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                 cout << fastaFileNames.size() << '\t' << qualFileNames.size() << endl;
 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                         cout << fastaFileNames[i] << endl;
 366
 367                         if (m->isBlank(fastaFileNames[i])) {  remove(fastaFileNames[i].c_str());        }
 368                         else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
 369                         else {
 370                                 ifstream inFASTA;
 371                                 string seqName;
 372                                 m->openInputFile(fastaFileNames[i], inFASTA);
 373                                 ofstream outGroups;
 374                                 string outGroupFilename = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "groups";
 375                                 m->openOutputFile(outGroupFilename, outGroups);
 376                                 outputNames.push_back(outGroupFilename); outputTypes["group"].push_back(outGroupFilename);
 377
 378                                 string thisGroup = "";
 379                                 if (i > comboStarts) {
 380                                         map<string, int>::iterator itCombo;
 381                                         for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 382                                                 if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 383                                         }
 384                                 }else{ thisGroup = groupVector[i]; }
 385
 386                                 while(!inFASTA.eof()){
 387                                         if(inFASTA.get() == '>'){
 388                                                 inFASTA >> seqName;
 389                                                 outGroups << seqName << '\t' << thisGroup << endl;
 390                                         }
 391                                         while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 392                                 }
 393                                 outGroups.close();
 394                                 inFASTA.close();
 395                         }
 396                 }
 397
 398                 if(qFileName != ""){
 399                         for(int i=0;i<qualFileNames.size();i++){
 400                                 cout << qualFileNames[i] << endl;
 401                                 if (m->isBlank(qualFileNames[i])) {  remove(qualFileNames[i].c_str());  }
 402                                 else if (filesToRemove.count(qualFileNames[i]) > 0) {  remove(qualFileNames[i].c_str()); }
 403                                 else {
 404                                         ifstream inQual;
 405                                         string seqName;
 406                                         m->openInputFile(qualFileNames[i], inQual);
 407 //                                      ofstream outGroups;
 408 //
 409 //                                      string thisGroup = "";
 410 //                                      if (i > comboStarts) {
 411 //                                              map<string, int>::iterator itCombo;
 412 //                                              for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
 413 //                                                      if(itCombo->second == i){       thisGroup = itCombo->first;     combos.erase(itCombo);  break;  }
 414 //                                              }
 415 //                                      }
 416 //                                      else{ thisGroup = groupVector[i]; }
 417
 418                                         inQual.close();
 419                                 }
 420                         }
 421                 }
 422
 423
 424                 if (m->control_pressed) {
 425                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 426                         return 0;
 427                 }
 428
 429                 m->mothurOutEndLine();
 430                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 431                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 432                 m->mothurOutEndLine();
 433
 434                 return 0;
 435
 436         }
 437         catch(exception& e) {
 438                 m->errorOut(e, "TrimSeqsCommand", "execute");
 439                 exit(1);
 440         }
 441 }
 442
 443 /**************************************************************************************/
 444
 445 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) {
 446
 447         try {
 448
 449                 ofstream outFASTA;
 450                 int able = m->openOutputFile(trimFile, outFASTA);
 451
 452                 ofstream scrapFASTA;
 453                 m->openOutputFile(scrapFile, scrapFASTA);
 454
 455                 ofstream outQual;
 456                 ofstream scrapQual;
 457                 if(qFileName != ""){
 458                         m->openOutputFile(trimQFile, outQual);
 459                         m->openOutputFile(scrapQFile, scrapQual);
 460                 }
 461
 462                 ofstream outGroups;
 463
 464                 if (oligoFile != "") {
 465                         m->openOutputFile(groupFile, outGroups);
 466                 }
 467
 468                 ifstream inFASTA;
 469                 m->openInputFile(filename, inFASTA);
 470                 inFASTA.seekg(line->start);
 471
 472                 ifstream qFile;
 473                 if(qFileName != "")     {       m->openInputFile(qFileName, qFile);     qFile.seekg(qline->start);  }
 474
 475                 bool done = false;
 476                 int count = 0;
 477
 478                 while (!done) {
 479
 480                         if (m->control_pressed) {
 481                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close();
 482                                 if (oligoFile != "") {   outGroups.close();   }
 483
 484                                 if(qFileName != ""){
 485                                         qFile.close();
 486                                 }
 487                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 488
 489                                 return 0;
 490                         }
 491
 492                         int success = 1;
 493
 494
 495                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 496
 497                         QualityScores currQual;
 498                         if(qFileName != ""){
 499                                 currQual = QualityScores(qFile, currSeq.getNumBases());  m->gobble(qFile);
 500                         }
 501
 502                         string origSeq = currSeq.getUnaligned();
 503                         if (origSeq != "") {
 504                                 int groupBar, groupPrime;
