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 #include "nast.hpp"
  13
  14 //***************************************************************************************************************
  15
  16 TrimSeqsCommand::TrimSeqsCommand(string option)  {
  17         try {
  18
  19                 abort = false;
  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", "qaverage", "allfiles", "qtrim","diffs", "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                         //check for optional parameter and set defaults
  85                         // ...at some point should added some additional type checking...
  86                         string temp;
  87                         temp = validParameter.validFile(parameters, "flip", false);
  88                         if (temp == "not found"){       flip = 0;       }
  89                         else if(isTrue(temp))   {       flip = 1;       }
  90
  91                         temp = validParameter.validFile(parameters, "oligos", true);
  92                         if (temp == "not found"){       oligoFile = "";         }
  93                         else if(temp == "not open"){    abort = true;   }
  94                         else                                    {       oligoFile = temp;               }
  95
  96                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
  97                         convert(temp, maxAmbig);
  98
  99                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "0"; }
 100                         convert(temp, maxHomoP);
 101
 102                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "0"; }
 103                         convert(temp, minLength);
 104
 105                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "0"; }
 106                         convert(temp, maxLength);
 107
 108                         temp = validParameter.validFile(parameters, "diffs", false);            if (temp == "not found") { temp = "0"; }
 109                         convert(temp, diffs);
 110
 111                         temp = validParameter.validFile(parameters, "qfile", true);
 112                         if (temp == "not found")        {       qFileName = "";         }
 113                         else if(temp == "not open")     {       abort = true;           }
 114                         else                                            {       qFileName = temp;       }
 115
 116                         temp = validParameter.validFile(parameters, "qthreshold", false);       if (temp == "not found") { temp = "0"; }
 117                         convert(temp, qThreshold);
 118
 119                         temp = validParameter.validFile(parameters, "qtrim", false);    if (temp == "not found") { temp = "F"; }
 120                         qtrim = isTrue(temp);
 121
 122                         temp = validParameter.validFile(parameters, "qaverage", false);         if (temp == "not found") { temp = "0"; }
 123                         convert(temp, qAverage);
 124
 125                         temp = validParameter.validFile(parameters, "allfiles", false);         if (temp == "not found") { temp = "F"; }
 126                         allFiles = isTrue(temp);
 127
 128                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 129                         convert(temp, processors);
 130
 131                         if(allFiles && oligoFile == ""){
 132                                 m->mothurOut("You selected allfiles, but didn't enter an oligos file.  Ignoring the allfiles request."); m->mothurOutEndLine();
 133                         }
 134                         if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
 135                                 m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
 136                                 qAverage=0;
 137                                 qThreshold=0;
 138                         }
 139                         if(!flip && oligoFile=="" && !maxLength && !minLength && (maxAmbig==-1) && !maxHomoP && qFileName == ""){
 140                                 m->mothurOut("You didn't set any options... quiting command."); m->mothurOutEndLine();
 141                                 abort = true;
 142                         }
 143                 }
 144
 145         }
 146         catch(exception& e) {
 147                 m->errorOut(e, "TrimSeqsCommand", "TrimSeqsCommand");
 148                 exit(1);
 149         }
 150 }
 151 //**********************************************************************************************************************
 152
 153 void TrimSeqsCommand::help(){
 154         try {
 155                 m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
 156                 m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim and allfiles.\n");
 157                 m->mothurOut("The fasta parameter is required.\n");
 158                 m->mothurOut("The flip parameter .... The default is 0.\n");
 159                 m->mothurOut("The oligos parameter .... The default is "".\n");
 160                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 161                 m->mothurOut("The maxhomop parameter .... The default is 0.\n");
 162                 m->mothurOut("The minlength parameter .... The default is 0.\n");
 163                 m->mothurOut("The maxlength parameter .... The default is 0.\n");
 164                 m->mothurOut("The diffs parameter .... The default is 0.\n");
 165                 m->mothurOut("The qfile parameter .....\n");
 166                 m->mothurOut("The qthreshold parameter .... The default is 0.\n");
 167                 m->mothurOut("The qaverage parameter .... The default is 0.\n");
 168                 m->mothurOut("The allfiles parameter .... The default is F.\n");
 169                 m->mothurOut("The qtrim parameter .... The default is F.\n");
 170                 m->mothurOut("The trim.seqs command should be in the following format: \n");
 171                 m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig,  \n");
 172                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 173                 m->mothurOut("Example trim.seqs(fasta=abrecovery.fasta, flip=..., oligos=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 174                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 175                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.seqs .\n\n");
