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