 505                                 string trashCode = "";
 506                                 int currentSeqsDiffs = 0;
 507
 508                                 if(qFileName != ""){
 509                                         if(qThreshold != 0)                     {       success = currQual.stripQualThreshold(currSeq, qThreshold);                     }
 510                                         else if(qAverage != 0)          {       success = currQual.cullQualAverage(currSeq, qAverage);                          }
 511                                         else if(qRollAverage != 0)      {       success = currQual.stripQualRollingAverage(currSeq, qRollAverage);      }
 512                                         else if(qWindowAverage != 0){   success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage);   }
 513
 514                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 515                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 516                                         }
 517
 518                                         if(!success)                            {       trashCode += 'q';       }
 519                                 }
 520
 521                                 if(barcodes.size() != 0){
 522                                         success = stripBarcode(currSeq, currQual, groupBar);
 523                                         if(success > bdiffs)            {       trashCode += 'b';       }
 524                                         else{ currentSeqsDiffs += success;  }
 525                                 }
 526
 527                                 if(numFPrimers != 0){
 528                                         success = stripForward(currSeq, currQual, groupPrime);
 529                                         if(success > pdiffs)            {       trashCode += 'f';       }
 530                                         else{ currentSeqsDiffs += success;  }
 531                                 }
 532
 533                                 if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
 534
 535                                 if(numRPrimers != 0){
 536                                         success = stripReverse(currSeq, currQual);
 537                                         if(!success)                            {       trashCode += 'r';       }
 538                                 }
 539
 540                                 if(minLength > 0 || maxLength > 0){
 541                                         success = cullLength(currSeq);
 542                                         if(!success)                            {       trashCode += 'l';       }
 543                                 }
 544                                 if(maxHomoP > 0){
 545                                         success = cullHomoP(currSeq);
 546                                         if(!success)                            {       trashCode += 'h';       }
 547                                 }
 548                                 if(maxAmbig != -1){
 549                                         success = cullAmbigs(currSeq);
 550                                         if(!success)                            {       trashCode += 'n';       }
 551                                 }
 552
 553                                 if(flip){       currSeq.reverseComplement();            currQual.flipQScores(); }               // should go last
 554
 555                                 if(trashCode.length() == 0){
 556                                         currSeq.setAligned(currSeq.getUnaligned());
 557                                         currSeq.printSequence(outFASTA);
 558                                         currQual.printQScores(outQual);
 559
 560                                         if(barcodes.size() != 0){
 561                                                 string thisGroup = groupVector[groupBar];
 562                                                 int indexToFastaFile = groupBar;
 563                                                 if (primers.size() != 0){
 564                                                         //does this primer have a group
 565                                                         if (groupVector[groupPrime] != "") {
 566                                                                 thisGroup += "." + groupVector[groupPrime];
 567                                                                 indexToFastaFile = combos[thisGroup];
 568                                                         }
 569                                                 }
 570                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 571                                                 if(allFiles){
 572                                                         ofstream outTemp;
 573                                                         m->openOutputFileAppend(fastaNames[indexToFastaFile], outTemp);
 574                                                         //currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 575                                                         currSeq.printSequence(outTemp);
 576                                                         outTemp.close();
 577
 578                                                         if(qFileName != ""){
 579                                                                 //currQual.printQScores(*qualFileNames[indexToFastaFile]);
 580                                                                 ofstream outTemp2;
 581                                                                 m->openOutputFileAppend(qualNames[indexToFastaFile], outTemp2);
 582                                                                 currQual.printQScores(outTemp2);
 583                                                                 outTemp2.close();
 584                                                         }
 585                                                 }
 586                                         }
 587
 588                                         if (groupfile != "") {
 589                                                 string thisGroup = groupMap->getGroup(currSeq.getName());
 590
 591                                                 if (thisGroup != "not found") {
 592                                                         outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 593                                                         if (allFiles) {
 594                                                                 ofstream outTemp;
 595                                                                 m->openOutputFileAppend(fastaNames[groupToIndex[thisGroup]], outTemp);
 596                                                                 currSeq.printSequence(outTemp);
 597                                                                 outTemp.close();