 176
 177         }
 178         catch(exception& e) {
 179                 m->errorOut(e, "TrimSeqsCommand", "help");
 180                 exit(1);
 181         }
 182 }
 183
 184
 185 //***************************************************************************************************************
 186
 187 TrimSeqsCommand::~TrimSeqsCommand(){    /*      do nothing      */      }
 188
 189 //***************************************************************************************************************
 190
 191 int TrimSeqsCommand::execute(){
 192         try{
 193
 194                 if (abort == true) { return 0; }
 195
 196                 numFPrimers = 0;  //this needs to be initialized
 197                 numRPrimers = 0;
 198
 199                 string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
 200                 outputNames.push_back(trimSeqFile);
 201                 string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
 202                 outputNames.push_back(scrapSeqFile);
 203                 string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
 204
 205                 vector<string> fastaFileNames;
 206                 if(oligoFile != ""){
 207                         outputNames.push_back(groupFile);
 208                         getOligos(fastaFileNames);
 209                 }
 210
 211                 if(qFileName != "")     {       setLines(qFileName, qLines);    }
 212
 213
 214                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 215                                 if(processors == 1){
 216                                         ifstream inFASTA;
 217                                         openInputFile(fastaFile, inFASTA);
 218                                         int numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 219                                         inFASTA.close();
 220
 221                                         lines.push_back(new linePair(0, numSeqs));
 222
 223                                         driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 224
 225                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 226                                                 rename((fastaFileNames[j] + toString(getpid()) + ".temp").c_str(), fastaFileNames[j].c_str());
 227                                         }
 228
 229                                 }else{
 230                                         setLines(fastaFile, lines);
 231                                         if(qFileName == "")     {       qLines = lines; }
 232
 233                                         createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames);
 234
 235                                         rename((trimSeqFile + toString(processIDS[0]) + ".temp").c_str(), trimSeqFile.c_str());
 236                                         rename((scrapSeqFile + toString(processIDS[0]) + ".temp").c_str(), scrapSeqFile.c_str());
 237                                         rename((groupFile + toString(processIDS[0]) + ".temp").c_str(), groupFile.c_str());
 238                                         for (int j = 0; j < fastaFileNames.size(); j++) {
 239                                                 rename((fastaFileNames[j] + toString(processIDS[0]) + ".temp").c_str(), fastaFileNames[j].c_str());
 240                                         }
 241                                         //append files
 242                                         for(int i=1;i<processors;i++){
 243                                                 appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
 244                                                 remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
 245                                                 appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
 246                                                 remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
 247                                                 appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
 248                                                 remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
 249                                                 for (int j = 0; j < fastaFileNames.size(); j++) {
 250                                                         appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
 251                                                         remove((fastaFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
 252                                                 }
 253                                         }
 254                                 }
 255
 256                                 if (m->control_pressed) {  return 0; }
 257                 #else
 258                                 ifstream inFASTA;
 259                                 openInputFile(fastafileNames[s], inFASTA);
 260                                 numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 261                                 inFASTA.close();
 262
 263                                 lines.push_back(new linePair(0, numSeqs));
 264
 265                                 driverCreateSummary(fastafile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
 266
 267                                 if (m->control_pressed) {  return 0; }
 268                 #endif
 269
 270
 271                 for(int i=0;i<fastaFileNames.size();i++){
 272                         ifstream inFASTA;
 273                         string seqName;
 274                         openInputFile(getRootName(fastaFile) + groupVector[i] + ".fasta", inFASTA);
 275                         ofstream outGroups;
 276                         openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
 277                         outputNames.push_back(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups");
 278
 279                         while(!inFASTA.eof()){
 280                                 if(inFASTA.get() == '>'){
 281                                         inFASTA >> seqName;
