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                 cout << "done with driver " << endl;
 364                 for(int i=0;i<fastaFileNames.size();i++){
 365                         cout << fastaFileNames[i] << endl;
 366
 367                         if (m->isBlank(fastaFileNames[i])) { cout << fastaFileNames[i] << " was blank" << endl; remove(fastaFileNames[i].c_str());      }
 368                         else if (filesToRemove.count(fastaFileNames[i]) > 0) { cout << fastaFileNames[i] << " was on the remove list" << endl; 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                                 cout << thisGroup << '\t' << i  << '\t' << comboStarts << endl;
 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         cout << "done with fastaFileNames " << endl;
 398                 if(qFileName != ""){
 399                         for(int i=0;i<qualFileNames.size();i++){
 400                                 cout << qualFileNames[i] << endl;
 401                                 if (m->isBlank(qualFileNames[i])) { cout << qualFileNames[i] << " was blank" << endl; remove(qualFileNames[i].c_str()); }
 402                                 else if (filesToRemove.count(qualFileNames[i]) > 0) { cout << qualFileNames[i] << " was on the remove list" << endl; 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                 cout << "done with qualFileNames " << endl;
 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                 //vector<ofstream*> fastaFileNames;
 464                 //vector<ofstream*> qualFileNames;
 465
 466                 if (oligoFile != "") {
 467                         m->openOutputFile(groupFile, outGroups);
 468                 }
 469
 470                 ifstream inFASTA;
 471                 m->openInputFile(filename, inFASTA);
 472                 inFASTA.seekg(line->start);
 473
 474                 ifstream qFile;
 475                 if(qFileName != "")     {       m->openInputFile(qFileName, qFile);     qFile.seekg(qline->start);  }
 476
 477                 bool done = false;
 478                 int count = 0;
 479
 480                 while (!done) {
 481
 482                         if (m->control_pressed) {
 483                                 inFASTA.close(); outFASTA.close(); scrapFASTA.close();
 484                                 if (oligoFile != "") {   outGroups.close();   }
 485
 486                                 if(qFileName != ""){
 487                                         qFile.close();
 488                                 }
 489                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 490
 491                                 return 0;
 492                         }
 493
 494                         int success = 1;
 495
 496
 497                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 498
 499                         QualityScores currQual;
 500                         if(qFileName != ""){
 501                                 currQual = QualityScores(qFile, currSeq.getNumBases());  m->gobble(qFile);
 502                         }
 503
 504                         string origSeq = currSeq.getUnaligned();
 505                         if (origSeq != "") {
 506                                 int groupBar, groupPrime;
 507                                 string trashCode = "";
 508                                 int currentSeqsDiffs = 0;
 509
 510                                 if(qFileName != ""){
 511                                         if(qThreshold != 0)                     {       success = currQual.stripQualThreshold(currSeq, qThreshold);                     }
 512                                         else if(qAverage != 0)          {       success = currQual.cullQualAverage(currSeq, qAverage);                          }
 513                                         else if(qRollAverage != 0)      {       success = currQual.stripQualRollingAverage(currSeq, qRollAverage);      }
 514                                         else if(qWindowAverage != 0){   success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage);   }
 515
 516                                         if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
 517                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 518                                         }
 519
 520                                         if(!success)                            {       trashCode += 'q';       }
 521                                 }
 522
 523                                 if(barcodes.size() != 0){
 524                                         success = stripBarcode(currSeq, currQual, groupBar);
 525                                         if(success > bdiffs)            {       trashCode += 'b';       }
 526                                         else{ currentSeqsDiffs += success;  }
 527                                 }
 528
 529                                 if(numFPrimers != 0){
 530                                         success = stripForward(currSeq, currQual, groupPrime);
 531                                         if(success > pdiffs)            {       trashCode += 'f';       }
 532                                         else{ currentSeqsDiffs += success;  }
 533                                 }
 534
 535                                 if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
 536
 537                                 if(numRPrimers != 0){
 538                                         success = stripReverse(currSeq, currQual);
 539                                         if(!success)                            {       trashCode += 'r';       }
 540                                 }
 541
 542                                 if(minLength > 0 || maxLength > 0){
 543                                         success = cullLength(currSeq);
 544                                         if(!success)                            {       trashCode += 'l';       }
 545                                 }
 546                                 if(maxHomoP > 0){
 547                                         success = cullHomoP(currSeq);
 548                                         if(!success)                            {       trashCode += 'h';       }
 549                                 }
 550                                 if(maxAmbig != -1){
 551                                         success = cullAmbigs(currSeq);
 552                                         if(!success)                            {       trashCode += 'n';       }
 553                                 }
 554