 598                                                                 if(qFileName != ""){
 599                                                                         ofstream outTemp2;
 600                                                                         m->openOutputFileAppend(qualNames[groupToIndex[thisGroup]], outTemp2);
 601                                                                         currQual.printQScores(outTemp2);
 602                                                                         outTemp2.close();
 603                                                                 }
 604                                                         }
 605                                                 }else{
 606                                                         m->mothurOut(currSeq.getName() + " is not in your groupfile, adding to group XXX."); m->mothurOutEndLine();
 607                                                         outGroups << currSeq.getName() << '\t' << "XXX" << endl;
 608                                                         if (allFiles) {
 609                                                                 m->mothurOut("[ERROR]: " + currSeq.getName() + " will not be added to any .group.fasta or .group.qual file."); m->mothurOutEndLine();
 610                                                         }
 611                                                 }
 612                                         }
 613                                 }
 614                                 else{
 615                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 616                                         currSeq.setUnaligned(origSeq);
 617                                         currSeq.setAligned(origSeq);
 618                                         currSeq.printSequence(scrapFASTA);
 619                                         currQual.printQScores(scrapQual);
 620                                 }
 621                                 count++;
 622                         }
 623
 624                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 625                                 unsigned long int pos = inFASTA.tellg();
 626                                 if ((pos == -1) || (pos >= line->end)) { break; }
 627                         #else
 628                                 if (inFASTA.eof()) { break; }
 629                         #endif
 630
 631                         //report progress
 632                         if((count) % 1000 == 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 633
 634                 }
 635                 //report progress
 636                 if((count) % 1000 != 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 637
 638
 639                 inFASTA.close();
 640                 outFASTA.close();
 641                 scrapFASTA.close();
 642                 if (oligoFile != "") {   outGroups.close();   }
 643                 if(qFileName != "")     {       qFile.close();  scrapQual.close(); outQual.close();     }
 644
 645                 return count;
 646         }
 647         catch(exception& e) {
 648                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 649                 exit(1);
 650         }
 651 }
 652
 653 /**************************************************************************************************/
 654
 655 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
 656         try {
 657 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 658                 int process = 1;
 659                 int exitCommand = 1;
 660                 processIDS.clear();
 661
 662                 //loop through and create all the processes you want
 663                 while (process != processors) {
 664                         int pid = fork();
 665
 666                         if (pid > 0) {
 667                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 668                                 process++;
 669                         }else if (pid == 0){
 670                                 for (int i = 0; i < fastaNames.size(); i++) {
 671                                         fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
 672                                         //clear old file if it exists
 673                                         ofstream temp;
 674                                         m->openOutputFile(fastaNames[i], temp);
 675                                         temp.close();
 676                                         if(qFileName != ""){
 677                                                 qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
 678                                                 //clear old file if it exists
 679                                                 ofstream temp2;
 680                                                 m->openOutputFile(qualNames[i], temp2);
 681                                                 temp2.close();
 682                                         }
 683                                 }
 684
 685                                 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]);
 686                                 exit(0);
 687                         }else {
 688                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 689                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 690                                 exit(0);
 691                         }
 692                 }
 693
 694                 //parent do my part
 695                 for (int i = 0; i < fastaNames.size(); i++) {
 696                         //clear old file if it exists
 697                         ofstream temp;
 698                         m->openOutputFile(fastaNames[i], temp);
 699                         temp.close();
 700                         if(qFileName != ""){
 701                                 //clear old file if it exists
 702                                 ofstream temp2;
 703                                 m->openOutputFile(qualNames[i], temp2);
 704                                 temp2.close();
 705                         }
 706                 }
 707
 708                 driverCreateTrim(filename, qFileName, trimFile, scrapFile, trimQFile, scrapQFile, groupFile, fastaNames, qualNames, lines[0], qLines[0]);
 709
 710
 711                 //force parent to wait until all the processes are done
 712                 for (int i=0;i<processIDS.size();i++) {
 713                         int temp = processIDS[i];
 714                         wait(&temp);
 715                 }
 716
 717                 //append files
 718                 for(int i=0;i<processIDS.size();i++){
 719                         m->mothurOut("Appending files from process " + processIDS[i]); m->mothurOutEndLine();
 720
 721                         m->appendFiles((trimFile + toString(processIDS[i]) + ".temp"), trimFile);
