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
 474                 for (int i = 0; i < fastaNames.size(); i++) { //clears old file
 475                         ofstream temp;
 476                         m->openOutputFile(fastaNames[i], temp);
 477                         temp.close();
 478                 }
 479                 for (int i = 0; i < qualNames.size(); i++) { //clears old file
 480                         ofstream temp;
 481                         m->openOutputFile(qualNames[i], temp);
 482                         temp.close();
 483                 }
 484
 485
 486                 bool done = false;
 487                 int count = 0;
 488
 489                 while (!done) {
 490
 491                         if (m->control_pressed) {
 492                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close();
 493                                 if (oligoFile != "") {   outGroups.close();   }
 494
 495                                 if(qFileName != ""){
 496                                         qFile.close();
 497                                 }
 498                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 499
 500                                 return 0;
 501                         }
 502
 503                         int success = 1;
 504
 505
 506                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 507
 508                         QualityScores currQual;
 509                         if(qFileName != ""){
 510                                 currQual = QualityScores(qFile, currSeq.getNumBases());  m->gobble(qFile);
 511                         }
 512
 513                         string origSeq = currSeq.getUnaligned();
 514                         if (origSeq != "") {
 515                                 int groupBar, groupPrime;
 516                                 string trashCode = "";
 517                                 int currentSeqsDiffs = 0;
 518
 519                                 if(qFileName != ""){
 520                                         if(qThreshold != 0)                     {       success = currQual.stripQualThreshold(currSeq, qThreshold);                     }
 521                                         else if(qAverage != 0)          {       success = currQual.cullQualAverage(currSeq, qAverage);                          }
 522                                         else if(qRollAverage != 0)      {       success = currQual.stripQualRollingAverage(currSeq, qRollAverage);      }
 523                                         else if(qWindowAverage != 0){   success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage);   }
 524
 525                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 526                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 527                                         }
 528
 529                                         if(!success)                            {       trashCode += 'q';       }
 530                                 }
 531
 532                                 if(barcodes.size() != 0){
 533                                         success = stripBarcode(currSeq, currQual, groupBar);
 534                                         if(success > bdiffs)            {       trashCode += 'b';       }
 535                                         else{ currentSeqsDiffs += success;  }
 536                                 }
 537
 538                                 if(numFPrimers != 0){
 539                                         success = stripForward(currSeq, currQual, groupPrime);
 540                                         if(success > pdiffs)            {       trashCode += 'f';       }
 541                                         else{ currentSeqsDiffs += success;  }
 542                                 }
 543
 544                                 if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
 545
 546                                 if(numRPrimers != 0){
 547                                         success = stripReverse(currSeq, currQual);
 548                                         if(!success)                            {       trashCode += 'r';       }
 549                                 }
 550
 551                                 if(minLength > 0 || maxLength > 0){
 552                                         success = cullLength(currSeq);
 553                                         if(!success)                            {       trashCode += 'l';       }
 554                                 }
 555                                 if(maxHomoP > 0){
 556                                         success = cullHomoP(currSeq);
 557                                         if(!success)                            {       trashCode += 'h';       }
 558                                 }
 559                                 if(maxAmbig != -1){
 560                                         success = cullAmbigs(currSeq);
 561                                         if(!success)                            {       trashCode += 'n';       }
 562                                 }
 563
 564                                 if(flip){       currSeq.reverseComplement();            currQual.flipQScores(); }               // should go last
 565
 566                                 if(trashCode.length() == 0){
 567                                         currSeq.setAligned(currSeq.getUnaligned());
 568                                         currSeq.printSequence(outFASTA);
 569                                         currQual.printQScores(outQual);
 570
 571                                         if(barcodes.size() != 0){
 572                                                 string thisGroup = groupVector[groupBar];
 573                                                 int indexToFastaFile = groupBar;
 574                                                 if (primers.size() != 0){
 575                                                         //does this primer have a group
 576                                                         if (groupVector[groupPrime] != "") {
 577                                                                 thisGroup += "." + groupVector[groupPrime];
 578                                                                 indexToFastaFile = combos[thisGroup];
 579                                                         }
 580                                                 }
 581                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 582                                                 if(allFiles){
 583                                                         ofstream outTemp;
 584                                                         m->openOutputFileAppend(fastaNames[indexToFastaFile], outTemp);
 585                                                         //currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 586                                                         currSeq.printSequence(outTemp);
 587                                                         outTemp.close();
 588
 589                                                         if(qFileName != ""){
 590                                                                 //currQual.printQScores(*qualFileNames[indexToFastaFile]);