 282                                         outGroups << seqName << '\t' << groupVector[i] << endl;
 283                                 }
 284                                 while (!inFASTA.eof())  {       char c = inFASTA.get(); if (c == 10 || c == 13){        break;  }       }
 285                         }
 286                         outGroups.close();
 287                         inFASTA.close();
 288                 }
 289
 290                 if (m->control_pressed) {
 291                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 292                         return 0;
 293                 }
 294
 295                 m->mothurOutEndLine();
 296                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 297                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 298                 m->mothurOutEndLine();
 299
 300                 return 0;
 301
 302         }
 303         catch(exception& e) {
 304                 m->errorOut(e, "TrimSeqsCommand", "execute");
 305                 exit(1);
 306         }
 307 }
 308
 309 /**************************************************************************************/
 310 int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames, linePair* line, linePair* qline) {
 311         try {
 312
 313                 ofstream outFASTA;
 314                 int able = openOutputFile(trimFile, outFASTA);
 315
 316                 ofstream scrapFASTA;
 317                 openOutputFile(scrapFile, scrapFASTA);
 318
 319                 ofstream outGroups;
 320                 vector<ofstream*> fastaFileNames;
 321                 if (oligoFile != "") {
 322                         openOutputFile(groupFile, outGroups);
 323                         for (int i = 0; i < fastaNames.size(); i++) {
 324                                 fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
 325                         }
 326                 }
 327
 328                 ifstream inFASTA;
 329                 openInputFile(filename, inFASTA);
 330
 331                 ifstream qFile;
 332                 if(qFileName != "")     {       openInputFile(qFileName, qFile);        }
 333
 334                 qFile.seekg(qline->start);
 335                 inFASTA.seekg(line->start);
 336
 337                 for(int i=0;i<line->num;i++){
 338
 339                         if (m->control_pressed) {
 340                                 inFASTA.close();
 341                                 outFASTA.close();
 342                                 scrapFASTA.close();
 343                                 if (oligoFile != "") {   outGroups.close();   }
 344                                 if(qFileName != "")     {       qFile.close();  }
 345                                 for(int i=0;i<fastaFileNames.size();i++){
 346                                         fastaFileNames[i]->close();
 347                                         delete fastaFileNames[i];
 348                                 }
 349                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 350                                 return 0;
 351                         }
 352
 353                         bool success = 1;
 354
 355                         Sequence currSeq(inFASTA);
 356
 357                         string origSeq = currSeq.getUnaligned();
 358                         if (origSeq != "") {
 359                                 int group;
 360                                 string trashCode = "";
 361
 362                                 if(qFileName != ""){
 363                                         if(qThreshold != 0)             {       success = stripQualThreshold(currSeq, qFile);   }
 364                                         else if(qAverage != 0)  {       success = cullQualAverage(currSeq, qFile);              }
 365                                         if ((!qtrim) && (origSeq.length() != currSeq.getUnaligned().length())) {
 366                                                 success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
 367                                         }
 368                                         if(!success)                    {       trashCode += 'q';                                                               }
 369                                 }
 370
 371                                 if(barcodes.size() != 0){
 372                                         success = stripBarcode(currSeq, group);
 373                                         if(!success){   trashCode += 'b';       }
 374                                 }
 375
 376                                 if(numFPrimers != 0){
 377                                         success = stripForward(currSeq);
 378                                         if(!success){   trashCode += 'f';       }
 379                                 }
 380
 381                                 if(numRPrimers != 0){
 382                                         success = stripReverse(currSeq);
 383                                         if(!success){   trashCode += 'r';       }
 384                                 }
 385
 386                                 if(minLength > 0 || maxLength > 0){
 387                                         success = cullLength(currSeq);
 388                                         if(!success){   trashCode += 'l'; }
 389                                 }
 390                                 if(maxHomoP > 0){
 391                                         success = cullHomoP(currSeq);
 392                                         if(!success){   trashCode += 'h';       }
 393                                 }
 394                                 if(maxAmbig != -1){
 395                                         success = cullAmbigs(currSeq);
 396                                         if(!success){   trashCode += 'n';       }
 397                                 }
 398
 399                                 if(flip){       currSeq.reverseComplement();    }               // should go last
 400
 401                                 if(trashCode.length() == 0){
 402                                         currSeq.setAligned(currSeq.getUnaligned());  //this is because of a modification we made to the sequence class to fix a bug.  all seqs have an aligned version, which is the version that gets printed.