 722                         remove((trimFile + toString(processIDS[i]) + ".temp").c_str());
 723                         m->appendFiles((scrapFile + toString(processIDS[i]) + ".temp"), scrapFile);
 724                         remove((scrapFile + toString(processIDS[i]) + ".temp").c_str());
 725
 726                         m->mothurOut("Done with fasta files"); m->mothurOutEndLine();
 727
 728                         if(qFileName != ""){
 729                                 m->appendFiles((trimQFile + toString(processIDS[i]) + ".temp"), trimQFile);
 730                                 remove((trimQFile + toString(processIDS[i]) + ".temp").c_str());
 731                                 m->appendFiles((scrapQFile + toString(processIDS[i]) + ".temp"), scrapQFile);
 732                                 remove((scrapQFile + toString(processIDS[i]) + ".temp").c_str());
 733
 734                                 m->mothurOut("Done with quality files"); m->mothurOutEndLine();
 735                         }
 736
 737                         m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 738                         remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 739
 740                         m->mothurOut("Done with group file"); m->mothurOutEndLine();
 741
 742                         for (int j = 0; j < fastaNames.size(); j++) {
 743                                 m->appendFiles((fastaNames[j] + toString(processIDS[i]) + ".temp"), fastaNames[j]);
 744                                 remove((fastaNames[j] + toString(processIDS[i]) + ".temp").c_str());
 745                         }
 746
 747                         if(qFileName != ""){
 748                                 for (int j = 0; j < qualNames.size(); j++) {
 749                                         m->appendFiles((qualNames[j] + toString(processIDS[i]) + ".temp"), qualNames[j]);
 750                                         remove((qualNames[j] + toString(processIDS[i]) + ".temp").c_str());
 751                                 }
 752                         }
 753
 754                         if (allFiles) { m->mothurOut("Done with allfiles"); m->mothurOutEndLine(); }
 755                 }
 756
 757                 return exitCommand;
 758 #endif
 759         }
 760         catch(exception& e) {
 761                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 762                 exit(1);
 763         }
 764 }
 765
 766 /**************************************************************************************************/
 767
 768 int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
 769         try {
 770
 771                 //set file positions for fasta file
 772                 fastaFilePos = m->divideFile(filename, processors);
 773
 774                 if (qfilename == "") { return processors; }
 775
 776                 //get name of first sequence in each chunk
 777                 map<string, int> firstSeqNames;
 778                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 779                         ifstream in;
 780                         m->openInputFile(filename, in);
 781                         in.seekg(fastaFilePos[i]);
 782
 783                         Sequence temp(in);
 784                         firstSeqNames[temp.getName()] = i;
 785
 786                         in.close();
 787                 }
 788
 789                 //seach for filePos of each first name in the qfile and save in qfileFilePos
 790                 ifstream inQual;
 791                 m->openInputFile(qfilename, inQual);
 792
 793                 string input;
 794                 while(!inQual.eof()){
 795                         input = m->getline(inQual);
 796
 797                         if (input.length() != 0) {
 798                                 if(input[0] == '>'){ //this is a sequence name line
 799                                         istringstream nameStream(input);
 800
 801                                         string sname = "";  nameStream >> sname;
 802                                         sname = sname.substr(1);
 803
 804                                         map<string, int>::iterator it = firstSeqNames.find(sname);
 805
 806                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
 807                                                 unsigned long int pos = inQual.tellg();
 808                                                 qfileFilePos.push_back(pos - input.length() - 1);
 809                                                 firstSeqNames.erase(it);
 810                                         }
 811                                 }
 812                         }
 813
 814                         if (firstSeqNames.size() == 0) { break; }
 815                 }
 816                 inQual.close();
 817
 818
 819                 if (firstSeqNames.size() != 0) {
 820                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
 821                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
 822                         }
 823                         qFileName = "";
 824                         return processors;
 825                 }
 826
 827                 //get last file position of qfile
 828                 FILE * pFile;
 829                 unsigned long int size;
 830
 831                 //get num bytes in file
 832                 pFile = fopen (qfilename.c_str(),"rb");
 833                 if (pFile==NULL) perror ("Error opening file");
 834                 else{
 835                         fseek (pFile, 0, SEEK_END);
 836                         size=ftell (pFile);
 837                         fclose (pFile);
 838                 }
 839
 840                 qfileFilePos.push_back(size);
 841
 842                 return processors;
 843         }
 844         catch(exception& e) {
 845                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 846                 exit(1);
 847         }
 848 }
 849 //***************************************************************************************************************
 850
 851 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
 852         try {
 853                 ifstream inOligos;
 854                 m->openInputFile(oligoFile, inOligos);
 855
 856                 ofstream test;
 857
 858                 string type, oligo, group;
 859                 int index=0;
 860                 //int indexPrimer = 0;
 861
 862                 while(!inOligos.eof()){
 863                         inOligos >> type;
 864
 865                         if(type[0] == '#'){
 866                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 867                         }
 868                         else{
 869                                 //make type case insensitive
 870                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 871
 872                                 inOligos >> oligo;