 591                                                                 ofstream outTemp2;
 592                                                                 m->openOutputFileAppend(qualNames[indexToFastaFile], outTemp2);
 593                                                                 currQual.printQScores(outTemp2);
 594                                                                 outTemp2.close();
 595                                                         }
 596                                                 }
 597                                         }
 598
 599                                         if (groupfile != "") {
 600                                                 string thisGroup = groupMap->getGroup(currSeq.getName());
 601
 602                                                 if (thisGroup != "not found") {
 603                                                         outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 604                                                         if (allFiles) {
 605                                                                 ofstream outTemp;
 606                                                                 m->openOutputFileAppend(fastaNames[groupToIndex[thisGroup]], outTemp);
 607                                                                 currSeq.printSequence(outTemp);
 608                                                                 outTemp.close();
 609                                                                 if(qFileName != ""){
 610                                                                         ofstream outTemp2;
 611                                                                         m->openOutputFileAppend(qualNames[groupToIndex[thisGroup]], outTemp2);
 612                                                                         currQual.printQScores(outTemp2);
 613                                                                         outTemp2.close();
 614                                                                 }
 615                                                         }
 616                                                 }else{
 617                                                         m->mothurOut(currSeq.getName() + " is not in your groupfile, adding to group XXX."); m->mothurOutEndLine();
 618                                                         outGroups << currSeq.getName() << '\t' << "XXX" << endl;
 619                                                         if (allFiles) {
 620                                                                 m->mothurOut("[ERROR]: " + currSeq.getName() + " will not be added to any .group.fasta or .group.qual file."); m->mothurOutEndLine();
 621                                                         }
 622                                                 }
 623                                         }
 624                                 }
 625                                 else{
 626                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 627                                         currSeq.setUnaligned(origSeq);
 628                                         currSeq.setAligned(origSeq);
 629                                         currSeq.printSequence(scrapFASTA);
 630                                         currQual.printQScores(scrapQual);
 631                                 }
 632                                 count++;
 633                         }
 634
 635                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 636                                 unsigned long int pos = inFASTA.tellg();
 637                                 if ((pos == -1) || (pos >= line->end)) { break; }
 638                         #else
 639                                 if (inFASTA.eof()) { break; }
 640                         #endif
 641
 642                         //report progress
 643                         if((count) % 1000 == 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 644
 645                 }
 646                 //report progress
 647                 if((count) % 1000 != 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 648
 649
 650                 inFASTA.close();
 651                 outFASTA.close();
 652                 scrapFASTA.close();
 653                 if (oligoFile != "") {   outGroups.close();   }
 654                 if(qFileName != "")     {       qFile.close();  scrapQual.close(); outQual.close();     }
 655
 656                 return count;
 657         }
 658         catch(exception& e) {
 659                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 660                 exit(1);
 661         }
 662 }
 663
 664 /**************************************************************************************************/
 665
 666 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
 667         try {
 668 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 669                 int process = 1;
 670                 int exitCommand = 1;
 671                 processIDS.clear();
 672
 673                 //loop through and create all the processes you want
 674                 while (process != processors) {
 675                         int pid = fork();
 676
 677                         if (pid > 0) {
 678                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 679                                 process++;
 680                         }else if (pid == 0){
 681                                 for (int i = 0; i < fastaNames.size(); i++) {
 682                                         fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
 683                                         //clear old file if it exists
 684                                         ofstream temp;
 685                                         m->openOutputFile(fastaNames[i], temp);
 686                                         temp.close();
 687                                         if(qFileName != ""){
 688                                                 qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
 689                                                 //clear old file if it exists
 690                                                 ofstream temp2;
 691                                                 m->openOutputFile(qualNames[i], temp2);
 692                                                 temp2.close();
 693                                         }
 694                                 }
 695
 696                                 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]);
 697                                 exit(0);
 698                         }else {
 699                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 700                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 701                                 exit(0);
 702                         }
 703                 }
 704
 705                 //parent do my part
 706                 for (int i = 0; i < fastaNames.size(); i++) {
 707                         //clear old file if it exists
 708                         ofstream temp;
 709                         m->openOutputFile(fastaNames[i], temp);