 403                                         currSeq.printSequence(outFASTA);
 404                                         if(barcodes.size() != 0){
 405                                                 outGroups << currSeq.getName() << '\t' << groupVector[group] << endl;
 406
 407                                                 if(allFiles){
 408                                                         currSeq.printSequence(*fastaFileNames[group]);
 409                                                 }
 410                                         }
 411                                 }
 412                                 else{
 413                                         currSeq.setName(currSeq.getName() + '|' + trashCode);
 414                                         currSeq.setUnaligned(origSeq);
 415                                         currSeq.printSequence(scrapFASTA);
 416                                 }
 417                         }
 418                         gobble(inFASTA);
 419                 }
 420
 421                 inFASTA.close();
 422                 outFASTA.close();
 423                 scrapFASTA.close();
 424                 if (oligoFile != "") {   outGroups.close();   }
 425                 if(qFileName != "")     {       qFile.close();  }
 426
 427                 for(int i=0;i<fastaFileNames.size();i++){
 428                         fastaFileNames[i]->close();
 429                         delete fastaFileNames[i];
 430                 }
 431
 432                 return 0;
 433         }
 434         catch(exception& e) {
 435                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 436                 exit(1);
 437         }
 438 }
 439 /**************************************************************************************************/
 440 int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames) {
 441         try {
 442 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 443                 int process = 0;
 444                 int exitCommand = 1;
 445                 processIDS.clear();
 446
 447                 //loop through and create all the processes you want
 448                 while (process != processors) {
 449                         int pid = fork();
 450
 451                         if (pid > 0) {
 452                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 453                                 process++;
 454                         }else if (pid == 0){
 455                                 driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, lines[process], qLines[process]);
 456                                 exit(0);
 457                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 458                 }
 459
 460                 //force parent to wait until all the processes are done
 461                 for (int i=0;i<processors;i++) {
 462                         int temp = processIDS[i];
 463                         wait(&temp);
 464                 }
 465
 466                 return exitCommand;
 467 #endif
 468         }
 469         catch(exception& e) {
 470                 m->errorOut(e, "TrimSeqsCommand", "createProcessesCreateTrim");
 471                 exit(1);
 472         }
 473 }
 474 /**************************************************************************************************/
 475
 476 int TrimSeqsCommand::setLines(string filename, vector<linePair*>& lines) {
 477         try {
 478
 479                 lines.clear();
 480
 481                 vector<long int> positions;
 482
 483                 ifstream inFASTA;
 484                 openInputFile(filename, inFASTA);
 485
 486                 string input;
 487                 while(!inFASTA.eof()){
 488                         input = getline(inFASTA);
 489
 490                         if (input.length() != 0) {
 491                                 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);     }
 492                         }
 493                 }
 494                 inFASTA.close();
 495
 496                 int numFastaSeqs = positions.size();
 497
 498                 FILE * pFile;
 499                 long size;
 500
 501                 //get num bytes in file
 502                 pFile = fopen (filename.c_str(),"rb");
 503                 if (pFile==NULL) perror ("Error opening file");
 504                 else{
 505                         fseek (pFile, 0, SEEK_END);
 506                         size=ftell (pFile);
 507                         fclose (pFile);
 508                 }
 509
 510                 int numSeqsPerProcessor = numFastaSeqs / processors;
 511
 512                 for (int i = 0; i < processors; i++) {
 513
 514                         long int startPos = positions[ i * numSeqsPerProcessor ];
 515                         if(i == processors - 1){
 516                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 517                         }else{
 518                                 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
 519                         }
 520                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 521                 }
 522
 523                 return numFastaSeqs;
 524         }
 525         catch(exception& e) {
 526                 m->errorOut(e, "TrimSeqsCommand", "setLines");
 527                 exit(1);
 528         }
 529 }
 530 //***************************************************************************************************************
 531
 532 void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec){ //vector<ofstream*>& outFASTAVec
 533         try {
 534                 ifstream inOligos;
 535                 openInputFile(oligoFile, inOligos);
 536
 537                 ofstream test;
 538
 539                 string type, oligo, group;
 540                 int index=0;
 541
 542                 while(!inOligos.eof()){
 543                         inOligos >> type;
 544
 545                         if(type[0] == '#'){
