makecontigscommand.cpp

   1 //
   2 //  makecontigscommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 5/15/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "makecontigscommand.h"
  10
  11 //**********************************************************************************************************************
  12 vector<string> MakeContigsCommand::setParameters(){
  13         try {
  14                 CommandParameter pfastq("ffastq", "InputTypes", "", "", "FastaFastqFile", "FastaFastqFile", "fastqGroup","fasta-qfile",false,false,true); parameters.push_back(pfastq);
  15         CommandParameter prfastq("rfastq", "InputTypes", "", "", "none", "none", "fastqGroup","fasta-qfile",false,false,true); parameters.push_back(prfastq);
  16         CommandParameter pfasta("ffasta", "InputTypes", "", "", "FastaFastqFile", "FastaFastqFile", "fastaGroup","fasta",false,false,true); parameters.push_back(pfasta);
  17         CommandParameter prfasta("rfasta", "InputTypes", "", "", "none", "none", "none","fastaGroup",false,false,true); parameters.push_back(prfasta);
  18         CommandParameter pfqual("fqfile", "InputTypes", "", "", "none", "none", "qfileGroup","",false,false,true); parameters.push_back(pfqual);
  19         CommandParameter prqual("rqfile", "InputTypes", "", "", "none", "none", "qfileGroup","",false,false,true); parameters.push_back(prqual);
  20         CommandParameter pfile("file", "InputTypes", "", "", "FastaFastqFile", "FastaFastqFile", "none","fasta-qfile",false,false,true); parameters.push_back(pfile);
  21         CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none","group",false,false,true); parameters.push_back(poligos);
  22                 CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(ppdiffs);
  23                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(pbdiffs);
  24 //        CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(pldiffs);
  25 //              CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(psdiffs);
  26         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
  27
  28         CommandParameter palign("align", "Multiple", "needleman-gotoh", "needleman", "", "", "","",false,false); parameters.push_back(palign);
  29         CommandParameter pallfiles("allfiles", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pallfiles);
  30                 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "","",false,false); parameters.push_back(pmatch);
  31                 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "","",false,false); parameters.push_back(pmismatch);
  32                 CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "","",false,false); parameters.push_back(pgapopen);
  33                 CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "","",false,false); parameters.push_back(pgapextend);
  34         CommandParameter pthreshold("insert", "Number", "", "25", "", "", "","",false,false); parameters.push_back(pthreshold);
  35         CommandParameter pdeltaq("deltaq", "Number", "", "6", "", "", "","",false,false); parameters.push_back(pdeltaq);
  36                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  37         CommandParameter pformat("format", "Multiple", "sanger-illumina-solexa-illumina1.8+", "illumina1.8+", "", "", "","",false,false,true); parameters.push_back(pformat);
  38                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  39                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  40
  41                 vector<string> myArray;
  42                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "MakeContigsCommand", "setParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 string MakeContigsCommand::getHelpString(){
  52         try {
  53                 string helpString = "";
  54                 helpString += "The make.contigs command reads a file, forward fastq file and a reverse fastq file or forward fasta and reverse fasta files and outputs new fasta.  It will also provide new quality files if the fastq or file parameter is used.\n";
  55         helpString += "If an oligos file is provided barcodes and primers will be trimmed, and a group file will be created.\n";
  56                 helpString += "The make.contigs command parameters are file, ffastq, rfastq, ffasta, rfasta, fqfile, rqfile, oligos, format, tdiffs, bdiffs, pdiffs, align, match, mismatch, gapopen, gapextend, insert, deltaq, allfiles and processors.\n";
  57                 helpString += "The ffastq and rfastq, file, or ffasta and rfasta parameters are required.\n";
  58         helpString += "The file parameter is 2 column file containing the forward fastq files in the first column and their matching reverse fastq files in the second column.  Mothur will process each pair and create a combined fasta and report file with all the sequences.\n";
  59         helpString += "The ffastq and rfastq parameters are used to provide a forward fastq and reverse fastq file to process.  If you provide one, you must provide the other.\n";
  60         helpString += "The ffasta and rfasta parameters are used to provide a forward fasta and reverse fasta file to process.  If you provide one, you must provide the other.\n";
  61         helpString += "The fqfile and rqfile parameters are used to provide a forward quality and reverse quality files to process with the ffasta and rfasta parameters.  If you provide one, you must provide the other.\n";
  62                 helpString += "The format parameter is used to indicate whether your sequences are sanger, solexa, illumina1.8+ or illumina, default=illumina1.8+.\n";
  63         helpString += "The align parameter allows you to specify the alignment method to use.  Your options are: gotoh and needleman. The default is needleman.\n";
  64         helpString += "The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs + sdiffs + ldiffs.\n";
  65                 helpString += "The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n";
  66                 helpString += "The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n";
  67         //helpString += "The ldiffs parameter is used to specify the number of differences allowed in the linker. The default is 0.\n";
  68                 //helpString += "The sdiffs parameter is used to specify the number of differences allowed in the spacer. The default is 0.\n";
  69                 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n";
  70                 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n";
  71         helpString += "The deltaq parameter allows you to specify the delta allowed between quality scores of a mismatched base.  For example in the overlap, if deltaq=5 and in the alignment seqA, pos 200 has a quality score of 30 and the same position in seqB has a quality score of 20, you take the base from seqA (30-20 >= 5).  If the quality score in seqB is 28 then the base in the consensus will be an N (30-28<5) The default is 6.\n";
  72                 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n";
  73                 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n";
  74         helpString += "The insert parameter allows you to set a quality scores threshold. In the case where we are trying to decide whether to keep a base or remove it because the base is compared to a gap in the other fragment, if the base has a quality score below the threshold we eliminate it. Default=25.\n";
  75         helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  76         helpString += "The allfiles parameter will create separate group and fasta file for each grouping. The default is F.\n";
  77         helpString += "The make.contigs command should be in the following format: \n";
  78                 helpString += "make.contigs(ffastq=yourForwardFastqFile, rfastq=yourReverseFastqFile, align=yourAlignmentMethod) \n";
  79                 helpString += "Note: No spaces between parameter labels (i.e. ffastq), '=' and parameters (i.e.yourForwardFastqFile).\n";
  80                 return helpString;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "MakeContigsCommand", "getHelpString");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 string MakeContigsCommand::getOutputPattern(string type) {
  89     try {
  90         string pattern = "";
  91
  92         if (type == "fasta") {  pattern = "[filename],[tag],contigs.fasta"; }
  93         else if (type == "group") {  pattern = "[filename],[tag],contigs.groups"; }
  94         else if (type == "report") {  pattern = "[filename],[tag],contigs.report"; }
  95         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  96
  97         return pattern;
  98     }
  99     catch(exception& e) {
 100         m->errorOut(e, "MakeContigsCommand", "getOutputPattern");
 101         exit(1);
 102     }
 103 }
 104 //**********************************************************************************************************************
 105 MakeContigsCommand::MakeContigsCommand(){
 106         try {
 107                 abort = true; calledHelp = true;
 108                 setParameters();
 109                 vector<string> tempOutNames;
 110                 outputTypes["fasta"] = tempOutNames;
 111         outputTypes["group"] = tempOutNames;
 112         outputTypes["report"] = tempOutNames;
 113         }
 114         catch(exception& e) {
 115                 m->errorOut(e, "MakeContigsCommand", "MakeContigsCommand");
 116                 exit(1);
 117         }
 118 }
 119 //**********************************************************************************************************************
 120 MakeContigsCommand::MakeContigsCommand(string option)  {
 121         try {
 122                 abort = false; calledHelp = false;
 123
 124                 //allow user to run help
 125                 if(option == "help") { help(); abort = true; calledHelp = true; }
 126                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 127
 128                 else {
 129                         vector<string> myArray = setParameters();
 130
 131                         OptionParser parser(option);
 132                         map<string, string> parameters = parser.getParameters();
 133
 134                         ValidParameters validParameter("pairwise.seqs");
 135                         map<string, string>::iterator it;
 136
 137                         //check to make sure all parameters are valid for command
 138                         for (it = parameters.begin(); it != parameters.end(); it++) {
 139                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 140                         }
 141
 142                         //initialize outputTypes
 143                         vector<string> tempOutNames;
 144                         outputTypes["fasta"] = tempOutNames;
 145             outputTypes["report"] = tempOutNames;
 146             outputTypes["group"] = tempOutNames;
 147
 148
 149                         //if the user changes the input directory command factory will send this info to us in the output parameter
 150                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 151                         if (inputDir == "not found"){   inputDir = "";          }
 152                         else {
 153                                 string path;
 154                 it = parameters.find("ffastq");
 155                                 //user has given a template file
 156                                 if(it != parameters.end()){
 157                                         path = m->hasPath(it->second);
 158                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                         if (path == "") {       parameters["ffastq"] = inputDir + it->second;           }
 160                                 }
 161
 162                 it = parameters.find("rfastq");
 163                                 //user has given a template file
 164                                 if(it != parameters.end()){
 165                                         path = m->hasPath(it->second);
 166                                         //if the user has not given a path then, add inputdir. else leave path alone.
 167                                         if (path == "") {       parameters["rfastq"] = inputDir + it->second;           }
 168                                 }
 169
 170                 it = parameters.find("ffasta");
 171                                 //user has given a template file
 172                                 if(it != parameters.end()){
 173                                         path = m->hasPath(it->second);
 174                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                         if (path == "") {       parameters["ffasta"] = inputDir + it->second;           }
 176                                 }
 177
 178                 it = parameters.find("rfasta");
 179                                 //user has given a template file
 180                                 if(it != parameters.end()){
 181                                         path = m->hasPath(it->second);
 182                                         //if the user has not given a path then, add inputdir. else leave path alone.
 183                                         if (path == "") {       parameters["rfasta"] = inputDir + it->second;           }
 184                                 }
 185
 186                 it = parameters.find("fqfile");
 187                                 //user has given a template file
 188                                 if(it != parameters.end()){
 189                                         path = m->hasPath(it->second);
 190                                         //if the user has not given a path then, add inputdir. else leave path alone.