 873
 874                                 for(int i=0;i<oligo.length();i++){
 875                                         oligo[i] = toupper(oligo[i]);
 876                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 877                                 }
 878
 879                                 if(type == "FORWARD"){
 880                                         group = "";
 881
 882                                         // get rest of line in case there is a primer name
 883                                         while (!inOligos.eof()) {
 884                                                 char c = inOligos.get();
 885                                                 if (c == 10 || c == 13){        break;  }
 886                                                 else if (c == 32 || c == 9){;} //space or tab
 887                                                 else {  group += c;  }
 888                                         }
 889
 890                                         //check for repeat barcodes
 891                                         map<string, int>::iterator itPrime = primers.find(oligo);
 892                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 893
 894                                         primers[oligo]=index; index++;
 895                                         groupVector.push_back(group);
 896
 897                                         if(allFiles){
 898                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 899                                                 if(qFileName != ""){
 900                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 901                                                 }
 902                                                 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
 903                                                         filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 904                                                         if(qFileName != ""){
 905                                                                 filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 906                                                         }
 907                                                 }else {
 908                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 909                                                         outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 910                                                         if(qFileName != ""){
 911                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 912                                                                 outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 913                                                         }
 914                                                 }
 915                                         }
 916
 917                                 }
 918                                 else if(type == "REVERSE"){
 919                                         Sequence oligoRC("reverse", oligo);
 920                                         oligoRC.reverseComplement();
 921                                         revPrimer.push_back(oligoRC.getUnaligned());
 922                                 }
 923                                 else if(type == "BARCODE"){
 924                                         inOligos >> group;
 925
 926                                         //check for repeat barcodes
 927                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 928                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 929
 930                                         barcodes[oligo]=index; index++;
 931                                         groupVector.push_back(group);
 932
 933                                         if(allFiles){
 934                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 935                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 936                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 937                                                 if(qFileName != ""){
 938                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 939                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 940                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 941                                                 }
 942                                         }
 943                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 944                         }
 945                         m->gobble(inOligos);
 946                 }
 947
 948                 inOligos.close();
 949
 950                 //add in potential combos
 951                 if(allFiles){
 952                         comboStarts = outFASTAVec.size()-1;
 953                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 954                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 955                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 956                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 957                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 958                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 959                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 960
 961                                                 if(qFileName != ""){
 962                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 963                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 964                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 965                                                 }
 966                                         }
 967                                 }
 968                         }
 969                 }
 970
 971                 numFPrimers = primers.size();
 972                 numRPrimers = revPrimer.size();
 973
 974         }
 975         catch(exception& e) {
 976                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 977                 exit(1);
 978         }
 979 }
 980 //***************************************************************************************************************
 981
 982 int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
 983         try {
 984
 985                 string rawSequence = seq.getUnaligned();
 986                 int success = bdiffs + 1;       //guilty until proven innocent
 987
 988                 //can you find the barcode
 989                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 990                         string oligo = it->first;
 991                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 992                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 993                                 break;
 994                         }
 995