 555                                 if(flip){       currSeq.reverseComplement();            currQual.flipQScores(); }               // should go last
 556
 557                                 if(trashCode.length() == 0){
 558                                         currSeq.setAligned(currSeq.getUnaligned());
 559                                         currSeq.printSequence(outFASTA);
 560                                         currQual.printQScores(outQual);
 561
 562                                         if(barcodes.size() != 0){
 563                                                 string thisGroup = groupVector[groupBar];
 564                                                 int indexToFastaFile = groupBar;
 565                                                 if (primers.size() != 0){
 566                                                         //does this primer have a group
 567                                                         if (groupVector[groupPrime] != "") {
 568                                                                 thisGroup += "." + groupVector[groupPrime];
 569                                                                 indexToFastaFile = combos[thisGroup];
 570                                                         }
 571                                                 }
 572                                                 outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 573                                                 if(allFiles){
 574                                                         ofstream outTemp;
 575                                                         m->openOutputFileAppend(fastaNames[indexToFastaFile], outTemp);
 576                                                         //currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
 577                                                         currSeq.printSequence(outTemp);
 578                                                         outTemp.close();
 579
 580                                                         if(qFileName != ""){
 581                                                                 //currQual.printQScores(*qualFileNames[indexToFastaFile]);
 582                                                                 ofstream outTemp2;
 583                                                                 m->openOutputFileAppend(qualNames[indexToFastaFile], outTemp2);
 584                                                                 currQual.printQScores(outTemp2);
 585                                                                 outTemp2.close();
 586                                                         }
 587                                                 }
 588                                         }
 589
 590                                         if (groupfile != "") {
 591                                                 string thisGroup = groupMap->getGroup(currSeq.getName());
 592
 593                                                 if (thisGroup != "not found") {
 594                                                         outGroups << currSeq.getName() << '\t' << thisGroup << endl;
 595                                                         if (allFiles) {
 596                                                                 ofstream outTemp;
 597                                                                 m->openOutputFileAppend(fastaNames[groupToIndex[thisGroup]], outTemp);
 598                                                                 currSeq.printSequence(outTemp);
 599                                                                 outTemp.close();
 600                                                                 if(qFileName != ""){
 601                                                                         ofstream outTemp2;
 602                                                                         m->openOutputFileAppend(qualNames[groupToIndex[thisGroup]], outTemp2);
 603                                                                         currQual.printQScores(outTemp2);
 604                                                                         outTemp2.close();
 605                                                                 }
 606                                                         }
 607                                                 }else{
 608                                                         m->mothurOut(currSeq.getName() + " is not in your groupfile, adding to group XXX."); m->mothurOutEndLine();
 609                                                         outGroups << currSeq.getName() << '\t' << "XXX" << endl;
 610                                                         if (allFiles) {
 611                                                                 m->mothurOut("[ERROR]: " + currSeq.getName() + " will not be added to any .group.fasta or .group.qual file."); m->mothurOutEndLine();
 612                                                         }
 613                                                 }
 614                                         }
 615                                 }
 616                                 else{
 617                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 618                                         currSeq.setUnaligned(origSeq);
 619                                         currSeq.setAligned(origSeq);
 620                                         currSeq.printSequence(scrapFASTA);
 621                                         currQual.printQScores(scrapQual);
 622                                 }
 623                                 count++;
 624                         }
 625
 626                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 627                                 unsigned long int pos = inFASTA.tellg();
 628                                 if ((pos == -1) || (pos >= line->end)) { break; }
 629                         #else
 630                                 if (inFASTA.eof()) { break; }
 631                         #endif
 632
 633                         //report progress
 634                         if((count) % 1000 == 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 635
 636                 }
 637                 //report progress
 638                 if((count) % 1000 != 0){        m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 639
 640
 641                 inFASTA.close();
 642                 outFASTA.close();
 643                 scrapFASTA.close();
 644                 if (oligoFile != "") {   outGroups.close();   }
 645                 if(qFileName != "")     {       qFile.close();  scrapQual.close(); outQual.close();     }