 191                                         if (path == "") {       parameters["fqfile"] = inputDir + it->second;           }
 192                                 }
 193
 194                 it = parameters.find("rqfile");
 195                                 //user has given a template file
 196                                 if(it != parameters.end()){
 197                                         path = m->hasPath(it->second);
 198                                         //if the user has not given a path then, add inputdir. else leave path alone.
 199                                         if (path == "") {       parameters["rqfile"] = inputDir + it->second;           }
 200                                 }
 201
 202                 it = parameters.find("file");
 203                                 //user has given a template file
 204                                 if(it != parameters.end()){
 205                                         path = m->hasPath(it->second);
 206                                         //if the user has not given a path then, add inputdir. else leave path alone.
 207                                         if (path == "") {       parameters["file"] = inputDir + it->second;             }
 208                                 }
 209
 210                 it = parameters.find("oligos");
 211                                 //user has given a template file
 212                                 if(it != parameters.end()){
 213                                         path = m->hasPath(it->second);
 214                                         //if the user has not given a path then, add inputdir. else leave path alone.
 215                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 216                                 }
 217             }
 218
 219             ffastqfile = validParameter.validFile(parameters, "ffastq", true);
 220                         if (ffastqfile == "not open") {  abort = true; }
 221                         else if (ffastqfile == "not found") { ffastqfile = ""; }
 222
 223                         rfastqfile = validParameter.validFile(parameters, "rfastq", true);
 224                         if (rfastqfile == "not open") {  abort = true; }
 225                         else if (rfastqfile == "not found") { rfastqfile = "";  }
 226
 227             ffastafile = validParameter.validFile(parameters, "ffasta", true);
 228                         if (ffastafile == "not open") {  abort = true; }
 229                         else if (ffastafile == "not found") { ffastafile = ""; }
 230
 231                         rfastafile = validParameter.validFile(parameters, "rfasta", true);
 232                         if (rfastafile == "not open") {  abort = true; }
 233                         else if (rfastafile == "not found") { rfastafile = "";  }
 234
 235             fqualfile = validParameter.validFile(parameters, "fqfile", true);
 236                         if (fqualfile == "not open") {  abort = true; }
 237                         else if (fqualfile == "not found") { fqualfile = ""; }
 238
 239                         rqualfile = validParameter.validFile(parameters, "rqfile", true);
 240                         if (rqualfile == "not open") {  abort = true; }
 241                         else if (rqualfile == "not found") { rqualfile = "";  }
 242
 243             file = validParameter.validFile(parameters, "file", true);
 244                         if (file == "not open") {  abort = true; }
 245                         else if (file == "not found") { file = "";  }
 246
 247             //provide at least
 248             if ((file == "") && (ffastafile == "") && (ffastqfile == "")) { abort = true; m->mothurOut("[ERROR]: The file, ffastq and rfastq or ffasta and rfasta parameters are required.\n"); }
 249             if ((file != "") && ((ffastafile != "") || (ffastqfile != ""))) { abort = true; m->mothurOut("[ERROR]: The file, ffastq and rfastq or ffasta and rfasta parameters are required.\n"); }
 250             if ((ffastqfile != "") && (rfastqfile == "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the ffastq, you must provide a rfastq file.\n"); }
 251             if ((ffastqfile == "") && (rfastqfile != "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the rfastq, you must provide a ffastq file.\n"); }
 252             if ((ffastafile != "") && (rfastafile == "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the ffasta, you must provide a rfasta file.\n"); }
 253             if ((ffastafile == "") && (rfastafile != "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the rfasta, you must provide a ffasta file.\n"); }
 254             if ((fqualfile != "") && (rqualfile == "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the fqfile, you must provide a rqfile file.\n"); }
 255             if ((fqualfile == "") && (rqualfile != "")) {  abort = true; m->mothurOut("[ERROR]: If you provide use the rqfile, you must provide a fqfile file.\n"); }
 256             if (((fqualfile != "") || (rqualfile != "")) && ((ffastafile == "") || (rfastafile == ""))) {
 257                 abort = true; m->mothurOut("[ERROR]: If you provide use the rqfile or fqfile file, you must provide the ffasta and rfasta parameters.\n");
 258             }
 259
 260             oligosfile = validParameter.validFile(parameters, "oligos", true);
 261                         if (oligosfile == "not found")      {   oligosfile = "";        }
 262                         else if(oligosfile == "not open")   {   abort = true;       }
 263                         else {   m->setOligosFile(oligosfile);          }
 264
 265             //if the user changes the output directory command factory will send this info to us in the output parameter
 266                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 267                  outputDir = "";
 268             }
 269
 270
 271                         //check for optional parameter and set defaults
 272                         // ...at some point should added some additional type checking...
 273                         string temp;
 274                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 275                         m->mothurConvert(temp, match);
 276
 277                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 278                         m->mothurConvert(temp, misMatch);
 279             if (misMatch > 0) { m->mothurOut("[ERROR]: mismatch must be negative.\n"); abort=true; }
 280
 281                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 282                         m->mothurConvert(temp, gapOpen);
 283             if (gapOpen > 0) { m->mothurOut("[ERROR]: gapopen must be negative.\n"); abort=true; }
 284
 285                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 286                         m->mothurConvert(temp, gapExtend);
 287             if (gapExtend > 0) { m->mothurOut("[ERROR]: gapextend must be negative.\n"); abort=true; }
 288
 289             temp = validParameter.validFile(parameters, "insert", false);       if (temp == "not found"){       temp = "25";                    }
 290                         m->mothurConvert(temp, insert);
 291             if ((insert < 0) || (insert > 40)) { m->mothurOut("[ERROR]: insert must be between 0 and 40.\n"); abort=true; }
 292
 293             temp = validParameter.validFile(parameters, "deltaq", false);       if (temp == "not found"){       temp = "6";                     }
 294                         m->mothurConvert(temp, deltaq);
 295
 296                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 297                         m->setProcessors(temp);
 298                         m->mothurConvert(temp, processors);
 299
 300             temp = validParameter.validFile(parameters, "bdiffs", false);               if (temp == "not found") { temp = "0"; }
 301                         m->mothurConvert(temp, bdiffs);
 302
 303                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found") { temp = "0"; }
 304                         m->mothurConvert(temp, pdiffs);
 305
 306   //          temp = validParameter.validFile(parameters, "ldiffs", false);             if (temp == "not found") { temp = "0"; }
 307 //                      m->mothurConvert(temp, ldiffs);
 308             ldiffs = 0;
 309
 310  //           temp = validParameter.validFile(parameters, "sdiffs", false);             if (temp == "not found") { temp = "0"; }
 311  //           m->mothurConvert(temp, sdiffs);
 312             sdiffs = 0;
 313
 314                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs;  temp = toString(tempTotal); }
 315                         m->mothurConvert(temp, tdiffs);
 316
 317                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs;       }  //+ ldiffs + sdiffs;
 318
 319             temp = validParameter.validFile(parameters, "allfiles", false);             if (temp == "not found") { temp = "F"; }
 320                         allFiles = m->isTrue(temp);
 321
 322                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 323                         if ((align != "needleman") && (align != "gotoh")) { m->mothurOut(align + " is not a valid alignment method. Options are needleman or gotoh. I will use needleman."); m->mothurOutEndLine(); align = "needleman"; }
 324
 325             format = validParameter.validFile(parameters, "format", false);             if (format == "not found"){     format = "illumina1.8+";        }
 326
 327             if ((format != "sanger") && (format != "illumina") && (format != "illumina1.8+") && (format != "solexa"))  {
 328                                 m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa, illumina1.8+ and illumina, aborting." ); m->mothurOutEndLine();
 329                                 abort=true;
 330                         }
 331
 332             //fill convert table - goes from solexa to sanger. Used fq_all2std.pl as a reference.
 333             for (int i = -64; i < 65; i++) {
 334                 char temp = (char) ((int)(33 + 10*log(1+pow(10,(i/10.0)))/log(10)+0.499));
 335                 convertTable.push_back(temp);
 336             }
 337         }
 338
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "MakeContigsCommand", "MakeContigsCommand");
 342                 exit(1);
 343         }
 344 }
 345 //**********************************************************************************************************************
 346 int MakeContigsCommand::execute(){
 347         try {
 348                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 349
 350         //read ffastq and rfastq files creating fasta and qual files.
 351         //this function will create a forward and reverse, fasta and qual files for each processor.
 352         //files has an entry for each processor. files[i][0] = forwardFasta, files[i][1] = forwardQual, files[i][2] = reverseFasta, files[i][3] = reverseQual.  filesToProcess is for each filepair in the file parameter file.  for ffastq and rfastq this will be size 1.