 710                         temp.close();
 711                         if(qFileName != ""){
 712                                 //clear old file if it exists
 713                                 ofstream temp2;
 714                                 m->openOutputFile(qualNames[i], temp2);
 715                                 temp2.close();
 716                         }
 717                 }
 718
 719                 driverCreateTrim(filename, qFileName, trimFile, scrapFile, trimQFile, scrapQFile, groupFile, fastaNames, qualNames, lines[0], qLines[0]);
 720
 721
 722                 //force parent to wait until all the processes are done
 723                 for (int i=0;i<processIDS.size();i++) {
 724                         int temp = processIDS[i];
 725                         wait(&temp);
 726                 }
 727
 728                 //append files
 729                 for(int i=0;i<processIDS.size();i++){
 730                         m->mothurOut("Appending files from process " + processIDS[i]); m->mothurOutEndLine();
 731
 732                         m->appendFiles((trimFile + toString(processIDS[i]) + ".temp"), trimFile);
 733                         remove((trimFile + toString(processIDS[i]) + ".temp").c_str());
 734                         m->appendFiles((scrapFile + toString(processIDS[i]) + ".temp"), scrapFile);
 735                         remove((scrapFile + toString(processIDS[i]) + ".temp").c_str());
 736
 737                         m->mothurOut("Done with fasta files"); m->mothurOutEndLine();
 738
 739                         if(qFileName != ""){
 740                                 m->appendFiles((trimQFile + toString(processIDS[i]) + ".temp"), trimQFile);
 741                                 remove((trimQFile + toString(processIDS[i]) + ".temp").c_str());
 742                                 m->appendFiles((scrapQFile + toString(processIDS[i]) + ".temp"), scrapQFile);
 743                                 remove((scrapQFile + toString(processIDS[i]) + ".temp").c_str());
 744
 745                                 m->mothurOut("Done with quality files"); m->mothurOutEndLine();
 746                         }
 747
 748                         m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 749                         remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 750
 751                         m->mothurOut("Done with group file"); m->mothurOutEndLine();
 752
 753                         for (int j = 0; j < fastaNames.size(); j++) {
 754                                 m->appendFiles((fastaNames[j] + toString(processIDS[i]) + ".temp"), fastaNames[j]);
 755                                 remove((fastaNames[j] + toString(processIDS[i]) + ".temp").c_str());
 756                         }
 757
 758                         if(qFileName != ""){
 759                                 for (int j = 0; j < qualNames.size(); j++) {
 760                                         m->appendFiles((qualNames[j] + toString(processIDS[i]) + ".temp"), qualNames[j]);
 761                                         remove((qualNames[j] + toString(processIDS[i]) + ".temp").c_str());
 762                                 }
 763                         }
 764
 765                         if (allFiles) { m->mothurOut("Done with allfiles"); m->mothurOutEndLine(); }
 766                 }
 767
 768                 return exitCommand;
 769 #endif
 770         }
 771         catch(exception& e) {
 772                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 773                 exit(1);
 774         }
 775 }
 776
 777 /**************************************************************************************************/
 778
 779 int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
 780         try {
 781
 782                 //set file positions for fasta file
 783                 fastaFilePos = m->divideFile(filename, processors);
 784
 785                 if (qfilename == "") { return processors; }
 786
 787                 //get name of first sequence in each chunk
 788                 map<string, int> firstSeqNames;
 789                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 790                         ifstream in;
 791                         m->openInputFile(filename, in);
 792                         in.seekg(fastaFilePos[i]);
 793
 794                         Sequence temp(in);
 795                         firstSeqNames[temp.getName()] = i;
 796
 797                         in.close();
 798                 }
 799
 800                 //seach for filePos of each first name in the qfile and save in qfileFilePos
 801                 ifstream inQual;
 802                 m->openInputFile(qfilename, inQual);
 803
 804                 string input;
 805                 while(!inQual.eof()){
 806                         input = m->getline(inQual);
 807
 808                         if (input.length() != 0) {
 809                                 if(input[0] == '>'){ //this is a sequence name line
 810                                         istringstream nameStream(input);
 811
 812                                         string sname = "";  nameStream >> sname;
 813                                         sname = sname.substr(1);
 814
 815                                         map<string, int>::iterator it = firstSeqNames.find(sname);
 816
 817                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
 818                                                 unsigned long int pos = inQual.tellg();
 819                                                 qfileFilePos.push_back(pos - input.length() - 1);
 820                                                 firstSeqNames.erase(it);
 821                                         }
 822                                 }
 823                         }
 824
 825                         if (firstSeqNames.size() == 0) { break; }
 826                 }
 827                 inQual.close();
 828
 829
 830                 if (firstSeqNames.size() != 0) {
 831                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
 832                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
 833                         }
 834                         qFileName = "";
 835                         return processors;
 836                 }
 837
 838                 //get last file position of qfile
 839                 FILE * pFile;
 840                 unsigned long int size;
 841
 842                 //get num bytes in file
 843                 pFile = fopen (qfilename.c_str(),"rb");
 844                 if (pFile==NULL) perror ("Error opening file");
 845                 else{
 846                         fseek (pFile, 0, SEEK_END);
 847                         size=ftell (pFile);
 848                         fclose (pFile);
 849                 }
 850
 851                 qfileFilePos.push_back(size);
 852
 853                 return processors;
 854         }
 855         catch(exception& e) {
 856                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 857                 exit(1);
 858         }
 859 }
 860 //***************************************************************************************************************