 646
 647                 return count;
 648         }
 649         catch(exception& e) {
 650                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 651                 exit(1);
 652         }
 653 }
 654
 655 /**************************************************************************************************/
 656
 657 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
 658         try {
 659 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 660                 int process = 1;
 661                 int exitCommand = 1;
 662                 processIDS.clear();
 663
 664                 //loop through and create all the processes you want
 665                 while (process != processors) {
 666                         int pid = fork();
 667
 668                         if (pid > 0) {
 669                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 670                                 process++;
 671                         }else if (pid == 0){
 672                                 for (int i = 0; i < fastaNames.size(); i++) {
 673                                         fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
 674                                         //clear old file if it exists
 675                                         ofstream temp;
 676                                         m->openOutputFile(fastaNames[i], temp);
 677                                         temp.close();
 678                                         if(qFileName != ""){
 679                                                 qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
 680                                                 //clear old file if it exists
 681                                                 ofstream temp2;
 682                                                 m->openOutputFile(qualNames[i], temp2);
 683                                                 temp2.close();
 684                                         }
 685                                 }
 686
 687                                 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]);
 688                                 exit(0);
 689                         }else {
 690                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 691                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 692                                 exit(0);
 693                         }
 694                 }
 695
 696                 //parent do my part
 697                 for (int i = 0; i < fastaNames.size(); i++) {
 698                         //clear old file if it exists
 699                         ofstream temp;
 700                         m->openOutputFile(fastaNames[i], temp);
 701                         temp.close();
 702                         if(qFileName != ""){
 703                                 //clear old file if it exists
 704                                 ofstream temp2;
 705                                 m->openOutputFile(qualNames[i], temp2);
 706                                 temp2.close();
 707                         }
 708                 }
 709
 710                 driverCreateTrim(filename, qFileName, trimFile, scrapFile, trimQFile, scrapQFile, groupFile, fastaNames, qualNames, lines[0], qLines[0]);
 711
 712
 713                 //force parent to wait until all the processes are done
 714                 for (int i=0;i<processIDS.size();i++) {
 715                         int temp = processIDS[i];
 716                         wait(&temp);
 717                 }
 718
 719                 //append files
 720                 for(int i=0;i<processIDS.size();i++){
 721                         m->mothurOut("Appending files from process " + processIDS[i]); m->mothurOutEndLine();
 722
 723                         m->appendFiles((trimFile + toString(processIDS[i]) + ".temp"), trimFile);
 724                         remove((trimFile + toString(processIDS[i]) + ".temp").c_str());
 725                         m->appendFiles((scrapFile + toString(processIDS[i]) + ".temp"), scrapFile);
 726                         remove((scrapFile + toString(processIDS[i]) + ".temp").c_str());
 727
 728                         m->mothurOut("Done with fasta files"); m->mothurOutEndLine();
 729
 730                         if(qFileName != ""){
 731                                 m->appendFiles((trimQFile + toString(processIDS[i]) + ".temp"), trimQFile);
 732                                 remove((trimQFile + toString(processIDS[i]) + ".temp").c_str());
 733                                 m->appendFiles((scrapQFile + toString(processIDS[i]) + ".temp"), scrapQFile);
 734                                 remove((scrapQFile + toString(processIDS[i]) + ".temp").c_str());
 735
 736                                 m->mothurOut("Done with quality files"); m->mothurOutEndLine();
 737                         }
 738
 739                         m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 740                         remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 741
 742                         m->mothurOut("Done with group file"); m->mothurOutEndLine();
 743
 744                         for (int j = 0; j < fastaNames.size(); j++) {
 745                                 m->appendFiles((fastaNames[j] + toString(processIDS[i]) + ".temp"), fastaNames[j]);
 746                                 remove((fastaNames[j] + toString(processIDS[i]) + ".temp").c_str());
 747                         }
 748
 749                         if(qFileName != ""){
 750                                 for (int j = 0; j < qualNames.size(); j++) {
 751                                         m->appendFiles((qualNames[j] + toString(processIDS[i]) + ".temp"), qualNames[j]);
 752                                         remove((qualNames[j] + toString(processIDS[i]) + ".temp").c_str());
 753                                 }
 754                         }
 755
 756                         if (allFiles) { m->mothurOut("Done with allfile"); m->mothurOutEndLine(); }
 757                 }
 758
 759                 return exitCommand;
 760 #endif
 761         }
 762         catch(exception& e) {
 763                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 764                 exit(1);
 765         }
 766 }
 767
 768 /**************************************************************************************************/
 769
 770 int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
 771         try {
 772
 773                 //set file positions for fasta file
 774                 fastaFilePos = m->divideFile(filename, processors);
 775