 353         unsigned long int numReads = 0;
 354         int start = time(NULL);
 355         longestBase = 1000;
 356         m->mothurOut("Reading fastq data...\n");
 357         vector < vector< vector<string> > > filesToProcess = preProcessData(numReads);
 358         m->mothurOut("Done.\n");
 359
 360         if (m->control_pressed) { return 0; }
 361
 362         map<string, string> cvars;
 363         string compOutputDir = outputDir;
 364         if (outputDir == "") { compOutputDir = m->hasPath(file); }
 365         cvars["[filename]"] = compOutputDir + m->getRootName(m->getSimpleName(file));
 366         cvars["[tag]"] = "";
 367         string compositeGroupFile = getOutputFileName("group",cvars);
 368         cvars["[tag]"] = "trim";
 369         string compositeFastaFile = getOutputFileName("fasta",cvars);
 370         cvars["[tag]"] = "scrap";
 371         string compositeScrapFastaFile = getOutputFileName("fasta",cvars);
 372         cvars["[tag]"] = "";
 373         string compositeMisMatchFile = getOutputFileName("report",cvars);
 374
 375         if (filesToProcess.size() > 1) { //clear files for append below
 376             ofstream outCTFasta, outCTQual, outCSFasta, outCSQual, outCMisMatch;
 377             m->openOutputFile(compositeFastaFile, outCTFasta); outCTFasta.close();
 378             m->openOutputFile(compositeScrapFastaFile, outCSFasta); outCSFasta.close();
 379             m->openOutputFile(compositeMisMatchFile, outCMisMatch); outCMisMatch.close();
 380             outputNames.push_back(compositeFastaFile); outputTypes["fasta"].push_back(compositeFastaFile);
 381             outputNames.push_back(compositeMisMatchFile); outputTypes["report"].push_back(compositeMisMatchFile);
 382             outputNames.push_back(compositeScrapFastaFile); outputTypes["fasta"].push_back(compositeScrapFastaFile);
 383         }
 384
 385         for (int l = 0; l < filesToProcess.size(); l++) {
 386
 387             m->mothurOut("\n>>>>>\tProcessing " + filesToProcess[l][0][0] + " (file " + toString(l+1) + " of " + toString(filesToProcess.size()) + ")\t<<<<<\n");
 388
 389             vector<vector<string> > fastaFileNames;
 390             createGroup = false;
 391             string outputGroupFileName;
 392             map<string, string> variables;
 393             string thisOutputDir = outputDir;
 394             if (outputDir == "") {  thisOutputDir = m->hasPath(filesToProcess[l][0][0]); }
 395             variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(filesToProcess[l][0][0]));
 396             variables["[tag]"] = "";
 397             if(oligosfile != ""){
 398                 createGroup = getOligos(fastaFileNames, variables["[filename]"]);
 399                 if (createGroup) {
 400                     outputGroupFileName = getOutputFileName("group",variables);
 401                     outputNames.push_back(outputGroupFileName); outputTypes["group"].push_back(outputGroupFileName);
 402                 }
 403             }
 404
 405             variables["[tag]"] = "trim";
 406             string outFastaFile = getOutputFileName("fasta",variables);
 407             variables["[tag]"] = "scrap";
 408             string outScrapFastaFile = getOutputFileName("fasta",variables);
 409             variables["[tag]"] = "";
 410             string outMisMatchFile = getOutputFileName("report",variables);
 411             outputNames.push_back(outFastaFile); outputTypes["fasta"].push_back(outFastaFile);
 412             outputNames.push_back(outScrapFastaFile); outputTypes["fasta"].push_back(outScrapFastaFile);
 413             outputNames.push_back(outMisMatchFile); outputTypes["report"].push_back(outMisMatchFile);
 414
 415             m->mothurOut("Making contigs...\n");
 416             createProcesses(filesToProcess[l], outFastaFile, outScrapFastaFile, outMisMatchFile, fastaFileNames);
 417             m->mothurOut("Done.\n");
 418
 419             //remove temp fasta and qual files
 420             for (int i = 0; i < processors; i++) { for(int j = 0; j < filesToProcess[l][i].size(); j++) { m->mothurRemove(filesToProcess[l][i][j]); }  }
 421
 422             if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]); }  return 0; }
 423
 424             if(allFiles){
 425                 map<string, string> uniqueFastaNames;// so we don't add the same groupfile multiple times
 426                 map<string, string>::iterator it;
 427                 set<string> namesToRemove;
 428                 for(int i=0;i<fastaFileNames.size();i++){
 429                     for(int j=0;j<fastaFileNames[0].size();j++){
 430                         if (fastaFileNames[i][j] != "") {
 431                             if (namesToRemove.count(fastaFileNames[i][j]) == 0) {
 432                                 if(m->isBlank(fastaFileNames[i][j])){
 433                                     m->mothurRemove(fastaFileNames[i][j]);
 434                                     namesToRemove.insert(fastaFileNames[i][j]);
 435                                 }else{
 436                                     it = uniqueFastaNames.find(fastaFileNames[i][j]);
 437                                     if (it == uniqueFastaNames.end()) {
 438                                         uniqueFastaNames[fastaFileNames[i][j]] = barcodeNameVector[i];
 439                                     }
 440                                 }
 441                             }
 442                         }
 443                     }
 444                 }
 445
 446                 //remove names for outputFileNames, just cleans up the output
 447                 vector<string> outputNames2;
 448                 for(int i = 0; i < outputNames.size(); i++) { if (namesToRemove.count(outputNames[i]) == 0) { outputNames2.push_back(outputNames[i]); } }
 449                 outputNames = outputNames2;
 450
 451                 for (it = uniqueFastaNames.begin(); it != uniqueFastaNames.end(); it++) {
 452                     ifstream in;
 453                     m->openInputFile(it->first, in);
 454
 455                     ofstream out;
 456                     string thisGroupName = thisOutputDir + m->getRootName(m->getSimpleName(it->first));
 457                     thisGroupName += getOutputFileName("group",variables); outputNames.push_back(thisGroupName); outputTypes["group"].push_back(thisGroupName);
 458                     m->openOutputFile(thisGroupName, out);
 459
 460                     while (!in.eof()){
 461                         if (m->control_pressed) { break; }
 462
 463                         Sequence currSeq(in); m->gobble(in);
 464                         out << currSeq.getName() << '\t' << it->second << endl;
 465                     }
 466                     in.close();
 467                     out.close();
 468                 }
 469             }
 470
 471             if (createGroup) {
 472                 ofstream outGroup;
 473                 m->openOutputFile(outputGroupFileName, outGroup);
 474                 for (map<string, string>::iterator itGroup = groupMap.begin(); itGroup != groupMap.end(); itGroup++) {
 475                     outGroup << itGroup->first << '\t' << itGroup->second << endl;
 476                 }
 477                 outGroup.close();
 478             }
 479
 480             if (filesToProcess.size() > 1) { //merge into large combo files
 481                 if (createGroup) {
 482                     if (l == 0) {
 483                         ofstream outCGroup;
 484                         m->openOutputFile(compositeGroupFile, outCGroup); outCGroup.close();
 485                         outputNames.push_back(compositeGroupFile); outputTypes["group"].push_back(compositeGroupFile);
 486                     }
 487                     m->appendFiles(outputGroupFileName, compositeGroupFile);
 488                 }
 489                 m->appendFiles(outMisMatchFile, compositeMisMatchFile);
 490                 m->appendFiles(outFastaFile, compositeFastaFile);
 491                 m->appendFiles(outScrapFastaFile, compositeScrapFastaFile);
 492             }
 493         }
 494         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to process " + toString(numReads) + " sequences.\n");
 495
 496         if (m->control_pressed) {       for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0;    }
 497
 498                 //output group counts
 499                 m->mothurOutEndLine();
 500                 int total = 0;
 501                 if (groupCounts.size() != 0) {  m->mothurOut("Group count: \n");  }
 502                 for (map<string, int>::iterator it = groupCounts.begin(); it != groupCounts.end(); it++) {
 503             total += it->second; m->mothurOut(it->first + "\t" + toString(it->second)); m->mothurOutEndLine();
 504                 }
 505                 if (total != 0) { m->mothurOut("Total of all groups is " + toString(total)); m->mothurOutEndLine(); }
 506
 507                 if (m->control_pressed) {       for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0;    }
 508
 509         string currentFasta = "";
 510                 itTypes = outputTypes.find("fasta");
 511                 if (itTypes != outputTypes.end()) {
 512                         if ((itTypes->second).size() != 0) { currentFasta = (itTypes->second)[0]; m->setFastaFile(currentFasta); }
 513                 }
 514
 515         string currentGroup = "";
 516                 itTypes = outputTypes.find("group");
 517                 if (itTypes != outputTypes.end()) {
 518                         if ((itTypes->second).size() != 0) { currentGroup = (itTypes->second)[0]; m->setGroupFile(currentGroup); }
 519                 }
 520
 521         //output files created by command
 522                 m->mothurOutEndLine();
 523                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 524                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 525                 m->mothurOutEndLine();
 526
 527         return 0;
 528     }
 529         catch(exception& e) {
 530                 m->errorOut(e, "MakeContigsCommand", "execute");
 531                 exit(1);
 532         }
 533 }
 534 //**********************************************************************************************************************
 535 vector< vector< vector<string> > > MakeContigsCommand::preProcessData(unsigned long int& numReads) {
 536         try {
 537         vector< vector< vector<string> > > filesToProcess;
 538
 539         if (ffastqfile != "") { //reading one file
 540             vector< vector<string> > files = readFastqFiles(numReads, ffastqfile, rfastqfile);
 541             //adjust for really large processors or really small files
 542             if (numReads == 0) {  m->mothurOut("[ERROR]: no good reads.\n"); m->control_pressed = true; }
 543             if (numReads < processors) {
 544                 for (int i = numReads; i < processors; i++) { for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } files[i].clear(); }
 545                 files.resize(numReads);
 546                 processors = numReads;
 547             }
 548             filesToProcess.push_back(files);
 549         }else if (file != "") { //reading multiple files
 550             //return only valid pairs
 551             vector< vector<string> > filePairsToProcess = readFileNames(file);
 552
 553             if (m->control_pressed) { return filesToProcess; }
 554
 555             if (filePairsToProcess.size() != 0) {
 556                 for (int i = 0; i < filePairsToProcess.size(); i++) {
 557
 558                     if (m->control_pressed) { for (int l = 0; l < filesToProcess.size(); l++) { for (int k = 0; k < filesToProcess[l].size(); k++) { for(int j = 0; j < filesToProcess[l][k].size(); j++) { m->mothurRemove(filesToProcess[l][k][j]); } filesToProcess[l][k].clear(); } return filesToProcess; } }
 559