 861
 862 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
 863         try {
 864                 ifstream inOligos;
 865                 m->openInputFile(oligoFile, inOligos);
 866
 867                 ofstream test;
 868
 869                 string type, oligo, group;
 870                 int index=0;
 871                 //int indexPrimer = 0;
 872
 873                 while(!inOligos.eof()){
 874                         inOligos >> type; m->gobble(inOligos);
 875
 876                         if(type[0] == '#'){
 877                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 878                         }
 879                         else{
 880                                 //make type case insensitive
 881                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 882
 883                                 inOligos >> oligo;
 884
 885                                 for(int i=0;i<oligo.length();i++){
 886                                         oligo[i] = toupper(oligo[i]);
 887                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 888                                 }
 889
 890                                 if(type == "FORWARD"){
 891                                         group = "";
 892
 893                                         // get rest of line in case there is a primer name
 894                                         while (!inOligos.eof()) {
 895                                                 char c = inOligos.get();
 896                                                 if (c == 10 || c == 13){        break;  }
 897                                                 else if (c == 32 || c == 9){;} //space or tab
 898                                                 else {  group += c;  }
 899                                         }
 900
 901                                         //check for repeat barcodes
 902                                         map<string, int>::iterator itPrime = primers.find(oligo);
 903                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 904
 905                                                 primers[oligo]=index; index++;
 906                                                 groupVector.push_back(group);
 907
 908                                                 if(allFiles){
 909                                                         outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 910                                                         if(qFileName != ""){
 911                                                                 outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 912                                                         }
 913                                                         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
 914                                                                 filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 915                                                                 if(qFileName != ""){
 916                                                                         filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 917                                                                 }
 918                                                         }else {
 919                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 920                                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 921                                                                 if(qFileName != ""){
 922                                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 923                                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 924                                                                 }
 925                                                         }
 926                                                 }
 927
 928                                 }
 929                                 else if(type == "REVERSE"){
 930                                         Sequence oligoRC("reverse", oligo);
 931                                         oligoRC.reverseComplement();
 932                                         revPrimer.push_back(oligoRC.getUnaligned());
 933                                 }
 934                                 else if(type == "BARCODE"){
 935                                         inOligos >> group;
 936
 937                                         //check for repeat barcodes
 938                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 939                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 940
 941                                                 barcodes[oligo]=index; index++;
 942                                                 groupVector.push_back(group);
 943
 944                                                 if(allFiles){
 945                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 946                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 947                                                         outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 948                                                         if(qFileName != ""){
 949                                                                 outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 950                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 951                                                                 outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 952                                                         }
 953                                                 }
 954
 955                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 956                         }
 957                         m->gobble(inOligos);
 958                 }
 959
 960                 inOligos.close();
 961
 962                 //add in potential combos
 963                 if(allFiles){
 964                         comboStarts = outFASTAVec.size()-1;
 965                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 966                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 967                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 968                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 969                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 970                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 971                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 972
 973                                                 if(qFileName != ""){
 974                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 975                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 976                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 977                                                 }
 978                                         }
 979                                 }
 980                         }
 981                 }
 982
 983                 numFPrimers = primers.size();
 984                 numRPrimers = revPrimer.size();
 985
 986         }
 987         catch(exception& e) {