 546                                 while (!inOligos.eof()) {       char c = inOligos.get(); if (c == 10 || c == 13){       break;  }       } // get rest of line if there's any crap there
 547                         }
 548                         else{
 549                                 inOligos >> oligo;
 550
 551                                 for(int i=0;i<oligo.length();i++){
 552                                         oligo[i] = toupper(oligo[i]);
 553                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 554                                 }
 555
 556                                 if(type == "forward"){
 557                                         forPrimer.push_back(oligo);
 558                                 }
 559                                 else if(type == "reverse"){
 560                                         Sequence oligoRC("reverse", oligo);
 561                                         oligoRC.reverseComplement();
 562                                         revPrimer.push_back(oligoRC.getUnaligned());
 563                                 }
 564                                 else if(type == "barcode"){
 565                                         inOligos >> group;
 566                                         barcodes[oligo]=index++;
 567                                         groupVector.push_back(group);
 568
 569                                         if(allFiles){
 570                                                 //outFASTAVec.push_back(new ofstream((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta").c_str(), ios::ate));
 571                                                 outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta"));
 572                                                 outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + group + ".fasta"));
 573                                         }
 574                                 }
 575                         }
 576                 }
 577
 578                 inOligos.close();
 579
 580                 numFPrimers = forPrimer.size();
 581                 numRPrimers = revPrimer.size();
 582
 583         }
 584         catch(exception& e) {
 585                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 586                 exit(1);
 587         }
 588 }
 589 //***************************************************************************************************************
 590
 591 bool TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
 592         try {
 593                 string rawSequence = seq.getUnaligned();
 594                 bool success = 0;       //guilty until proven innocent
 595
 596                 //can you find the barcode
 597                 for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 598                         string oligo = it->first;
 599                         if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 600                                 success = 0;
 601                                 break;
 602                         }
 603
 604                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 605                                 group = it->second;
 606                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 607                                 success = 1;
 608                                 break;
 609                         }
 610                 }
 611
 612                 //if you found the barcode or if you don't want to allow for diffs
 613                 if ((diffs == 0) || (success == 1)) { return success;  }
 614
 615                 else { //try aligning and see if you can find it
 616                         //can you find the barcode
 617                         for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
 618                                 string oligo = it->first;
 619                                 if(rawSequence.length() < oligo.length()){      //let's just assume that the barcodes are the same length
 620                                         success = 0;
 621                                         break;
 622                                 }
 623
 624                                 //use needleman to align first barcode.length()+numdiffs of sequence to each barcode
 625                                 Alignment* alignment = new NeedlemanOverlap(-2.0, 1.0, -1.0, (oligo.length()+diffs+1));
 626                                 Sequence* templateSeq = new Sequence("temp", rawSequence.substr(0,(oligo.length()+diffs)));
 627                                 Sequence* candidateSeq = new Sequence("temp2", oligo);
 628                                 Nast nast(alignment, candidateSeq, templateSeq);
 629
 630                                 oligo = candidateSeq->getAligned();
 631         cout << "barcode = " << oligo << " raw = " << rawSequence.substr(0,(oligo.length())) << endl;
 632                                 delete alignment;
 633                                 delete templateSeq;
 634                                 delete candidateSeq;
 635
 636                                 if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 637                                         group = it->second;
 638                                         seq.setUnaligned(rawSequence.substr(0,oligo.length()));
 639                                         success = 1;
 640                                         break;
 641                                 }
 642                         }
 643                 }
 644                 return success;
 645
 646         }
 647         catch(exception& e) {
 648                 m->errorOut(e, "TrimSeqsCommand", "stripBarcode");
 649                 exit(1);
 650         }
 651
 652 }
 653
 654 //***************************************************************************************************************
 655
 656 bool TrimSeqsCommand::stripForward(Sequence& seq){
 657         try {
 658                 string rawSequence = seq.getUnaligned();
 659                 bool success = 0;       //guilty until proven innocent