 560                     unsigned long int thisFilesReads;
 561                     vector< vector<string> > files = readFastqFiles(thisFilesReads, filePairsToProcess[i][0], filePairsToProcess[i][1]);
 562
 563                     //adjust for really large processors or really small files
 564                     if (thisFilesReads < processors) {
 565                         m->mothurOut("[ERROR]: " + filePairsToProcess[i][0] + " has less than " + toString(processors) + " good reads, skipping\n");
 566                         for (int k = 0; k < files.size(); k++) { for(int j = 0; j < files[k].size(); j++) { m->mothurRemove(files[k][j]); } files[k].clear(); }
 567                     }else {
 568                         filesToProcess.push_back(files);
 569                         numReads += thisFilesReads;
 570                     }
 571                 }
 572                 //all files are bad
 573                 if (numReads == 0) {  m->control_pressed = true; }
 574             }
 575         }else if (ffastafile != "") {
 576             vector< vector<string> > files = readFastaFiles(numReads, ffastafile, rfastafile);
 577             //adjust for really large processors or really small files
 578             if (numReads == 0) {  m->mothurOut("[ERROR]: no good reads.\n"); m->control_pressed = true; }
 579             if (numReads < processors) {
 580                 for (int i = numReads; i < processors; i++) { for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } files[i].clear(); }
 581                 files.resize(numReads);
 582                 processors = numReads;
 583             }
 584             filesToProcess.push_back(files);
 585         }else { m->control_pressed = true; } //should not get here
 586
 587         return filesToProcess;
 588     }
 589         catch(exception& e) {
 590                 m->errorOut(e, "MakeContigsCommand", "preProcessData");
 591                 exit(1);
 592         }
 593 }
 594 //**********************************************************************************************************************
 595 int MakeContigsCommand::createProcesses(vector< vector<string> > files, string outputFasta, string outputScrapFasta, string outputMisMatches, vector<vector<string> > fastaFileNames) {
 596         try {
 597                 int num = 0;
 598                 vector<int> processIDS;
 599 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 600                 int process = 0;
 601
 602                 //loop through and create all the processes you want
 603                 while (process != processors-1) {
 604                         int pid = fork();
 605
 606                         if (pid > 0) {
 607                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 608                                 process++;
 609                         }else if (pid == 0){
 610                 vector<vector<string> > tempFASTAFileNames = fastaFileNames;
 611
 612                                 if(allFiles){
 613                                         ofstream temp;
 614
 615                                         for(int i=0;i<tempFASTAFileNames.size();i++){
 616                                                 for(int j=0;j<tempFASTAFileNames[i].size();j++){
 617                                                         if (tempFASTAFileNames[i][j] != "") {
 618                                                                 tempFASTAFileNames[i][j] += toString(getpid()) + ".temp";
 619                                                                 m->openOutputFile(tempFASTAFileNames[i][j], temp);                      temp.close();
 620                                                         }
 621                                                 }
 622                                         }
 623                                 }
 624
 625                                 num = driver(files[process],
 626                              outputFasta + toString(getpid()) + ".temp",
 627                              outputScrapFasta + toString(getpid()) + ".temp",
 628                              outputMisMatches + toString(getpid()) + ".temp",
 629                              tempFASTAFileNames, process);
 630
 631                                 //pass groupCounts to parent
 632                 ofstream out;
 633                 string tempFile = toString(getpid()) + ".num.temp";
 634                 m->openOutputFile(tempFile, out);
 635                 out << num << endl;
 636                                 if(createGroup){
 637                                         out << groupCounts.size() << endl;
 638
 639                                         for (map<string, int>::iterator it = groupCounts.begin(); it != groupCounts.end(); it++) {
 640                                                 out << it->first << '\t' << it->second << endl;
 641                                         }
 642
 643                     out << groupMap.size() << endl;
 644                     for (map<string, string>::iterator it = groupMap.begin(); it != groupMap.end(); it++) {
 645                                                 out << it->first << '\t' << it->second << endl;
 646                                         }
 647                                 }
 648                 out.close();
 649
 650                                 exit(0);
 651                         }else {
 652                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 653                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 654                                 exit(0);
 655                         }
 656                 }
 657
 658         ofstream temp;
 659                 m->openOutputFile(outputFasta, temp);           temp.close();
 660         m->openOutputFile(outputScrapFasta, temp);              temp.close();
 661
 662                 //do my part
 663                 num = driver(files[processors-1], outputFasta, outputScrapFasta,  outputMisMatches, fastaFileNames, processors-1);
 664
 665                 //force parent to wait until all the processes are done
 666                 for (int i=0;i<processIDS.size();i++) {
 667                         int temp = processIDS[i];
 668                         wait(&temp);
 669                 }
 670
 671                 for (int i = 0; i < processIDS.size(); i++) {
 672             ifstream in;
 673             string tempFile = toString(processIDS[i]) + ".num.temp";
 674             m->openInputFile(tempFile, in);
 675             int tempNum;
 676             in >> tempNum; num += tempNum; m->gobble(in);
 677
 678                         if(createGroup){
 679                                 string group;
 680                                 in >> tempNum; m->gobble(in);
 681
 682                                 if (tempNum != 0) {
 683                                         for (int j = 0; j < tempNum; j++) {
 684                         int groupNum;
 685                                                 in >> group >> groupNum; m->gobble(in);
 686
 687                                                 map<string, int>::iterator it = groupCounts.find(group);
 688                                                 if (it == groupCounts.end()) {  groupCounts[group] = groupNum; }
 689                                                 else { groupCounts[it->first] += groupNum; }
 690                                         }
 691                                 }
 692                 in >> tempNum; m->gobble(in);
 693                 if (tempNum != 0) {
 694                                         for (int j = 0; j < tempNum; j++) {
 695                         string group, seqName;
 696                                                 in >> seqName >> group; m->gobble(in);
 697
 698                                                 map<string, string>::iterator it = groupMap.find(seqName);
 699                                                 if (it == groupMap.end()) {     groupMap[seqName] = group; }
 700                                                 else { m->mothurOut("[ERROR]: " + seqName + " is in your fasta file more than once. Sequence names must be unique. please correct.\n");  }
 701                                         }
 702                                 }
 703                         }
 704             in.close(); m->mothurRemove(tempFile);
 705         }
 706     #else
 707
 708         //////////////////////////////////////////////////////////////////////////////////////////////////////
 709                 //Windows version shared memory, so be careful when passing variables through the contigsData struct.
 710                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 711                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 712
 713                 vector<contigsData*> pDataArray;
 714                 DWORD   dwThreadIdArray[processors-1];
 715                 HANDLE  hThreadArray[processors-1];
 716
 717                 //Create processor worker threads.
 718                 for( int h=0; h<processors-1; h++ ){
 719                         string extension = "";
 720                         if (h != 0) { extension = toString(h) + ".temp"; processIDS.push_back(h); }
 721             vector<vector<string> > tempFASTAFileNames = fastaFileNames;
 722
 723             if(allFiles){
 724                 ofstream temp;
 725
 726                 for(int i=0;i<tempFASTAFileNames.size();i++){
 727                     for(int j=0;j<tempFASTAFileNames[i].size();j++){
 728                         if (tempFASTAFileNames[i][j] != "") {
 729                             tempFASTAFileNames[i][j] += extension;
 730                             m->openOutputFile(tempFASTAFileNames[i][j], temp);                  temp.close();
 731                         }
 732                     }
 733                 }
 734             }
 735
 736
 737                         contigsData* tempcontig = new contigsData(files[h], (outputFasta + extension), (outputScrapFasta + extension), (outputMisMatches + extension), align, m, match, misMatch, gapOpen, gapExtend, insert, deltaq, barcodes, primers, tempFASTAFileNames, barcodeNameVector, primerNameVector, pdiffs, bdiffs, tdiffs, createGroup, allFiles, h);
 738                         pDataArray.push_back(tempcontig);
 739
 740                         hThreadArray[h] = CreateThread(NULL, 0, MyContigsThreadFunction, pDataArray[h], 0, &dwThreadIdArray[h]);
 741                 }
 742
 743         vector<vector<string> > tempFASTAFileNames = fastaFileNames;
 744
 745         if(allFiles){
 746             ofstream temp;
 747             string extension = toString(processors-1) + ".temp";
 748
 749             for(int i=0;i<tempFASTAFileNames.size();i++){
 750                 for(int j=0;j<tempFASTAFileNames[i].size();j++){
 751                     if (tempFASTAFileNames[i][j] != "") {
 752                         tempFASTAFileNames[i][j] += extension;
 753                         m->openOutputFile(tempFASTAFileNames[i][j], temp);                      temp.close();
 754                     }
 755                 }
 756             }
 757         }
 758
 759                 //parent do my part
 760                 ofstream temp;
 761                 m->openOutputFile(outputFasta, temp);           temp.close();
 762         m->openOutputFile(outputScrapFasta, temp);              temp.close();
 763
 764         //do my part
 765         processIDS.push_back(processors-1);
 766                 num = driver(files[processors-1], (outputFasta+ toString(processors-1) + ".temp"),  (outputScrapFasta+ toString(processors-1) + ".temp"),  (outputMisMatches+ toString(processors-1) + ".temp"), tempFASTAFileNames, processors-1);
 767
 768                 //Wait until all threads have terminated.
 769                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 770
 771                 //Close all thread handles and free memory allocations.