 776                 if (qfilename == "") { return processors; }
 777
 778                 //get name of first sequence in each chunk
 779                 map<string, int> firstSeqNames;
 780                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 781                         ifstream in;
 782                         m->openInputFile(filename, in);
 783                         in.seekg(fastaFilePos[i]);
 784
 785                         Sequence temp(in);
 786                         firstSeqNames[temp.getName()] = i;
 787
 788                         in.close();
 789                 }
 790
 791                 //seach for filePos of each first name in the qfile and save in qfileFilePos
 792                 ifstream inQual;
 793                 m->openInputFile(qfilename, inQual);
 794
 795                 string input;
 796                 while(!inQual.eof()){
 797                         input = m->getline(inQual);
 798
 799                         if (input.length() != 0) {
 800                                 if(input[0] == '>'){ //this is a sequence name line
 801                                         istringstream nameStream(input);
 802
 803                                         string sname = "";  nameStream >> sname;
 804                                         sname = sname.substr(1);
 805
 806                                         map<string, int>::iterator it = firstSeqNames.find(sname);
 807
 808                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
 809                                                 unsigned long int pos = inQual.tellg();
 810                                                 qfileFilePos.push_back(pos - input.length() - 1);
 811                                                 firstSeqNames.erase(it);
 812                                         }
 813                                 }
 814                         }
 815
 816                         if (firstSeqNames.size() == 0) { break; }
 817                 }
 818                 inQual.close();
 819
 820
 821                 if (firstSeqNames.size() != 0) {
 822                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
 823                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
 824                         }
 825                         qFileName = "";
 826                         return processors;
 827                 }
 828
 829                 //get last file position of qfile
 830                 FILE * pFile;
 831                 unsigned long int size;
 832
 833                 //get num bytes in file
 834                 pFile = fopen (qfilename.c_str(),"rb");
 835                 if (pFile==NULL) perror ("Error opening file");
 836                 else{
 837                         fseek (pFile, 0, SEEK_END);
 838                         size=ftell (pFile);
 839                         fclose (pFile);
 840                 }
 841
 842                 qfileFilePos.push_back(size);
 843
 844                 return processors;
 845         }
 846         catch(exception& e) {
 847                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 848                 exit(1);
 849         }
 850 }
 851 //***************************************************************************************************************
 852
 853 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
 854         try {
 855                 ifstream inOligos;
 856                 m->openInputFile(oligoFile, inOligos);
 857
 858                 ofstream test;
 859
 860                 string type, oligo, group;
 861                 int index=0;
 862                 //int indexPrimer = 0;
 863
 864                 while(!inOligos.eof()){
 865                         inOligos >> type;
 866
 867                         if(type[0] == '#'){
 868                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 869                         }
 870                         else{
 871                                 //make type case insensitive
 872                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 873
 874                                 inOligos >> oligo;
 875
 876                                 for(int i=0;i<oligo.length();i++){
 877                                         oligo[i] = toupper(oligo[i]);
 878                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 879                                 }
 880
 881                                 if(type == "FORWARD"){
 882                                         group = "";
 883
 884                                         // get rest of line in case there is a primer name
 885                                         while (!inOligos.eof()) {
 886                                                 char c = inOligos.get();
 887                                                 if (c == 10 || c == 13){        break;  }
 888                                                 else if (c == 32 || c == 9){;} //space or tab
 889                                                 else {  group += c;  }
 890                                         }
 891
 892                                         //check for repeat barcodes
 893                                         map<string, int>::iterator itPrime = primers.find(oligo);
 894                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 895
 896                                         primers[oligo]=index; index++;
 897                                         groupVector.push_back(group);
 898
 899                                         if(allFiles){
 900                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 901                                                 if(qFileName != ""){
 902                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 903                                                 }
 904                                                 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
 905                                                         filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 906                                                         if(qFileName != ""){
 907                                                                 filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 908                                                         }
 909                                                 }else {
 910                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 911                                                         outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 912                                                         if(qFileName != ""){