 996                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 997                                 group = it->second;
 998                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 999
1000                                 if(qual.getName() != ""){
1001                                         qual.trimQScores(oligo.length(), -1);
1002                                 }
1003
1004                                 success = 0;
1005                                 break;
1006                         }
1007                 }
1008
1009                 //if you found the barcode or if you don't want to allow for diffs
1010 //              cout << success;
1011                 if ((bdiffs == 0) || (success == 0)) { return success;  }
1012
1013                 else { //try aligning and see if you can find it
1014 //                      cout << endl;
1015
1016                         int maxLength = 0;
1017
1018                         Alignment* alignment;
1019                         if (barcodes.size() > 0) {
1020                                 map<string,int>::iterator it=barcodes.begin();
1021
1022                                 for(it;it!=barcodes.end();it++){
1023                                         if(it->first.length() > maxLength){
1024                                                 maxLength = it->first.length();
1025                                         }
1026                                 }
1027                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
1028
1029                         }else{ alignment = NULL; }
1030
1031                         //can you find the barcode
1032                         int minDiff = 1e6;
1033                         int minCount = 1;
1034                         int minGroup = -1;
1035                         int minPos = 0;
1036
1037                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
1038                                 string oligo = it->first;
1039 //                              int length = oligo.length();
1040
1041                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
1042                                         success = bdiffs + 10;
1043                                         break;
1044                                 }
1045
1046                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1047                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
1048                                 oligo = alignment->getSeqAAln();
1049                                 string temp = alignment->getSeqBAln();
1050
1051                                 int alnLength = oligo.length();
1052
1053                                 for(int i=oligo.length()-1;i>=0;i--){
1054                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1055                                 }
1056                                 oligo = oligo.substr(0,alnLength);
1057                                 temp = temp.substr(0,alnLength);
1058
1059                                 int newStart=0;
1060                                 int numDiff = countDiffs(oligo, temp);
1061
1062 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1063
1064                                 if(numDiff < minDiff){
1065                                         minDiff = numDiff;
1066                                         minCount = 1;
1067                                         minGroup = it->second;
1068                                         minPos = 0;
1069                                         for(int i=0;i<alnLength;i++){
1070                                                 if(temp[i] != '-'){
1071                                                         minPos++;
1072                                                 }
1073                                         }
1074                                 }
1075                                 else if(numDiff == minDiff){
1076                                         minCount++;
1077                                 }
1078
1079                         }
1080
1081                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
1082                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
1083                         else{                                                                                                   //use the best match
1084                                 group = minGroup;
1085                                 seq.setUnaligned(rawSequence.substr(minPos));
1086
1087                                 if(qual.getName() != ""){
1088                                         qual.trimQScores(minPos, -1);
1089                                 }
1090                                 success = minDiff;
1091                         }
1092
1093                         if (alignment != NULL) {  delete alignment;  }
1094
1095                 }
1096 //              cout << success << endl;
1097
1098                 return success;
1099
1100         }
1101         catch(exception& e) {
1102                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
1103                 exit(1);
1104         }
1105
1106 }
1107
1108 //***************************************************************************************************************
1109
1110 int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
1111         try {
1112                 string rawSequence = seq.getUnaligned();
1113                 int success = pdiffs + 1;       //guilty until proven innocent
1114
1115                 //can you find the primer
1116                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1117                         string oligo = it->first;
1118                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
1119                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
1120                                 break;
1121                         }
1122
1123                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1124                                 group = it->second;
1125                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1126                                 if(qual.getName() != ""){
1127                                         qual.trimQScores(oligo.length(), -1);
1128
1129                                 }
1130                                 success = 0;
1131                                 break;
1132                         }
1133                 }
1134
1135                 //if you found the barcode or if you don't want to allow for diffs
1136 //              cout << success;
1137                 if ((pdiffs == 0) || (success == 0)) { return success;  }
1138
1139                 else { //try aligning and see if you can find it
1140 //                      cout << endl;
1141
1142                         int maxLength = 0;
1143
1144                         Alignment* alignment;
1145                         if (primers.size() > 0) {