 772                 for(int i=0; i < pDataArray.size(); i++){
 773                         num += pDataArray[i]->count;
 774             if (!pDataArray[i]->done) {
 775                 m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of sequences assigned to it, quitting. \n"); m->control_pressed = true;
 776             }
 777             for (map<string, int>::iterator it = pDataArray[i]->groupCounts.begin(); it != pDataArray[i]->groupCounts.end(); it++) {
 778                 map<string, int>::iterator it2 = groupCounts.find(it->first);
 779                 if (it2 == groupCounts.end()) { groupCounts[it->first] = it->second; }
 780                 else { groupCounts[it->first] += it->second; }
 781             }
 782             for (map<string, string>::iterator it = pDataArray[i]->groupMap.begin(); it != pDataArray[i]->groupMap.end(); it++) {
 783                 map<string, string>::iterator it2 = groupMap.find(it->first);
 784                 if (it2 == groupMap.end()) {    groupMap[it->first] = it->second; }
 785                 else { m->mothurOut("[ERROR]: " + it->first + " is in your fasta file more than once. Sequence names must be unique. please correct.\n");  }
 786             }
 787             CloseHandle(hThreadArray[i]);
 788                         delete pDataArray[i];
 789         }
 790
 791     #endif
 792
 793         for (int i = 0; i < processIDS.size(); i++) {
 794                         m->appendFiles((outputFasta + toString(processIDS[i]) + ".temp"), outputFasta);
 795                         m->mothurRemove((outputFasta + toString(processIDS[i]) + ".temp"));
 796
 797                         m->appendFiles((outputScrapFasta + toString(processIDS[i]) + ".temp"), outputScrapFasta);
 798                         m->mothurRemove((outputScrapFasta + toString(processIDS[i]) + ".temp"));
 799
 800             m->appendFiles((outputMisMatches + toString(processIDS[i]) + ".temp"), outputMisMatches);
 801                         m->mothurRemove((outputMisMatches + toString(processIDS[i]) + ".temp"));
 802
 803             if(allFiles){
 804                                 for(int j=0;j<fastaFileNames.size();j++){
 805                                         for(int k=0;k<fastaFileNames[j].size();k++){
 806                                                 if (fastaFileNames[j][k] != "") {
 807                                                         m->appendFiles((fastaFileNames[j][k] + toString(processIDS[i]) + ".temp"), fastaFileNames[j][k]);
 808                                                         m->mothurRemove((fastaFileNames[j][k] + toString(processIDS[i]) + ".temp"));
 809                                                 }
 810                                         }
 811                                 }
 812                         }
 813                 }
 814
 815                 return num;
 816         }
 817         catch(exception& e) {
 818                 m->errorOut(e, "MakeContigsCommand", "createProcesses");
 819                 exit(1);
 820         }
 821 }
 822 //**********************************************************************************************************************
 823 int MakeContigsCommand::driver(vector<string> files, string outputFasta, string outputScrapFasta, string outputMisMatches, vector<vector<string> > fastaFileNames, int process){
 824     try {
 825
 826         Alignment* alignment;
 827         if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 828                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 829
 830         int num = 0;
 831         string thisffastafile = files[0];
 832         string thisfqualfile = files[1];
 833         string thisrfastafile = files[2];
 834         string thisrqualfile = files[3];
 835
 836         if (m->debug) {  m->mothurOut("[DEBUG]: ffasta = " + thisffastafile + ".\n[DEBUG]: fqual = " + thisfqualfile + ".\n[DEBUG]: rfasta = " + thisrfastafile + ".\n[DEBUG]: rqual = " + thisrqualfile + ".\n"); }
 837
 838         ifstream inFFasta, inRFasta, inFQual, inRQual;
 839         ofstream outFasta, outMisMatch, outScrapFasta;
 840         m->openInputFile(thisffastafile, inFFasta);
 841         m->openInputFile(thisrfastafile, inRFasta);
 842         if (thisfqualfile != "") {
 843             m->openInputFile(thisfqualfile, inFQual);
 844             m->openInputFile(thisrqualfile, inRQual);
 845         }
 846         m->openOutputFile(outputFasta, outFasta);
 847         m->openOutputFile(outputScrapFasta, outScrapFasta);
 848         m->openOutputFile(outputMisMatches, outMisMatch);
 849         if (process == 0) { outMisMatch << "Name\tLength\tOverlap_Length\tOverlap_Start\tOverlap_End\tMisMatches\tNum_Ns\n";  }
 850
 851         TrimOligos trimOligos(pdiffs, bdiffs, 0, 0, primers, barcodes);
 852
 853         while ((!inFFasta.eof()) && (!inRFasta.eof())) {
 854
 855             if (m->control_pressed) { break; }
 856
 857             int success = 1;
 858             string trashCode = "";
 859             int currentSeqsDiffs = 0;
 860
 861             //read seqs and quality info
 862             Sequence fSeq(inFFasta); m->gobble(inFFasta);
 863             Sequence rSeq(inRFasta); m->gobble(inRFasta);
 864             QualityScores* fQual = NULL; QualityScores* rQual = NULL;
 865             if (thisfqualfile != "") {
 866                 fQual = new QualityScores(inFQual); m->gobble(inFQual);
 867                 rQual = new QualityScores(inRQual); m->gobble(inRQual);
 868             }
 869
 870             int barcodeIndex = 0;
 871             int primerIndex = 0;
 872
 873             if(barcodes.size() != 0){
 874                 if (thisfqualfile != "") {
 875                     success = trimOligos.stripBarcode(fSeq, rSeq, *fQual, *rQual, barcodeIndex);
 876                 }else {
 877                     success = trimOligos.stripBarcode(fSeq, rSeq, barcodeIndex);
 878                 }
 879                 if(success > bdiffs)            {       trashCode += 'b';       }
 880                 else{ currentSeqsDiffs += success;  }
 881             }
 882
 883             if(primers.size() != 0){
 884                 if (thisfqualfile != "") {
 885                     success = trimOligos.stripForward(fSeq, rSeq, *fQual, *rQual, primerIndex);
 886                 }else {
 887                     success = trimOligos.stripForward(fSeq, rSeq, primerIndex);
 888                 }
 889                 if(success > pdiffs)            {       trashCode += 'f';       }
 890                 else{ currentSeqsDiffs += success;  }
 891             }
 892
 893             if (currentSeqsDiffs > tdiffs)      {       trashCode += 't';   }
 894
 895             //flip the reverse reads
 896             rSeq.reverseComplement();
 897             if (thisfqualfile != "") { rQual->flipQScores(); }
 898
 899             //pairwise align
 900             alignment->align(fSeq.getUnaligned(), rSeq.getUnaligned());
 901             map<int, int> ABaseMap = alignment->getSeqAAlnBaseMap();
 902             map<int, int> BBaseMap = alignment->getSeqBAlnBaseMap();
 903             fSeq.setAligned(alignment->getSeqAAln());
 904             rSeq.setAligned(alignment->getSeqBAln());
 905             int length = fSeq.getAligned().length();
 906
 907             //traverse alignments merging into one contiguous seq
 908             string contig = "";
 909             int numMismatches = 0;
 910             string seq1 = fSeq.getAligned();
 911             string seq2 = rSeq.getAligned();
 912             vector<int> scores1, scores2;
 913             if (thisfqualfile != "") {
 914                 scores1 = fQual->getQualityScores();
 915                 scores2 = rQual->getQualityScores();
 916                 delete fQual; delete rQual;
 917             }
 918
 919             // if (num < 5) {  cout << fSeq.getStartPos() << '\t' << fSeq.getEndPos() << '\t' << rSeq.getStartPos() << '\t' << rSeq.getEndPos() << endl; }
 920             int overlapStart = fSeq.getStartPos();
 921             int seq2Start = rSeq.getStartPos();
 922             //bigger of the 2 starting positions is the location of the overlapping start
 923             if (overlapStart < seq2Start) { //seq2 starts later so take from 0 to seq2Start from seq1
 924                 overlapStart = seq2Start;
 925                 for (int i = 0; i < overlapStart; i++) { contig += seq1[i];  }
 926             }else { //seq1 starts later so take from 0 to overlapStart from seq2
 927                 for (int i = 0; i < overlapStart; i++) {  contig += seq2[i]; }
 928             }
 929
 930             int seq1End = fSeq.getEndPos();
 931             int seq2End = rSeq.getEndPos();
 932             int overlapEnd = seq1End;
 933             if (seq2End < overlapEnd) { overlapEnd = seq2End; }  //smallest end position is where overlapping ends
 934
 935             int oStart = contig.length();
 936             for (int i = overlapStart; i < overlapEnd; i++) {
 937                 if (seq1[i] == seq2[i]) { //match, add base and choose highest score
 938                     contig += seq1[i];
 939                 }else if (((seq1[i] == '.') || (seq1[i] == '-')) && ((seq2[i] != '-') && (seq2[i] != '.'))) { //seq1 is a gap and seq2 is a base, choose seq2, unless quality score for base is below insert. In that case eliminate base
 940                     if (thisfqualfile != "") {
 941                         if (scores2[BBaseMap[i]] < insert) { } //
 942                         else { contig += seq2[i];  }
 943                     }else { contig += seq2[i]; } //with no quality info, then we keep it?
 944                 }else if (((seq2[i] == '.') || (seq2[i] == '-')) && ((seq1[i] != '-') && (seq1[i] != '.'))) { //seq2 is a gap and seq1 is a base, choose seq1, unless quality score for base is below insert. In that case eliminate base
 945                     if (thisfqualfile != "") {
 946                         if (scores1[ABaseMap[i]] < insert) { } //
 947                         else { contig += seq1[i];  }
 948                     }else { contig += seq1[i]; } //with no quality info, then we keep it?