 913                                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 914                                                                 outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 915                                                         }
 916                                                 }
 917                                         }
 918
 919                                 }
 920                                 else if(type == "REVERSE"){
 921                                         Sequence oligoRC("reverse", oligo);
 922                                         oligoRC.reverseComplement();
 923                                         revPrimer.push_back(oligoRC.getUnaligned());
 924                                 }
 925                                 else if(type == "BARCODE"){
 926                                         inOligos >> group;
 927
 928                                         //check for repeat barcodes
 929                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 930                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 931
 932                                         barcodes[oligo]=index; index++;
 933                                         groupVector.push_back(group);
 934
 935                                         if(allFiles){
 936                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 937                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 938                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
 939                                                 if(qFileName != ""){
 940                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 941                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 942                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
 943                                                 }
 944                                         }
 945                                 }else{  m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();  }
 946                         }
 947                         m->gobble(inOligos);
 948                 }
 949
 950                 inOligos.close();
 951
 952                 //add in potential combos
 953                 if(allFiles){
 954                         comboStarts = outFASTAVec.size()-1;
 955                         for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
 956                                 for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
 957                                         if (groupVector[itPrime->second] != "") { //there is a group for this primer
 958                                                 outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 959                                                 outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 960                                                 outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
 961                                                 combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
 962
 963                                                 if(qFileName != ""){
 964                                                         outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 965                                                         outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 966                                                         outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
 967                                                 }
 968                                         }
 969                                 }
 970                         }
 971                 }
 972
 973                 numFPrimers = primers.size();
 974                 numRPrimers = revPrimer.size();
 975
 976         }
 977         catch(exception& e) {
 978                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 979                 exit(1);
 980         }
 981 }
 982 //***************************************************************************************************************
 983
 984 int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
 985         try {
 986
 987                 string rawSequence = seq.getUnaligned();
 988                 int success = bdiffs + 1;       //guilty until proven innocent
 989
 990                 //can you find the barcode
 991                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 992                         string oligo = it->first;
 993                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 994                                 success = bdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
 995                                 break;
 996                         }
 997
 998                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 999                                 group = it->second;
1000                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1001
1002                                 if(qual.getName() != ""){
1003                                         qual.trimQScores(oligo.length(), -1);
1004                                 }
1005
1006                                 success = 0;
1007                                 break;
1008                         }
1009                 }
1010
1011                 //if you found the barcode or if you don't want to allow for diffs
1012 //              cout << success;
1013                 if ((bdiffs == 0) || (success == 0)) { return success;  }
1014
1015                 else { //try aligning and see if you can find it
1016 //                      cout << endl;
1017
1018                         int maxLength = 0;
1019
1020                         Alignment* alignment;
1021                         if (barcodes.size() > 0) {
1022                                 map<string,int>::iterator it=barcodes.begin();
1023
1024                                 for(it;it!=barcodes.end();it++){
1025                                         if(it->first.length() > maxLength){
1026                                                 maxLength = it->first.length();
1027                                         }
1028                                 }
1029                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+bdiffs+1));
1030
1031                         }else{ alignment = NULL; }
1032
1033                         //can you find the barcode
1034                         int minDiff = 1e6;
1035                         int minCount = 1;
1036                         int minGroup = -1;
1037                         int minPos = 0;