1146                                 map<string,int>::iterator it=primers.begin();
1147
1148                                 for(it;it!=primers.end();it++){
1149                                         if(it->first.length() > maxLength){
1150                                                 maxLength = it->first.length();
1151                                         }
1152                                 }
1153                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
1154
1155                         }else{ alignment = NULL; }
1156
1157                         //can you find the barcode
1158                         int minDiff = 1e6;
1159                         int minCount = 1;
1160                         int minGroup = -1;
1161                         int minPos = 0;
1162
1163                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1164                                 string oligo = it->first;
1165 //                              int length = oligo.length();
1166
1167                                 if(rawSequence.length() < maxLength){
1168                                         success = pdiffs + 100;
1169                                         break;
1170                                 }
1171
1172                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1173                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
1174                                 oligo = alignment->getSeqAAln();
1175                                 string temp = alignment->getSeqBAln();
1176
1177                                 int alnLength = oligo.length();
1178
1179                                 for(int i=oligo.length()-1;i>=0;i--){
1180                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1181                                 }
1182                                 oligo = oligo.substr(0,alnLength);
1183                                 temp = temp.substr(0,alnLength);
1184
1185                                 int newStart=0;
1186                                 int numDiff = countDiffs(oligo, temp);
1187
1188 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1189
1190                                 if(numDiff < minDiff){
1191                                         minDiff = numDiff;
1192                                         minCount = 1;
1193                                         minGroup = it->second;
1194                                         minPos = 0;
1195                                         for(int i=0;i<alnLength;i++){
1196                                                 if(temp[i] != '-'){
1197                                                         minPos++;
1198                                                 }
1199                                         }
1200                                 }
1201                                 else if(numDiff == minDiff){
1202                                         minCount++;
1203                                 }
1204
1205                         }
1206
1207                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
1208                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
1209                         else{                                                                                                   //use the best match
1210                                 group = minGroup;
1211                                 seq.setUnaligned(rawSequence.substr(minPos));
1212                                 if(qual.getName() != ""){
1213                                         qual.trimQScores(minPos, -1);
1214                                 }
1215                                 success = minDiff;
1216                         }
1217
1218                         if (alignment != NULL) {  delete alignment;  }
1219
1220                 }
1221
1222                 return success;
1223
1224         }
1225         catch(exception& e) {
1226                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
1227                 exit(1);
1228         }
1229 }
1230
1231 //***************************************************************************************************************
1232
1233 bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
1234         try {
1235                 string rawSequence = seq.getUnaligned();
1236                 bool success = 0;       //guilty until proven innocent
1237
1238                 for(int i=0;i<numRPrimers;i++){
1239                         string oligo = revPrimer[i];
1240
1241                         if(rawSequence.length() < oligo.length()){
1242                                 success = 0;
1243                                 break;
1244                         }
1245
1246                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
1247                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
1248                                 if(qual.getName() != ""){
1249                                         qual.trimQScores(-1, rawSequence.length()-oligo.length());
1250                                 }
1251                                 success = 1;
1252                                 break;
1253                         }
1254                 }
1255                 return success;
1256
1257         }
1258         catch(exception& e) {
1259                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
1260                 exit(1);
1261         }
1262 }
1263
1264 //***************************************************************************************************************
1265
1266 bool TrimSeqsCommand::cullLength(Sequence& seq){
1267         try {
1268
1269                 int length = seq.getNumBases();
1270                 bool success = 0;       //guilty until proven innocent
1271
1272                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
1273                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
1274                 else                                                                                            {       success = 0;    }
1275
1276                 return success;
1277
1278         }
1279         catch(exception& e) {
1280                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
1281                 exit(1);
1282         }
1283
1284 }
1285
1286 //***************************************************************************************************************
1287
1288 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
1289         try {
1290                 int longHomoP = seq.getLongHomoPolymer();
1291                 bool success = 0;       //guilty until proven innocent
1292
1293                 if(longHomoP <= maxHomoP){      success = 1;    }
1294                 else                                    {       success = 0;    }
1295
1296                 return success;
1297         }
1298         catch(exception& e) {
1299                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