 949                 }else if (((seq1[i] != '-') && (seq1[i] != '.')) && ((seq2[i] != '-') && (seq2[i] != '.'))) { //both bases choose one with better quality
 950                     if (thisfqualfile != "") {
 951                         if (abs(scores1[ABaseMap[i]] - scores2[BBaseMap[i]]) >= deltaq) { //is the difference in qual scores >= deltaq, if yes choose base with higher score
 952                             char c = seq1[i];
 953                             if (scores1[ABaseMap[i]] < scores2[BBaseMap[i]]) { c = seq2[i]; }
 954                             contig += c;
 955                         }else { //if no, base becomes n
 956                             contig += 'N';
 957                         }
 958                         numMismatches++;
 959                     }else { numMismatches++; } //cant decide, so eliminate and mark as mismatch
 960                 }else { //should never get here
 961                     m->mothurOut("[ERROR]: case I didn't think of seq1 = " + toString(seq1[i]) + " and seq2 = " + toString(seq2[i]) + "\n");
 962                 }
 963             }
 964             int oend = contig.length();
 965             if (seq1End < seq2End) { //seq1 ends before seq2 so take from overlap to length from seq2
 966                 for (int i = overlapEnd; i < length; i++) { contig += seq2[i];  }
 967             }else { //seq2 ends before seq1 so take from overlap to length from seq1
 968                 for (int i = overlapEnd; i < length; i++) {  contig += seq1[i]; }
 969             }
 970
 971             if(trashCode.length() == 0){
 972                 bool ignore = false;
 973
 974                 if (m->debug) { m->mothurOut(fSeq.getName()); }
 975
 976                 if (createGroup) {
 977                     if(barcodes.size() != 0){
 978                         string thisGroup = barcodeNameVector[barcodeIndex];
 979                         if (primers.size() != 0) {
 980                             if (primerNameVector[primerIndex] != "") {
 981                                 if(thisGroup != "") {
 982                                     thisGroup += "." + primerNameVector[primerIndex];
 983                                 }else {
 984                                     thisGroup = primerNameVector[primerIndex];
 985                                 }
 986                             }
 987                         }
 988
 989                         if (m->debug) { m->mothurOut(", group= " + thisGroup + "\n"); }
 990
 991                         int pos = thisGroup.find("ignore");
 992                         if (pos == string::npos) {
 993                             groupMap[fSeq.getName()] = thisGroup;
 994
 995                             map<string, int>::iterator it = groupCounts.find(thisGroup);
 996                             if (it == groupCounts.end()) {      groupCounts[thisGroup] = 1; }
 997                             else { groupCounts[it->first] ++; }
 998                         }else { ignore = true; }
 999
1000                     }
1001                 }
1002                 if (m->debug) { m->mothurOut("\n"); }
1003
1004                 if(allFiles && !ignore){
1005                     ofstream output;
1006                     m->openOutputFileAppend(fastaFileNames[barcodeIndex][primerIndex], output);
1007                     output << ">" << fSeq.getName() << endl << contig << endl;
1008                     output.close();
1009                 }
1010
1011                 //output
1012                 outFasta << ">" << fSeq.getName() << endl << contig << endl;
1013                 int numNs = 0;
1014                 for (int i = 0; i < contig.length(); i++) { if (contig[i] == 'N') { numNs++; }  }
1015                 outMisMatch << fSeq.getName() << '\t' << contig.length() << '\t' << (oend-oStart) << '\t' << oStart << '\t' << oend << '\t' << numMismatches << '\t' << numNs << endl;
1016             }else {
1017                 //output
1018                 outScrapFasta << ">" << fSeq.getName() << " | " << trashCode << endl << contig << endl;
1019             }
1020             num++;
1021
1022                         //report progress
1023                         if((num) % 1000 == 0){  m->mothurOut(toString(num)); m->mothurOutEndLine();             }
1024                 }
1025
1026                 //report progress
1027                 if((num) % 1000 != 0){  m->mothurOut(toString(num)); m->mothurOutEndLine();             }
1028
1029         inFFasta.close();
1030         inRFasta.close();
1031         outFasta.close();
1032         outScrapFasta.close();
1033         outMisMatch.close();
1034         if (thisfqualfile != "") {
1035             inFQual.close();
1036             inRQual.close();
1037         }
1038         delete alignment;
1039
1040         if (m->control_pressed) {  m->mothurRemove(outputFasta); m->mothurRemove(outputScrapFasta);m->mothurRemove(outputMisMatches);  }
1041
1042         return num;
1043     }
1044         catch(exception& e) {
1045                 m->errorOut(e, "MakeContigsCommand", "driver");
1046                 exit(1);
1047         }
1048 }
1049 //**********************************************************************************************************************
1050 vector< vector<string> > MakeContigsCommand::readFastqFiles(unsigned long int& count, string ffastq, string rfastq){
1051     try {
1052         vector< vector<string> > files;
1053         //maps processors number to file pointer
1054         map<int, vector<ofstream*> > tempfiles;  //tempfiles[0] = forwardFasta, [1] = forwardQual, [2] = reverseFasta, [3] = reverseQual
1055         map<int, vector<ofstream*> >::iterator it;
1056
1057         //create files to write to
1058         for (int i = 0; i < processors; i++) {
1059             vector<ofstream*> temp;
1060             ofstream* outFF = new ofstream;     temp.push_back(outFF);
1061             ofstream* outFQ = new ofstream;     temp.push_back(outFQ);
1062             ofstream* outRF = new ofstream;     temp.push_back(outRF);
1063             ofstream* outRQ = new ofstream;     temp.push_back(outRQ);
1064             tempfiles[i] = temp;
1065
1066             vector<string> names;
1067             string thisOutputDir = outputDir;
1068             if (outputDir == "") { thisOutputDir = m->hasPath(ffastq); }
1069             string ffastafilename = thisOutputDir + m->getRootName(m->getSimpleName(ffastq)) + toString(i) + "ffastatemp";
1070             string rfastafilename = thisOutputDir + m->getRootName(m->getSimpleName(rfastq)) + toString(i) + "rfastatemp";
1071             string fqualfilename = thisOutputDir + m->getRootName(m->getSimpleName(ffastq)) + toString(i) + "fqualtemp";
1072             string rqualfilename = thisOutputDir + m->getRootName(m->getSimpleName(rfastq)) + toString(i) + "rqualtemp";
1073             names.push_back(ffastafilename); names.push_back(fqualfilename);
1074             names.push_back(rfastafilename); names.push_back(rqualfilename);
1075             files.push_back(names);
1076
1077             m->openOutputFile(ffastafilename, *outFF);
1078             m->openOutputFile(rfastafilename, *outRF);
1079             m->openOutputFile(fqualfilename, *outFQ);
1080             m->openOutputFile(rqualfilename, *outRQ);
1081         }
1082
1083         if (m->control_pressed) {
1084             //close files, delete ofstreams
1085             for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
1086             //remove files
1087             for (int i = 0; i < files.size(); i++) {
1088                 for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); }
1089             }
1090         }
1091
1092         ifstream inForward;
1093         m->openInputFile(ffastq, inForward);
1094
1095         ifstream inReverse;
1096         m->openInputFile(rfastq, inReverse);
1097
1098         count = 0;
1099         map<string, fastqRead> uniques;
1100         map<string, fastqRead>::iterator itUniques;
1101         while ((!inForward.eof()) || (!inReverse.eof())) {
1102
1103             if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inForward.close(); inReverse.close(); return files; }
1104
1105             //get a read from forward and reverse fastq files
1106             bool ignoref, ignorer;
1107             fastqRead thisFread, thisRread;
1108             if (!inForward.eof()) {  thisFread = readFastq(inForward, ignoref); }
1109             else { ignoref = true; }
1110             if (!inReverse.eof()) { thisRread = readFastq(inReverse, ignorer);  }
1111             else { ignorer = true; }
1112
1113             vector<pairFastqRead> reads = getReads(ignoref, ignorer, thisFread, thisRread, uniques);
1114
1115             for (int i = 0; i < reads.size(); i++) {
1116                 fastqRead fread = reads[i].forward;
1117                 fastqRead rread = reads[i].reverse;
1118
1119                 if (m->debug) { m->mothurOut(toString(count) + '\t' + fread.name + '\t' + rread.name + '\n'); }
1120
1121                 //if (checkReads(fread, rread, ffastq, rfastq)) {
1122                     if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inForward.close(); inReverse.close(); return files; }
1123
1124                     //if the reads are okay write to output files
1125                     int process = count % processors;
1126
1127                     *(tempfiles[process][0]) << ">" << fread.name << endl << fread.sequence << endl;
1128                     *(tempfiles[process][1]) << ">" << fread.name << endl;
1129                     for (int i = 0; i < fread.scores.size(); i++) { *(tempfiles[process][1]) << fread.scores[i] << " "; }
1130                     *(tempfiles[process][1]) << endl;
1131                     *(tempfiles[process][2]) << ">" << rread.name << endl << rread.sequence << endl;
1132                     *(tempfiles[process][3]) << ">" << rread.name << endl;
1133                     for (int i = 0; i < rread.scores.size(); i++) { *(tempfiles[process][3]) << rread.scores[i] << " "; }
1134                     *(tempfiles[process][3]) << endl;
1135
1136                     count++;
1137
1138                     //report progress
1139                     if((count) % 10000 == 0){   m->mothurOut(toString(count)); m->mothurOutEndLine();           }
1140                 //}
1141             }
1142                 }
1143                 //report progress
1144                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
1145
1146         if (uniques.size() != 0) {
1147             for (itUniques = uniques.begin(); itUniques != uniques.end(); itUniques++) {
1148                 m->mothurOut("[WARNING]: did not find paired read for " + itUniques->first + ", ignoring.\n");
1149             }
1150             m->mothurOutEndLine();
1151         }
1152
1153         //close files, delete ofstreams
1154         for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
1155         inForward.close();
1156         inReverse.close();
1157
1158         return files;
1159     }
1160     catch(exception& e) {
1161         m->errorOut(e, "MakeContigsCommand", "readFastqFiles");
1162         exit(1);
1163     }
1164 }
1165 //**********************************************************************************************************************
1166 vector< vector<string> > MakeContigsCommand::readFastaFiles(unsigned long int& count, string ffasta, string rfasta){
1167     try {
1168         vector< vector<string> > files;
1169         //maps processors number to file pointer
1170         map<int, vector<ofstream*> > tempfiles;  //tempfiles[0] = forwardFasta, [1] = forwardQual, [2] = reverseFasta, [3] = reverseQual
1171         map<int, vector<ofstream*> >::iterator it;
1172
1173         //create files to write to
1174         for (int i = 0; i < processors; i++) {
1175             vector<ofstream*> temp;
1176             ofstream* outFF = new ofstream;     temp.push_back(outFF);
1177             ofstream* outFQ = new ofstream;     temp.push_back(outFQ);
