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