1038
1039                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
1040                                 string oligo = it->first;
1041 //                              int length = oligo.length();
1042
1043                                 if(rawSequence.length() < maxLength){   //let's just assume that the barcodes are the same length
1044                                         success = bdiffs + 10;
1045                                         break;
1046                                 }
1047
1048                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1049                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+bdiffs));
1050                                 oligo = alignment->getSeqAAln();
1051                                 string temp = alignment->getSeqBAln();
1052
1053                                 int alnLength = oligo.length();
1054
1055                                 for(int i=oligo.length()-1;i>=0;i--){
1056                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1057                                 }
1058                                 oligo = oligo.substr(0,alnLength);
1059                                 temp = temp.substr(0,alnLength);
1060
1061                                 int newStart=0;
1062                                 int numDiff = countDiffs(oligo, temp);
1063
1064 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1065
1066                                 if(numDiff < minDiff){
1067                                         minDiff = numDiff;
1068                                         minCount = 1;
1069                                         minGroup = it->second;
1070                                         minPos = 0;
1071                                         for(int i=0;i<alnLength;i++){
1072                                                 if(temp[i] != '-'){
1073                                                         minPos++;
1074                                                 }
1075                                         }
1076                                 }
1077                                 else if(numDiff == minDiff){
1078                                         minCount++;
1079                                 }
1080
1081                         }
1082
1083                         if(minDiff > bdiffs)    {       success = minDiff;              }       //no good matches
1084                         else if(minCount > 1)   {       success = bdiffs + 100; }       //can't tell the difference between multiple barcodes
1085                         else{                                                                                                   //use the best match
1086                                 group = minGroup;
1087                                 seq.setUnaligned(rawSequence.substr(minPos));
1088
1089                                 if(qual.getName() != ""){
1090                                         qual.trimQScores(minPos, -1);
1091                                 }
1092                                 success = minDiff;
1093                         }
1094
1095                         if (alignment != NULL) {  delete alignment;  }
1096
1097                 }
1098 //              cout << success << endl;
1099
1100                 return success;
1101
1102         }
1103         catch(exception& e) {
1104                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
1105                 exit(1);
1106         }
1107
1108 }
1109
1110 //***************************************************************************************************************
1111
1112 int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
1113         try {
1114                 string rawSequence = seq.getUnaligned();
1115                 int success = pdiffs + 1;       //guilty until proven innocent
1116
1117                 //can you find the primer
1118                 for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1119                         string oligo = it->first;
1120                         if(rawSequence.length() < oligo.length()){      //let's just assume that the primers are the same length
1121                                 success = pdiffs + 10;                                  //if the sequence is shorter than the barcode then bail out
1122                                 break;
1123                         }
1124
1125                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
1126                                 group = it->second;
1127                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
1128                                 if(qual.getName() != ""){
1129                                         qual.trimQScores(oligo.length(), -1);
1130
1131                                 }
1132                                 success = 0;
1133                                 break;
1134                         }
1135                 }
1136
1137                 //if you found the barcode or if you don't want to allow for diffs
1138 //              cout << success;
1139                 if ((pdiffs == 0) || (success == 0)) { return success;  }
1140
1141                 else { //try aligning and see if you can find it
1142 //                      cout << endl;
1143
1144                         int maxLength = 0;
1145
1146                         Alignment* alignment;
1147                         if (primers.size() > 0) {
1148                                 map<string,int>::iterator it=primers.begin();
1149
1150                                 for(it;it!=primers.end();it++){
1151                                         if(it->first.length() > maxLength){
1152                                                 maxLength = it->first.length();
1153                                         }
1154                                 }
1155                                 alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (maxLength+pdiffs+1));
1156
1157                         }else{ alignment = NULL; }
1158
1159                         //can you find the barcode
1160                         int minDiff = 1e6;
1161                         int minCount = 1;
1162                         int minGroup = -1;
1163                         int minPos = 0;
1164
1165                         for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
1166                                 string oligo = it->first;
1167 //                              int length = oligo.length();
1168
1169                                 if(rawSequence.length() < maxLength){
1170                                         success = pdiffs + 100;