 660
 661                 for(int i=0;i<numFPrimers;i++){
 662                         string oligo = forPrimer[i];
 663
 664                         if(rawSequence.length() < oligo.length()){
 665                                 success = 0;
 666                                 break;
 667                         }
 668
 669                         if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 670                                 seq.setUnaligned(rawSequence.substr(oligo.length()));
 671                                 success = 1;
 672                                 break;
 673                         }
 674                 }
 675
 676                 //if you found the primer or if you don't want to allow for diffs
 677                 if ((diffs == 0) || (success == 1)) { return success;  }
 678
 679                 else { //try aligning and see if you can find it
 680                         //can you find the primer
 681                         for(int i=0;i<numFPrimers;i++){
 682                                 string oligo = forPrimer[i];
 683                                 if(rawSequence.length() < oligo.length()){
 684                                         success = 0;
 685                                         break;
 686                                 }
 687
 688                                 //use needleman to align first primer.length()+numdiffs of sequence to each primer
 689                                 Alignment* alignment = new NeedlemanOverlap(-2.0, 1.0, -1.0, (oligo.length()+diffs+1));
 690                                 Sequence* templateSeq = new Sequence("temp", rawSequence.substr(0,(oligo.length()+diffs)));
 691                                 Sequence* candidateSeq = new Sequence("temp2", oligo);
 692                                 Nast nast(alignment, candidateSeq, templateSeq);
 693
 694                                 oligo = candidateSeq->getAligned();
 695
 696                                 delete alignment;
 697                                 delete templateSeq;
 698                                 delete candidateSeq;
 699
 700                                 if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
 701                                         seq.setUnaligned(rawSequence.substr(0,oligo.length()));
 702                                         success = 1;
 703                                         break;
 704                                 }
 705                         }
 706                 }
 707
 708                 return success;
 709
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "TrimSeqsCommand", "stripForward");
 713                 exit(1);
 714         }
 715 }
 716
 717 //***************************************************************************************************************
 718
 719 bool TrimSeqsCommand::stripReverse(Sequence& seq){
 720         try {
 721                 string rawSequence = seq.getUnaligned();
 722                 bool success = 0;       //guilty until proven innocent
 723
 724                 for(int i=0;i<numRPrimers;i++){
 725                         string oligo = revPrimer[i];
 726
 727                         if(rawSequence.length() < oligo.length()){
 728                                 success = 0;
 729                                 break;
 730                         }
 731
 732                         if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
 733                                 seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
 734                                 success = 1;
 735                                 break;
 736                         }
 737                 }
 738                 return success;
 739
 740         }
 741         catch(exception& e) {
 742                 m->errorOut(e, "TrimSeqsCommand", "stripReverse");
 743                 exit(1);
 744         }
 745 }
 746
 747 //***************************************************************************************************************
 748
 749 bool TrimSeqsCommand::cullLength(Sequence& seq){
 750         try {
 751
 752                 int length = seq.getNumBases();
 753                 bool success = 0;       //guilty until proven innocent
 754
 755                 if(length >= minLength && maxLength == 0)                       {       success = 1;    }
 756                 else if(length >= minLength && length <= maxLength)     {       success = 1;    }
 757                 else                                                                                            {       success = 0;    }
 758
 759                 return success;
 760
 761         }
 762         catch(exception& e) {
 763                 m->errorOut(e, "TrimSeqsCommand", "cullLength");
 764                 exit(1);
 765         }
 766
 767 }
 768
 769 //***************************************************************************************************************
 770
 771 bool TrimSeqsCommand::cullHomoP(Sequence& seq){
 772         try {
 773                 int longHomoP = seq.getLongHomoPolymer();
 774                 bool success = 0;       //guilty until proven innocent
 775
 776                 if(longHomoP <= maxHomoP){      success = 1;    }
 777                 else                                    {       success = 0;    }
 778
 779                 return success;
 780         }
 781         catch(exception& e) {
 782                 m->errorOut(e, "TrimSeqsCommand", "cullHomoP");
 783                 exit(1);
 784         }
 785
 786 }
 787
 788 //***************************************************************************************************************
 789
 790 bool TrimSeqsCommand::cullAmbigs(Sequence& seq){
 791         try {
 792                 int numNs = seq.getAmbigBases();
 793                 bool success = 0;       //guilty until proven innocent
 794
 795                 if(numNs <= maxAmbig)   {       success = 1;    }
 796                 else                                    {       success = 0;    }
