         int minGroup = -1;
1160                         int minPos = 0;
1161
1162                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1163                                 string oligo = it->first;
1164 //                              int length = oligo.length();
1165
1166                                 if(rawSequence.length() < maxLength){
1167                                         success = pdiffs + 100;
1168                                         break;
1169                                 }
1170
1171                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1172                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
1173                                 oligo = alignment->getSeqAAln();
1174                                 string temp = alignment->getSeqBAln();
1175
1176                                 int alnLength = oligo.length();
1177
1178                                 for(int i=oligo.length()-1;i>=0;i--){
1179                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1180                                 }
1181                                 oligo = oligo.substr(0,alnLength);
1182                                 temp = temp.substr(0,alnLength);
1183
1184                                 int newStart=0;
1185                                 int numDiff = countDiffs(oligo, temp);
1186
1187 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1188
1189                                 if(numDiff < minDiff){
1190                                         minDiff = numDiff;
1191                                         minCount = 1;
1192                                         minGroup = it->second;
1193                                         minPos = 0;
1194                                         for(int i=0;i<alnLength;i++){
1195                                                 if(temp[i] != '-'){
1196                                                         minPos++;
1197                                                 }
1198                                         }
1199                                 }
1200                                 else if(numDiff == minDiff){
1201                                         minCount++;
1202                                 }
1203
1204                         }
1205
1206                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
1207                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
1208                         else{                                                                                                   //use the best match
1209                                 group = minGroup;
1210                                 seq.setUnaligned(rawSequence.substr(minPos));
1211                                 if(qual.getName() != ""){
1212                                         qual.trimQScores(minPos, -1);
1213                                 }
1214                                 success = minDiff;
1215                         }
1216
1217                         if (alignment != NULL) {  delete alignment;  }
1218
1219                 }
1220
1221                 return success;
1222
1223         }
1224         catch(exception& e) {
1225                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
1226                 exit(1);
1227         }
1228 }
1229
1230 //***************************************************************************************************************
1231
1232 bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
1233         try {
1234                 string rawSequence = seq.getUnaligned();
1235                 bool success = 0;       //guilty until proven innocent
1236
1237                 for(int i=0;i<numRPrimers;i++){
1238                         string oligo = revPrimer[i];
1239
1240                         if(rawSequence.length() < oligo.length()){
1241                                 success = 0;
1242                                 break;
1243                         }
1244
1245                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
1246                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
1247                                 if(qual.getName() != ""){
1248                                         qual.trimQScores(-1, rawSequence.length()-oligo.length());
1249                                 }
1250                                 success = 1;
1251                                 break;
1252                         }
1253                 }
1254                 return success;
1255
1256         }
1257         catch(exception& e) {
1258                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
1259                 exit(1);
1260         }
1261 }
1262
1263 //***************************************************************************************************************
1264
1265 bool TrimSeqsCommand::cullLength(Sequence& seq){
1266         try {
1267
1268                 int length = seq.getNumBases();
1269                 bool success = 0;       //guilty until proven innocent
1270
1271                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
1272                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
1273                 else                                                                                            {       success = 0;    }
1274
1275                 return success;
1276
1277         }
1278         catch(exception& e) {
1279                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
1280                 exit(1);
1281         }
1282
1283 }
1284
1285 //***************************************************************************************************************
1286
1287 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
1288         try {
1289                 int longHomoP = seq.getLongHomoPolymer();
1290                 bool success = 0;       //guilty until proven innocent
1291
1292                 if(longHomoP <= maxHomoP){      success = 1;    }
1293                 else                                    {       success = 0;    }
1294
1295                 return success;
1296         }
1297         catch(exception& e) {
1298                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
1299                 exit(1);
1300         }
1301
1302 }
1303
1304 //***************************************************************************************************************
1305
1306 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
1307         try {
1308                 int numNs = seq.getAmbigBases();
1309                 bool success = 0;       //guilty until proven innocent
1310
1311                 if(numNs <= maxAmbig)   {       success = 1;    }
1312                 else                                    {       success = 0;    }
1313
1314                 return success;
1315         }
1316         catch(exception& e) {
1317                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
1318                 exit(1);
1319         }
1320
1321 }
1322
1323 //***************************************************************************************************************
1324
1325 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1326         try {
1327                 bool success = 1;
1328                 int length = oligo.length();
1329
1330                 for(int i=0;i<length;i++){
1331
1332                         if(oligo[i] != seq[i]){
1333                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1334                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1335                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1336                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1337                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1338                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1339                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1340                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1341                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1342                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1343                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1344                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1345
1346                                 if(success == 0)        {       break;   }
1347                         }
1348                         else{
1349                                 success = 1;
1350                         }
1351                 }
1352
1353                 return success;
1354         }
1355         catch(exception& e) {
1356                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1357                 exit(1);
1358         }
1359
1360 }
1361 //***************************************************************************************************************
1362
1363 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1364         try {
1365
1366                 int length = oligo.length();
1367                 int countDiffs = 0;
1368
1369                 for(int i=0;i<length;i++){
1370
1371                         if(oligo[i] != seq[i]){
1372                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1373                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1374                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1375                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1376                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1377                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1378                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1379                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1380                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1381                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1382                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1383                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1384                         }
1385
1386                 }
1387
1388                 return countDiffs;
1389         }
1390         catch(exception& e) {
1391                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1392                 exit(1);
1393         }
1394
1395 }
1396 //***************************************************************************************************************
1397
1398 //bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1399 //      try {
1400 //
1401 //              string rawSequence = seq.getUnaligned();
1402 //              int seqLength = seq.getNumBases();
1403 //              bool success = 0;       //guilty until proven innocent
1404 //              string name;
1405 //
1406 //              qFile >> name;
1407 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1408 //
1409 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1410 //
1411 //              int score;
1412 //              int end = seqLength;
1413 //
1414 //              for(int i=0;i<seqLength;i++){
1415 //                      qFile >> score;
1416 //
1417 //                      if(score < qThreshold){
1418 //                              end = i;
1419 //                              break;
1420 //                      }
1421 //              }
1422 //              for(int i=end+1;i<seqLength;i++){
1423 //                      qFile >> score;
1424 //              }
1425 //
1426 //              seq.setUnaligned(rawSequence.substr(0,end));
1427 //
1428 //              return 1;
1429 //      }
1430 //      catch(exception& e) {
1431 //              m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1432 //              exit(1);
1433 //      }
1434 //}
1435
1436 //***************************************************************************************************************
1437
1438 //bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1439 //      try {
1440 //              string rawSequence = seq.getUnaligned();
1441 //              int seqLength = seq.getNumBases();
1442 //              bool success = 0;       //guilty until proven innocent
1443 //              string name;
1444 //
1445 //              qFile >> name;
1446 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1447 //
1448 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1449 //
1450 //              float score;
1451 //              float average = 0;
1452 //
1453 //              for(int i=0;i<seqLength;i++){
1454 //                      qFile >> score;
1455 //                      average += score;
1456 //              }
1457 //              average /= seqLength;
1458 //
1459 //              if(average >= qAverage) {       success = 1;    }
1460 //              else                                    {       success = 0;    }
1461 //
1462 //              return success;
1463 //      }
1464 //      catch(exception& e) {
1465 //              m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1466 //              exit(1);
1467 //      }
1468 //}
1469
1470 //***************************************************************************************************************