1171                                         break;
1172                                 }
1173
1174                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
1175                                 alignment->align(oligo, rawSequence.substr(0,oligo.length()+pdiffs));
1176                                 oligo = alignment->getSeqAAln();
1177                                 string temp = alignment->getSeqBAln();
1178
1179                                 int alnLength = oligo.length();
1180
1181                                 for(int i=oligo.length()-1;i>=0;i--){
1182                                         if(oligo[i] != '-'){    alnLength = i+1;        break;  }
1183                                 }
1184                                 oligo = oligo.substr(0,alnLength);
1185                                 temp = temp.substr(0,alnLength);
1186
1187                                 int newStart=0;
1188                                 int numDiff = countDiffs(oligo, temp);
1189
1190 //                              cout << oligo << '\t' << temp << '\t' << numDiff << endl;
1191
1192                                 if(numDiff < minDiff){
1193                                         minDiff = numDiff;
1194                                         minCount = 1;
1195                                         minGroup = it->second;
1196                                         minPos = 0;
1197                                         for(int i=0;i<alnLength;i++){
1198                                                 if(temp[i] != '-'){
1199                                                         minPos++;
1200                                                 }
1201                                         }
1202                                 }
1203                                 else if(numDiff == minDiff){
1204                                         minCount++;
1205                                 }
1206
1207                         }
1208
1209                         if(minDiff > pdiffs)    {       success = minDiff;              }       //no good matches
1210                         else if(minCount > 1)   {       success = pdiffs + 10;  }       //can't tell the difference between multiple primers
1211                         else{                                                                                                   //use the best match
1212                                 group = minGroup;
1213                                 seq.setUnaligned(rawSequence.substr(minPos));
1214                                 if(qual.getName() != ""){
1215                                         qual.trimQScores(minPos, -1);
1216                                 }
1217                                 success = minDiff;
1218                         }
1219
1220                         if (alignment != NULL) {  delete alignment;  }
1221
1222                 }
1223
1224                 return success;
1225
1226         }
1227         catch(exception& e) {
1228                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
1229                 exit(1);
1230         }
1231 }
1232
1233 //***************************************************************************************************************
1234
1235 bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
1236         try {
1237                 string rawSequence = seq.getUnaligned();
1238                 bool success = 0;       //guilty until proven innocent
1239
1240                 for(int i=0;i<numRPrimers;i++){
1241                         string oligo = revPrimer[i];
1242
1243                         if(rawSequence.length() < oligo.length()){
1244                                 success = 0;
1245                                 break;
1246                         }
1247
1248                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
1249                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
1250                                 if(qual.getName() != ""){
1251                                         qual.trimQScores(-1, rawSequence.length()-oligo.length());
1252                                 }
1253                                 success = 1;
1254                                 break;
1255                         }
1256                 }
1257                 return success;
1258
1259         }
1260         catch(exception& e) {
1261                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
1262                 exit(1);
1263         }
1264 }
1265
1266 //***************************************************************************************************************
1267
1268 bool TrimSeqsCommand::cullLength(Sequence& seq){
1269         try {
1270
1271                 int length = seq.getNumBases();
1272                 bool success = 0;       //guilty until proven innocent
1273
1274                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
1275                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
1276                 else                                                                                            {       success = 0;    }
1277
1278                 return success;
1279
1280         }
1281         catch(exception& e) {
1282                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
1283                 exit(1);
1284         }
1285
1286 }
1287
1288 //***************************************************************************************************************
1289
1290 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
1291         try {
1292                 int longHomoP = seq.getLongHomoPolymer();
1293                 bool success = 0;       //guilty until proven innocent
1294
1295                 if(longHomoP <= maxHomoP){      success = 1;    }
1296                 else                                    {       success = 0;    }
1297
1298                 return success;
1299         }
1300         catch(exception& e) {
1301                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
1302                 exit(1);
1303         }
1304
1305 }
1306
1307 //***************************************************************************************************************
1308
1309 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
1310         try {
1311                 int numNs = seq.getAmbigBases();
1312                 bool success = 0;       //guilty until proven innocent
1313
1314                 if(numNs <= maxAmbig)   {       success = 1;    }
1315                 else                                    {       success = 0;    }
1316
1317                 return success;
1318         }
1319         catch(exception& e) {