 797
 798                 return success;
 799         }
 800         catch(exception& e) {
 801                 m->errorOut(e, "TrimSeqsCommand", "cullAmbigs");
 802                 exit(1);
 803         }
 804
 805 }
 806
 807 //***************************************************************************************************************
 808
 809 bool TrimSeqsCommand::compareDNASeq(string oligo, string seq){
 810         try {
 811                 bool success = 1;
 812                 int length = oligo.length();
 813
 814                 for(int i=0;i<length;i++){
 815
 816                         if(oligo[i] != seq[i]){
 817                                 if(oligo[i] == 'A' || oligo[i] == 'T' || oligo[i] == 'G' || oligo[i] == 'C')    {       success = 0;    }
 818                                 else if((oligo[i] == 'N' || oligo[i] == 'I') && (seq[i] == 'N'))                                {       success = 0;    }
 819                                 else if(oligo[i] == 'R' && (seq[i] != 'A' && seq[i] != 'G'))                                    {       success = 0;    }
 820                                 else if(oligo[i] == 'Y' && (seq[i] != 'C' && seq[i] != 'T'))                                    {       success = 0;    }
 821                                 else if(oligo[i] == 'M' && (seq[i] != 'C' && seq[i] != 'A'))                                    {       success = 0;    }
 822                                 else if(oligo[i] == 'K' && (seq[i] != 'T' && seq[i] != 'G'))                                    {       success = 0;    }
 823                                 else if(oligo[i] == 'W' && (seq[i] != 'T' && seq[i] != 'A'))                                    {       success = 0;    }
 824                                 else if(oligo[i] == 'S' && (seq[i] != 'C' && seq[i] != 'G'))                                    {       success = 0;    }
 825                                 else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 826                                 else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G'))   {       success = 0;    }
 827                                 else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C'))   {       success = 0;    }
 828                                 else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G'))   {       success = 0;    }
 829
 830                                 if(success == 0)        {       break;   }
 831                         }
 832                         else{
 833                                 success = 1;
 834                         }
 835                 }
 836
 837                 return success;
 838         }
 839         catch(exception& e) {
 840                 m->errorOut(e, "TrimSeqsCommand", "compareDNASeq");
 841                 exit(1);
 842         }
 843
 844 }
 845
 846 //***************************************************************************************************************
 847
 848 bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
 849         try {
 850                 string rawSequence = seq.getUnaligned();
 851                 int seqLength;  // = rawSequence.length();
 852                 string name, temp, temp2;
 853
 854                 qFile >> name >> temp;
 855
 856                 splitAtEquals(temp2, temp); //separates length=242, temp=length, temp2=242
 857                 convert(temp, seqLength); //converts string to int
 858
 859                 if (name.length() != 0) {  if(name.substr(1) != seq.getName())  {       m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); }  }
 860                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 861
 862                 int score;
 863                 int end = seqLength;
 864
 865                 for(int i=0;i<seqLength;i++){
 866                         qFile >> score;
 867
 868                         if(score <= qThreshold){
 869                                 end = i;
 870                                 break;
 871                         }
 872                 }
 873                 for(int i=end+1;i<seqLength;i++){
 874                         qFile >> score;
 875                 }
 876
 877                 seq.setUnaligned(rawSequence.substr(0,end));
 878
 879                 return 1;
 880         }
 881         catch(exception& e) {
 882                 m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
 883                 exit(1);
 884         }
 885 }
 886
 887 //***************************************************************************************************************
 888
 889 bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
 890         try {
 891                 string rawSequence = seq.getUnaligned();
 892                 int seqLength = seq.getNumBases();
 893                 bool success = 0;       //guilty until proven innocent
 894                 string name;
 895
 896                 qFile >> name;
 897                 if (name[0] == '>') {  if(name.substr(1) != seq.getName())      {       m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); }  }
 898
 899                 while (!qFile.eof())    {       char c = qFile.get(); if (c == 10 || c == 13){  break;  }       }
 900
 901                 float score;
 902                 float average = 0;
 903
 904                 for(int i=0;i<seqLength;i++){
 905                         qFile >> score;
 906                         average += score;
 907                 }
 908                 average /= seqLength;
 909
 910                 if(average >= qAverage) {       success = 1;    }
 911                 else                                    {       success = 0;    }
 912
 913                 return success;
 914         }
 915         catch(exception& e) {
 916                 m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
 917                 exit(1);
 918         }
 919 }
 920
 921 //***************************************************************************************************************