1178             ofstream* outRF = new ofstream;     temp.push_back(outRF);
1179             ofstream* outRQ = new ofstream;     temp.push_back(outRQ);
1180             tempfiles[i] = temp;
1181
1182             vector<string> names;
1183             string thisOutputDir = outputDir;
1184             if (outputDir == "") { thisOutputDir = m->hasPath(ffasta); }
1185             string ffastafilename = thisOutputDir + m->getRootName(m->getSimpleName(ffasta)) + toString(i) + "ffastatemp";
1186             string rfastafilename = thisOutputDir + m->getRootName(m->getSimpleName(rfasta)) + toString(i) + "rfastatemp";
1187             string fqualfilename = "";
1188             if (fqualfile != "") { fqualfilename = thisOutputDir + m->getRootName(m->getSimpleName(fqualfile)) + toString(i) + "fqual.temp";  m->openOutputFile(fqualfilename, *outFQ); }
1189             string rqualfilename = "";
1190             if (rqualfile != "") { rqualfilename = thisOutputDir + m->getRootName(m->getSimpleName(rqualfile)) + toString(i) + "rqual.temp"; m->openOutputFile(rqualfilename, *outRQ); }
1191             names.push_back(ffastafilename); names.push_back(fqualfilename);
1192             names.push_back(rfastafilename); names.push_back(rqualfilename);
1193             files.push_back(names);
1194
1195             m->openOutputFile(ffastafilename, *outFF);
1196             m->openOutputFile(rfastafilename, *outRF);
1197         }
1198
1199         if (m->control_pressed) {
1200             //close files, delete ofstreams
1201             for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
1202             //remove files
1203             for (int i = 0; i < files.size(); i++) {
1204                 for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); }
1205             }
1206         }
1207
1208         ifstream inForwardFasta;
1209         m->openInputFile(ffasta, inForwardFasta);
1210
1211         ifstream inReverseFasta;
1212         m->openInputFile(rfasta, inReverseFasta);
1213
1214         ifstream inForwardQual, inReverseQual;
1215         if (fqualfile != "") { m->openInputFile(fqualfile, inForwardQual); m->openInputFile(rqualfile, inReverseQual); }
1216
1217         count = 0;
1218         map<string, fastqRead> uniques;
1219         map<string, fastqRead>::iterator itUniques;
1220         while ((!inForwardFasta.eof()) || (!inReverseFasta.eof())) {
1221
1222             if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inReverseFasta.close(); inForwardFasta.close(); if (fqualfile != "") { inReverseQual.close(); inReverseQual.close(); } return files; }
1223
1224             //get a reads from forward and reverse fasta files
1225             bool ignoref, ignorer;
1226             fastqRead thisFread, thisRread;
1227             if (!inForwardFasta.eof()) {
1228                 ignoref = false;
1229                 Sequence temp(inForwardFasta); m->gobble(inForwardFasta);
1230                 thisFread.name = temp.getName();
1231                 thisFread.sequence = temp.getUnaligned();
1232             }else { ignoref = true; }
1233             if (!inReverseFasta.eof()) {
1234                 ignorer = false;
1235                 Sequence temp(inReverseFasta); m->gobble(inReverseFasta);
1236                 thisRread.name = temp.getName();
1237                 thisRread.sequence = temp.getUnaligned();
1238             }else { ignorer = true; }
1239
1240             //get qual reads if given
1241             if (fqualfile != "") {
1242                 if (!inForwardQual.eof() && !ignoref) {
1243                     QualityScores temp(inForwardQual); m->gobble(inForwardQual);
1244                     //if forward files dont match ignore read
1245                     if (thisFread.name != temp.getName()) { ignoref = true; }
1246                     else { thisFread.scores = temp.getQualityScores(); }
1247                 }else { ignoref = true; }
1248                 if (!inReverseQual.eof() && !ignorer) {
1249                     QualityScores temp(inReverseQual); m->gobble(inReverseQual);
1250                     //if reverse files dont match ignore read
1251                     if (thisRread.name != temp.getName()) { ignorer = true; }
1252                     else { thisRread.scores = temp.getQualityScores(); }
1253                 }else { ignorer = true; }
1254             }
1255
1256             vector<pairFastqRead> reads = getReads(ignoref, ignorer, thisFread, thisRread, uniques);
1257
1258             for (int i = 0; i < reads.size(); i++) {
1259                 fastqRead fread = reads[i].forward;
1260                 fastqRead rread = reads[i].reverse;
1261
1262                 if (m->debug) { m->mothurOut(toString(count) + '\t' + fread.name + '\t' + rread.name + '\n'); }
1263
1264                // if (checkReads(fread, rread, ffasta, rfasta)) {
1265                     if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inReverseFasta.close(); inForwardFasta.close(); if (fqualfile != "") { inReverseQual.close(); inReverseQual.close(); } return files; }
1266
1267                     //if the reads are okay write to output files
1268                     int process = count % processors;
1269
1270                     *(tempfiles[process][0]) << ">" << fread.name << endl << fread.sequence << endl;
1271                     *(tempfiles[process][2]) << ">" << rread.name << endl << rread.sequence << endl;
1272                     if (fqualfile != "") { //if you have quality info, print it
1273                         *(tempfiles[process][1]) << ">" << fread.name << endl;
1274                         for (int i = 0; i < fread.scores.size(); i++) { *(tempfiles[process][1]) << fread.scores[i] << " "; }
1275                         *(tempfiles[process][1]) << endl;
1276                         *(tempfiles[process][3]) << ">" << rread.name << endl;
1277                         for (int i = 0; i < rread.scores.size(); i++) { *(tempfiles[process][3]) << rread.scores[i] << " "; }
1278                         *(tempfiles[process][3]) << endl;
1279                     }
1280                     count++;
1281
1282                     //report progress
1283                     if((count) % 10000 == 0){   m->mothurOut(toString(count)); m->mothurOutEndLine();           }
1284                 //}
1285             }
1286                 }
1287                 //report progress
1288                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
1289
1290         if (uniques.size() != 0) {
1291             for (itUniques = uniques.begin(); itUniques != uniques.end(); itUniques++) {
1292                 m->mothurOut("[WARNING]: did not find paired read for " + itUniques->first + ", ignoring.\n");
1293             }
1294             m->mothurOutEndLine();
1295         }
1296
1297         //close files, delete ofstreams
1298         for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
1299         inReverseFasta.close();
1300         inForwardFasta.close();
1301         if (fqualfile != "") { inReverseQual.close(); inReverseQual.close(); }
1302
1303         return files;
1304     }
1305     catch(exception& e) {
1306         m->errorOut(e, "MakeContigsCommand", "readFastaFiles");
1307         exit(1);
1308     }
1309 }
1310 //**********************************************************************************************************************
1311 vector<pairFastqRead> MakeContigsCommand::getReads(bool ignoref, bool ignorer, fastqRead forward, fastqRead reverse, map<string, fastqRead>& uniques){
1312     try {
1313         vector<pairFastqRead> reads;
1314         map<string, fastqRead>::iterator itUniques;
1315
1316         if (!ignoref && !ignorer) {
1317             if (forward.name == reverse.name) {
1318                 pairFastqRead temp(forward, reverse);
1319                 reads.push_back(temp);
1320             }else {
1321                 bool match = false;
1322                 //if no match are the names only different by 1 and 2?
1323                 string tempFRead = forward.name.substr(0, forward.name.length()-1);
1324                 string tempRRead = reverse.name.substr(0, reverse.name.length()-1);
1325                 if (tempFRead == tempRRead) {
1326                     if ((forward.name[forward.name.length()-1] == '1') && (reverse.name[reverse.name.length()-1] == '2')) {
1327                         forward.name = tempFRead;
1328                         reverse.name = tempRRead;
1329                         pairFastqRead temp(forward, reverse);
1330                         reads.push_back(temp);
1331                         match = true;
1332                     }
1333                 }
1334
1335                 if (!match) {
1336                     //look for forward pair
1337                     itUniques = uniques.find(forward.name);
1338                     if (itUniques != uniques.end()) {  //we have the pair for this read
1339                         pairFastqRead temp(forward, itUniques->second);
1340                         reads.push_back(temp);
1341                         uniques.erase(itUniques);
1342                     }else { //save this read for later
1343                         uniques[forward.name] = forward;
1344                     }
1345
1346                     //look for reverse pair
1347                     itUniques = uniques.find(reverse.name);
1348                     if (itUniques != uniques.end()) {  //we have the pair for this read
1349                         pairFastqRead temp(itUniques->second, reverse);
1350                         reads.push_back(temp);
1351                         uniques.erase(itUniques);
1352                     }else { //save this read for later
1353                         uniques[reverse.name] = reverse;
1354                     }
1355                 }
1356
1357             }
1358         }else if (!ignoref && ignorer) { //ignore reverse keep forward
1359             //look for forward pair
1360             itUniques = uniques.find(forward.name);
1361             if (itUniques != uniques.end()) {  //we have the pair for this read
1362                 pairFastqRead temp(forward, itUniques->second);
1363                 reads.push_back(temp);
1364                 uniques.erase(itUniques);
1365             }else { //save this read for later
1366                 uniques[forward.name] = forward;
1367             }
1368
1369         }else if (ignoref && !ignorer) { //ignore forward keep reverse
1370             //look for reverse pair
1371             itUniques = uniques.find(reverse.name);
1372             if (itUniques != uniques.end()) {  //we have the pair for this read
1373                 pairFastqRead temp(itUniques->second, reverse);
1374                 reads.push_back(temp);
1375                 uniques.erase(itUniques);
1376             }else { //save this read for later
1377                 uniques[reverse.name] = reverse;
1378             }
1379         }//else ignore both and do nothing
1380
1381         return reads;
1382     }
1383     catch(exception& e) {
1384         m->errorOut(e, "MakeContigsCommand", "readFastqFiles");
1385         exit(1);
1386     }
1387 }
1388 //**********************************************************************************************************************
1389 fastqRead MakeContigsCommand::readFastq(ifstream& in, bool& ignore){
1390     try {
1391         fastqRead read;
1392
1393         ignore = false;
1394
1395         //read sequence name
1396         string line = m->getline(in); m->gobble(in);
1397         vector<string> pieces = m->splitWhiteSpace(line);
1398         string name = "";  if (pieces.size() != 0) { name = pieces[0]; }
1399         if (name == "") {  m->mothurOut("[WARNING]: Blank fasta name, ignoring read."); m->mothurOutEndLine(); ignore=true;  }
1400         else if (name[0] != '@') { m->mothurOut("[WARNING]: reading " + name + " expected a name with @ as a leading character, ignoring read."); m->mothurOutEndLine(); ignore=true; }
1401         else { name = name.substr(1); }
1402