1320                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
1321                 exit(1);
1322         }
1323
1324 }
1325
1326 //***************************************************************************************************************
1327
1328 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
1329         try {
1330                 bool success = 1;
1331                 int length = oligo.length();
1332
1333                 for(int i=0;i<length;i++){
1334
1335                         if(oligo[i] != seq[i]){
1336                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
1337                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
1338                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
1339                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
1340                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
1341                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
1342                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
1343                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
1344                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1345                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
1346                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
1347                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
1348
1349                                 if(success == 0)        {       break;   }
1350                         }
1351                         else{
1352                                 success = 1;
1353                         }
1354                 }
1355
1356                 return success;
1357         }
1358         catch(exception& e) {
1359                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
1360                 exit(1);
1361         }
1362
1363 }
1364 //***************************************************************************************************************
1365
1366 int TrimSeqsCommand::countDiffs(string oligo, string seq){
1367         try {
1368
1369                 int length = oligo.length();
1370                 int countDiffs = 0;
1371
1372                 for(int i=0;i<length;i++){
1373
1374                         if(oligo[i] != seq[i]){
1375                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C' || oligo[i] == '-' || oligo[i] == '.')      {       countDiffs++;   }
1376                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       countDiffs++;   }
1377                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       countDiffs++;   }
1378                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       countDiffs++;   }
1379                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       countDiffs++;   }
1380                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       countDiffs++;   }
1381                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       countDiffs++;   }
1382                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       countDiffs++;   }
1383                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1384                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       countDiffs++;   }
1385                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       countDiffs++;   }
1386                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       countDiffs++;   }
1387                         }
1388
1389                 }
1390
1391                 return countDiffs;
1392         }
1393         catch(exception& e) {
1394                 m->errorOut(e, "TrimSeqsCommand", "countDiffs");
1395                 exit(1);
1396         }
1397
1398 }
1399 //***************************************************************************************************************
1400
1401 //bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
1402 //      try {
1403 //
1404 //              string rawSequence = seq.getUnaligned();
1405 //              int seqLength = seq.getNumBases();
1406 //              bool success = 0;       //guilty until proven innocent
1407 //              string name;
1408 //
1409 //              qFile >> name;
1410 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1411 //
1412 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1413 //
1414 //              int score;
1415 //              int end = seqLength;
1416 //
1417 //              for(int i=0;i<seqLength;i++){
1418 //                      qFile >> score;
1419 //
1420 //                      if(score < qThreshold){
1421 //                              end = i;
1422 //                              break;
1423 //                      }
1424 //              }
1425 //              for(int i=end+1;i<seqLength;i++){
1426 //                      qFile >> score;
1427 //              }
1428 //
1429 //              seq.setUnaligned(rawSequence.substr(0,end));
1430 //
1431 //              return 1;
1432 //      }
1433 //      catch(exception& e) {
1434 //              m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
1435 //              exit(1);
1436 //      }
1437 //}
1438
1439 //***************************************************************************************************************
1440
1441 //bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
1442 //      try {
1443 //              string rawSequence = seq.getUnaligned();
1444 //              int seqLength = seq.getNumBases();
1445 //              bool success = 0;       //guilty until proven innocent
1446 //              string name;
1447 //
1448 //              qFile >> name;
1449 //              if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
1450 //
1451 //              while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
1452 //
1453 //              float score;
1454 //              float average = 0;
1455 //
1456 //              for(int i=0;i<seqLength;i++){
1457 //                      qFile >> score;
1458 //                      average += score;
1459 //              }
1460 //              average /= seqLength;
1461 //
1462 //              if(average >= qAverage) {       success = 1;    }
1463 //              else                                    {       success = 0;    }
1464 //
1465 //              return success;
1466 //      }
1467 //      catch(exception& e) {
1468 //              m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
1469 //              exit(1);
1470 //      }
1471 //}
1472
1473 //***************************************************************************************************************