 988                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 989                 exit(1);
 990         }
 991 }
 992 //***************************************************************************************************************
 993
 994 int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
 995         try {
 996
 997                 string rawSequence = seq.getUnaligned();
 998                 int success = bdiffs + 1;       //guilty until proven innocent
 999
1000                 //can you find the barcode
1001                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
1002                         string oligo = it->first;
1003                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
1004                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
1005                                 break;
1006                         }
1007
1008                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1009                                 group = it->second;
1010                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1011
1012                                 if(qual.getName() != ""){
1013                                         qual.trimQScores(oligo.length(), -1);
1014                                 }
1015
1016                                 success = 0;
1017                                 break;
1018                         }
1019                 }
1020
1021                 //if you found the barcode or if you don't want to allow for diffs
1022 //              cout << success;
1023                 if ((bdiffs == 0) || (success == 0)) { return success;  }
1024
1025                 else { //try aligning and see if you can find it
1026 //                      cout << endl;
1027
1028                         int maxLength = 0;
1029
1030                         Alignment* alignment;
1031                         if (barcodes.size() > 0) {
1032                                 map<string,int>::iterator it=barcodes.begin();
1033
1034                                 for(it;it!=barcodes.end();it++){
1035                                         if(it->first.length() > maxLength){
1036                                                 maxLength = it->first.length();
1037                                         }
1038                                 }
1039                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
1040
1041                         }else{ alignment = NULL; }
1042
1043                         //can you find the barcode
1044                         int minDiff = 1e6;
1045                         int minCount = 1;
1046                         int minGroup = -1;
1047                         int minPos = 0;
1048
1049                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
1050                                 string oligo = it->first;
1051 //                              int length = oligo.length();
1052
1053                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
1054                                         success = bdiffs + 10;
1055                                         break;
1056                                 }
1057
1058                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1059                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
1060                                 oligo = alignment->getSeqAAln();
1061                                 string temp = alignment->getSeqBAln();
1062
1063                                 int alnLength = oligo.length();
1064
1065                                 for(int i=oligo.length()-1;i>=0;i--){
1066                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1067                                 }
1068                                 oligo = oligo.substr(0,alnLength);
1069                                 temp = temp.substr(0,alnLength);
1070
1071                                 int newStart=0;
1072                                 int numDiff = countDiffs(oligo, temp);
1073
1074 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1075
1076                                 if(numDiff < minDiff){
1077                                         minDiff = numDiff;
1078                                         minCount = 1;
1079                                         minGroup = it->second;
1080                                         minPos = 0;
1081                                         for(int i=0;i<alnLength;i++){
1082                                                 if(temp[i] != '-'){
1083                                                         minPos++;
1084                                                 }
1085                                         }
1086                                 }
1087                                 else if(numDiff == minDiff){
1088                                         minCount++;
1089                                 }
1090
1091                         }
1092
1093                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
1094                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
1095                         else{                                                                                                   //use the best match
1096                                 group = minGroup;
1097                                 seq.setUnaligned(rawSequence.substr(minPos));
1098
1099                                 if(qual.getName() != ""){
1100                                         qual.trimQScores(minPos, -1);
1101                                 }
1102                                 success = minDiff;
1103                         }
1104
1105                         if (alignment != NULL) {  delete alignment;  }
1106
1107                 }
1108 //              cout << success << endl;
1109
1110                 return success;
1111
1112         }
1113         catch(exception& e) {
1114                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
1115                 exit(1);
1116         }
1117
1118 }
1119
1120 //***************************************************************************************************************
1121
1122 int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
1123         try {
1124                 string rawSequence = seq.getUnaligned();
1125                 int success = pdiffs + 1;       //guilty until proven innocent
1126
1127                 //can you find the primer
1128                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1129                         string oligo = it->first;
1130                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
1131                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
1132                                 break;
1133                         }
1134
1135                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1136                                 group = it->second;
1137                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1138                                 if(qual.getName() != ""){
1139                                         qual.trimQScores(oligo.length(), -1);
1140
1141                                 }
1142                                 success = 0;