1300                 exit(1);
1301         }
1302
1303 }
1304
1305 //***************************************************************************************************************
1306
1307 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
1308         try {
1309                 int numNs = seq.getAmbigBases();
1310                 bool success = 0;       //guilty until proven innocent
1311
1312                 if(numNs <= maxAmbig)   {       success = 1;    }
1313                 else                                    {       success = 0;    }
1314
1315                 return success;
1316         }
1317         catch(exception& e) {
1318                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
1319                 exit(1);
1320         }
1321
1322 }
1323
1324 //***************************************************************************************************************
1325
1326 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1327         try {
1328                 bool success = 1;
1329                 int length = oligo.length();
1330
1331                 for(int i=0;i<length;i++){
1332
1333                         if(oligo[i] != seq[i]){
1334                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1335                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1336                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1337                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1338                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1339                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1340                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1341                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1342                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1343                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1344                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1345                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1346
1347                                 if(success == 0)        {       break;   }
1348                         }
1349                         else{
1350                                 success = 1;
1351                         }
1352                 }
1353
1354                 return success;
1355         }
1356         catch(exception& e) {
1357                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1358                 exit(1);
1359         }
1360
1361 }
1362 //***************************************************************************************************************
1363
1364 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1365         try {
1366
1367                 int length = oligo.length();
1368                 int countDiffs = 0;
1369
1370                 for(int i=0;i<length;i++){
1371
1372                         if(oligo[i] != seq[i]){
1373                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1374                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1375                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1376                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1377                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1378                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1379                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1380                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1381                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1382                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1383                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1384                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1385                         }
1386
1387                 }
1388
1389                 return countDiffs;
1390         }
1391         catch(exception& e) {
1392                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1393                 exit(1);
1394         }
1395
1396 }
1397 //***************************************************************************************************************
1398
1399 //bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1400 //      try {
1401 //
1402 //              string rawSequence = seq.getUnaligned();
1403 //              int seqLength = seq.getNumBases();
1404 //              bool success = 0;       //guilty until proven innocent
1405 //              string name;
1406 //
1407 //              qFile >> name;
1408 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1409 //
1410 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1411 //
1412 //              int score;
1413 //              int end = seqLength;
1414 //
1415 //              for(int i=0;i<seqLength;i++){
1416 //                      qFile >> score;
1417 //
1418 //                      if(score < qThreshold){
1419 //                              end = i;
1420 //                              break;
1421 //                      }
1422 //              }
1423 //              for(int i=end+1;i<seqLength;i++){
1424 //                      qFile >> score;
1425 //              }
1426 //
1427 //              seq.setUnaligned(rawSequence.substr(0,end));
1428 //
1429 //              return 1;
1430 //      }
1431 //      catch(exception& e) {
1432 //              m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1433 //              exit(1);
1434 //      }
1435 //}
1436
1437 //***************************************************************************************************************
1438
1439 //bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1440 //      try {
1441 //              string rawSequence = seq.getUnaligned();
1442 //              int seqLength = seq.getNumBases();
1443 //              bool success = 0;       //guilty until proven innocent
1444 //              string name;
1445 //
1446 //              qFile >> name;
1447 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1448 //
1449 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1450 //
1451 //              float score;
1452 //              float average = 0;
1453 //
1454 //              for(int i=0;i<seqLength;i++){
1455 //                      qFile >> score;
1456 //                      average += score;
1457 //              }
1458 //              average /= seqLength;
1459 //
1460 //              if(average >= qAverage) {       success = 1;    }
1461 //              else                                    {       success = 0;    }
1462 //
1463 //              return success;
1464 //      }
1465 //      catch(exception& e) {
1466 //              m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1467 //              exit(1);
1468 //      }
1469 //}
1470
1471 //***************************************************************************************************************