1143                                 break;
1144                         }
1145                 }
1146
1147                 //if you found the barcode or if you don't want to allow for diffs
1148 //              cout << success;
1149                 if ((pdiffs == 0) || (success == 0)) { return success;  }
1150
1151                 else { //try aligning and see if you can find it
1152 //                      cout << endl;
1153
1154                         int maxLength = 0;
1155
1156                         Alignment* alignment;
1157                         if (primers.size() > 0) {
1158                                 map<string,int>::iterator it=primers.begin();
1159
1160                                 for(it;it!=primers.end();it++){
1161                                         if(it->first.length() > maxLength){
1162                                                 maxLength = it->first.length();
1163                                         }
1164                                 }
1165                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
1166
1167                         }else{ alignment = NULL; }
1168
1169                         //can you find the barcode
1170                         int minDiff = 1e6;
1171                         int minCount = 1;
1172                         int minGroup = -1;
1173                         int minPos = 0;
1174
1175                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1176                                 string oligo = it->first;
1177 //                              int length = oligo.length();
1178
1179                                 if(rawSequence.length() < maxLength){
1180                                         success = pdiffs + 100;
1181                                         break;
1182                                 }
1183
1184                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1185                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
1186                                 oligo = alignment->getSeqAAln();
1187                                 string temp = alignment->getSeqBAln();
1188
1189                                 int alnLength = oligo.length();
1190
1191                                 for(int i=oligo.length()-1;i>=0;i--){
1192                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1193                                 }
1194                                 oligo = oligo.substr(0,alnLength);
1195                                 temp = temp.substr(0,alnLength);
1196
1197                                 int newStart=0;
1198                                 int numDiff = countDiffs(oligo, temp);
1199
1200 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1201
1202                                 if(numDiff < minDiff){
1203                                         minDiff = numDiff;
1204                                         minCount = 1;
1205                                         minGroup = it->second;
1206                                         minPos = 0;
1207                                         for(int i=0;i<alnLength;i++){
1208                                                 if(temp[i] != '-'){
1209                                                         minPos++;
1210                                                 }
1211                                         }
1212                                 }
1213                                 else if(numDiff == minDiff){
1214                                         minCount++;
1215                                 }
1216
1217                         }
1218
1219                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
1220                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
1221                         else{                                                                                                   //use the best match
1222                                 group = minGroup;
1223                                 seq.setUnaligned(rawSequence.substr(minPos));
1224                                 if(qual.getName() != ""){
1225                                         qual.trimQScores(minPos, -1);
1226                                 }
1227                                 success = minDiff;
1228                         }
1229
1230                         if (alignment != NULL) {  delete alignment;  }
1231
1232                 }
1233
1234                 return success;
1235
1236         }
1237         catch(exception& e) {
1238                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
1239                 exit(1);
1240         }
1241 }
1242
1243 //***************************************************************************************************************
1244
1245 bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
1246         try {
1247                 string rawSequence = seq.getUnaligned();
1248                 bool success = 0;       //guilty until proven innocent
1249
1250                 for(int i=0;i<numRPrimers;i++){
1251                         string oligo = revPrimer[i];
1252
1253                         if(rawSequence.length() < oligo.length()){
1254                                 success = 0;
1255                                 break;
1256                         }
1257
1258                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
1259                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
1260                                 if(qual.getName() != ""){
1261                                         qual.trimQScores(-1, rawSequence.length()-oligo.length());
1262                                 }
1263                                 success = 1;
1264                                 break;
1265                         }
1266                 }
1267                 return success;
1268
1269         }
1270         catch(exception& e) {
1271                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
1272                 exit(1);
1273         }
1274 }
1275
1276 //***************************************************************************************************************
1277
1278 bool TrimSeqsCommand::cullLength(Sequence& seq){
1279         try {
1280
1281                 int length = seq.getNumBases();
1282                 bool success = 0;       //guilty until proven innocent
1283
1284                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
1285                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
1286                 else                                                                                            {       success = 0;    }
1287
1288                 return success;
1289
1290         }
1291         catch(exception& e) {
1292                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
1293                 exit(1);
1294         }
1295
1296 }
1297
1298 //***************************************************************************************************************
1299
1300 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
1301         try {
1302                 int longHomoP = seq.getLongHomoPolymer();
1303                 bool success = 0;       //guilty until proven innocent
1304
1305                 if(longHomoP <= maxHomoP){      success = 1;    }
1306                 else                                    {       success = 0;    }
1307
1308                 return success;
1309         }
1310         catch(exception& e) {
1311                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
1312                 exit(1);
1313         }
1314
1315 }
1316
1317 //***************************************************************************************************************
1318
1319 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
1320         try {
1321                 int numNs = seq.getAmbigBases();
1322                 bool success = 0;       //guilty until proven innocent
1323
1324                 if(numNs <= maxAmbig)   {       success = 1;    }
1325                 else                                    {       success = 0;    }
1326
1327                 return success;
1328         }
1329         catch(exception& e) {
1330                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
1331                 exit(1);
1332         }
1333
1334 }
1335
1336 //***************************************************************************************************************
1337
1338 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1339         try {
1340                 bool success = 1;
1341                 int length = oligo.length();
1342
1343                 for(int i=0;i<length;i++){
1344
1345                         if(oligo[i] != seq[i]){
1346                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1347                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1348                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1349                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1350                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1351                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1352                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1353                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1354                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1355                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1356                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1357                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1358
1359                                 if(success == 0)        {       break;   }
1360                         }
1361                         else{
1362                                 success = 1;
1363                         }
1364                 }
1365
1366                 return success;
1367         }
1368         catch(exception& e) {
1369                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1370                 exit(1);
1371         }
1372
1373 }
1374 //***************************************************************************************************************
1375
1376 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1377         try {
1378
1379                 int length = oligo.length();
1380                 int countDiffs = 0;
1381
1382                 for(int i=0;i<length;i++){
1383
1384                         if(oligo[i] != seq[i]){
1385                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1386                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1387                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1388                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1389                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1390                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1391                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1392                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1393                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1394                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1395                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1396                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1397                         }
1398
1399                 }
1400
1401                 return countDiffs;
1402         }
1403         catch(exception& e) {
1404                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1405                 exit(1);
1406         }
1407
1408 }
1409 //***************************************************************************************************************
1410
1411 //bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1412 //      try {
1413 //
1414 //              string rawSequence = seq.getUnaligned();
1415 //              int seqLength = seq.getNumBases();
1416 //              bool success = 0;       //guilty until proven innocent
1417 //              string name;
1418 //
1419 //              qFile >> name;
1420 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1421 //
1422 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1423 //
1424 //              int score;
1425 //              int end = seqLength;
1426 //
1427 //              for(int i=0;i<seqLength;i++){
1428 //                      qFile >> score;
1429 //
1430 //                      if(score < qThreshold){
1431 //                              end = i;
1432 //                              break;
1433 //                      }
1434 //              }
1435 //              for(int i=end+1;i<seqLength;i++){
1436 //                      qFile >> score;
1437 //              }
1438 //
1439 //              seq.setUnaligned(rawSequence.substr(0,end));
1440 //
1441 //              return 1;
1442 //      }
1443 //      catch(exception& e) {
1444 //              m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1445 //              exit(1);
1446 //      }
1447 //}
1448
1449 //***************************************************************************************************************
1450
1451 //bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1452 //      try {
1453 //              string rawSequence = seq.getUnaligned();
1454 //              int seqLength = seq.getNumBases();
1455 //              bool success = 0;       //guilty until proven innocent
1456 //              string name;
1457 //
1458 //              qFile >> name;
1459 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1460 //
1461 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1462 //
1463 //              float score;
1464 //              float average = 0;
1465 //
1466 //              for(int i=0;i<seqLength;i++){
1467 //                      qFile >> score;
1468 //                      average += score;
1469 //              }
1470 //              average /= seqLength;
1471 //
1472 //              if(average >= qAverage) {       success = 1;    }
1473 //              else                                    {       success = 0;    }
1474 //
1475 //              return success;
1476 //      }
1477 //      catch(exception& e) {
1478 //              m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1479 //              exit(1);
1480 //      }
1481 //}
1482
1483 //***************************************************************************************************************