1403         //read sequence
1404         string sequence = m->getline(in); m->gobble(in);
1405         if (sequence == "") {  m->mothurOut("[WARNING]: missing sequence for " + name + ", ignoring."); ignore=true; }
1406
1407         //read sequence name
1408         line = m->getline(in); m->gobble(in);
1409         pieces = m->splitWhiteSpace(line);
1410         string name2 = "";  if (pieces.size() != 0) { name2 = pieces[0]; }
1411         if (name2 == "") {  m->mothurOut("[WARNING]: expected a name with + as a leading character, ignoring."); ignore=true; }
1412         else if (name2[0] != '+') { m->mothurOut("[WARNING]: reading " + name2 + " expected a name with + as a leading character, ignoring."); ignore=true; }
1413         else { name2 = name2.substr(1); if (name2 == "") { name2 = name; } }
1414
1415         //read quality scores
1416         string quality = m->getline(in); m->gobble(in);
1417         if (quality == "") {  m->mothurOut("[WARNING]: missing quality for " + name2 + ", ignoring."); ignore=true; }
1418
1419         //sanity check sequence length and number of quality scores match
1420         if (name2 != "") { if (name != name2) { m->mothurOut("[WARNING]: names do not match. read " + name + " for fasta and " + name2 + " for quality, ignoring."); ignore=true; } }
1421         if (quality.length() != sequence.length()) { m->mothurOut("[WARNING]: Lengths do not match for sequence " + name + ". Read " + toString(sequence.length()) + " characters for fasta and " + toString(quality.length()) + " characters for quality scores, ignoring read."); ignore=true; }
1422
1423         vector<int> qualScores = convertQual(quality);
1424
1425         read.name = name;
1426         read.sequence = sequence;
1427         read.scores = qualScores;
1428
1429         return read;
1430     }
1431     catch(exception& e) {
1432         m->errorOut(e, "MakeContigsCommand", "readFastq");
1433         exit(1);
1434     }
1435 }
1436 /**********************************************************************************************************************
1437 bool MakeContigsCommand::checkReads(fastqRead& forward, fastqRead& reverse, string ffile, string rfile){
1438     try {
1439         bool good = true;
1440
1441         //do sequence lengths match
1442         if (forward.sequence.length() != reverse.sequence.length()) {
1443             m->mothurOut("[WARNING]: For sequence " + forward.name + " I read a sequence of length " + toString(forward.sequence.length()) + " from " + ffile + ", but read a sequence of length " + toString(reverse.sequence.length()) + " from " + rfile + ", ignoring.\n");
1444             good = false;
1445         }
1446
1447         //do number of qual scores match
1448         if (forward.scores.size() != reverse.scores.size()) {
1449             m->mothurOut("[WARNING]: For sequence " + forward.name + " I read " + toString(forward.scores.size()) + " quality scores from " + ffile + ", but read  " + toString(reverse.scores.size()) + " quality scores from " + rfile + ", ignoring.\n");
1450             good = false;
1451         }
1452
1453         return good;
1454     }
1455     catch(exception& e) {
1456         m->errorOut(e, "MakeContigsCommand", "checkReads");
1457         exit(1);
1458     }
1459 }*/
1460 //***************************************************************************************************************
1461 vector< vector<string> > MakeContigsCommand::readFileNames(string filename){
1462         try {
1463         vector< vector<string> > files;
1464         string forward, reverse;
1465
1466         ifstream in;
1467         m->openInputFile(filename, in);
1468
1469         while(!in.eof()) {
1470
1471             if (m->control_pressed) { return files; }
1472
1473             in >> forward; m->gobble(in);
1474             in >> reverse; m->gobble(in);
1475
1476             //check to make sure both are able to be opened
1477             ifstream in2;
1478             int openForward = m->openInputFile(forward, in2, "noerror");
1479
1480             //if you can't open it, try default location
1481             if (openForward == 1) {
1482                 if (m->getDefaultPath() != "") { //default path is set
1483                     string tryPath = m->getDefaultPath() + m->getSimpleName(forward);
1484                     m->mothurOut("Unable to open " + forward + ". Trying default " + tryPath); m->mothurOutEndLine();
1485                     ifstream in3;
1486                     openForward = m->openInputFile(tryPath, in3, "noerror");
1487                     in3.close();
1488                     forward = tryPath;
1489                 }
1490             }
1491
1492             //if you can't open it, try output location
1493             if (openForward == 1) {
1494                 if (m->getOutputDir() != "") { //default path is set
1495                     string tryPath = m->getOutputDir() + m->getSimpleName(forward);
1496                     m->mothurOut("Unable to open " + forward + ". Trying output directory " + tryPath); m->mothurOutEndLine();
1497                     ifstream in4;
1498                     openForward = m->openInputFile(tryPath, in4, "noerror");
1499                     forward = tryPath;
1500                     in4.close();
1501                 }
1502             }
1503
1504             if (openForward == 1) { //can't find it
1505                 m->mothurOut("[WARNING]: can't find " + forward + ", ignoring pair.\n");
1506             }else{  in2.close();  }
1507
1508             ifstream in3;
1509             int openReverse = m->openInputFile(reverse, in3, "noerror");
1510
1511             //if you can't open it, try default location
1512             if (openReverse == 1) {
1513                 if (m->getDefaultPath() != "") { //default path is set
1514                     string tryPath = m->getDefaultPath() + m->getSimpleName(reverse);
1515                     m->mothurOut("Unable to open " + reverse + ". Trying default " + tryPath); m->mothurOutEndLine();
1516                     ifstream in3;
1517                     openReverse = m->openInputFile(tryPath, in3, "noerror");
1518                     in3.close();
1519                     reverse = tryPath;
1520                 }
1521             }
1522
1523             //if you can't open it, try output location
1524             if (openReverse == 1) {
1525                 if (m->getOutputDir() != "") { //default path is set
1526                     string tryPath = m->getOutputDir() + m->getSimpleName(reverse);
1527                     m->mothurOut("Unable to open " + reverse + ". Trying output directory " + tryPath); m->mothurOutEndLine();
1528                     ifstream in4;
1529                     openReverse = m->openInputFile(tryPath, in4, "noerror");
1530                     reverse = tryPath;
1531                     in4.close();
1532                 }
1533             }
1534
1535             if (openReverse == 1) { //can't find it
1536                 m->mothurOut("[WARNING]: can't find " + reverse + ", ignoring pair.\n");
1537             }else{  in3.close();  }
1538
1539             if ((openForward != 1) && (openReverse != 1)) { //good pair
1540                 vector<string> pair;
1541                 pair.push_back(forward);
1542                 pair.push_back(reverse);
1543                 files.push_back(pair);
1544             }
1545
1546         }
1547         in.close();
1548
1549         return files;
1550     }
1551     catch(exception& e) {
1552         m->errorOut(e, "MakeContigsCommand", "checkReads");
1553         exit(1);
1554     }
1555 }
1556 //***************************************************************************************************************
1557 //illumina data requires paired forward and reverse data
1558 //BARCODE   atgcatgc   atgcatgc    groupName
1559 //PRIMER   atgcatgc   atgcatgc    groupName
1560 //PRIMER   atgcatgc   atgcatgc
1561 bool MakeContigsCommand::getOligos(vector<vector<string> >& fastaFileNames, string rootname){
1562         try {
1563                 ifstream in;
1564                 m->openInputFile(oligosfile, in);
1565
1566                 ofstream test;
1567
1568                 string type, foligo, roligo, group;
1569
1570                 int indexPrimer = 0;
1571                 int indexBarcode = 0;
1572         set<string> uniquePrimers;
1573         set<string> uniqueBarcodes;
1574
1575                 while(!in.eof()){
1576
1577                         in >> type;
1578
1579                         if (m->debug) { m->mothurOut("[DEBUG]: reading type - " + type + ".\n"); }
1580
1581                         if(type[0] == '#'){
1582                                 while (!in.eof())       {       char c = in.get();  if (c == 10 || c == 13){    break;  }       } // get rest of line if there's any crap there
1583                                 m->gobble(in);
1584                         }
1585                         else{
1586                                 m->gobble(in);
1587                                 //make type case insensitive
1588                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
1589
1590                                 in >> foligo;
1591
1592                 if (m->debug) { m->mothurOut("[DEBUG]: reading - " + foligo + ".\n"); }
1593
1594                                 for(int i=0;i<foligo.length();i++){
1595                                         foligo[i] = toupper(foligo[i]);
1596                                         if(foligo[i] == 'U')    {       foligo[i] = 'T';        }
1597                                 }
1598
1599                                 if(type == "FORWARD"){
1600                                         m->gobble(in);
1601
1602                     in >> roligo;
1603
1604                     for(int i=0;i<roligo.length();i++){
1605                         roligo[i] = toupper(roligo[i]);
1606                         if(roligo[i] == 'U')    {       roligo[i] = 'T';        }
1607                     }
1608                     //roligo = reverseOligo(roligo);
1609
1610                     group = "";
1611
1612                                         // get rest of line in case there is a primer name
1613                                         while (!in.eof())       {
1614                                                 char c = in.get();
1615                                                 if (c == 10 || c == 13){        break;  }
1616                                                 else if (c == 32 || c == 9){;} //space or tab
1617                                                 else {  group += c;  }
1618                                         }
1619
1620                     oligosPair newPrimer(foligo, roligo);
1621
1622                                         //check for repeat barcodes
1623                     string tempPair = foligo+roligo;
1624                     if (uniquePrimers.count(tempPair) != 0) { m->mothurOut("primer pair " + newPrimer.forward + " " + newPrimer.reverse + " is in your oligos file already."); m->mothurOutEndLine();  }
1625                     else { uniquePrimers.insert(tempPair); }
1626
1627                     if (m->debug) {  if (group != "") { m->mothurOut("[DEBUG]: reading group " + group + ".\n"); }else{ m->mothurOut("[DEBUG]: no group for primer pair " + newPrimer.forward + " " + newPrimer.reverse + ".\n"); }  }
1628
1629                                         primers[indexPrimer]=newPrimer; indexPrimer++;
1630                                         primerNameVector.push_back(group);
1631                                 }else if(type == "BARCODE"){
1632                                         m->gobble(in);
1633
1634                     in >> roligo;
1635
1636                     for(int i=0;i<roligo.length();i++){
1637                         roligo[i] = toupper(roligo[i]);
1638                         if(roligo[i] == 'U')    {       roligo[i] = 'T';        }
1639                     }
1640                     //roligo = reverseOligo(roligo);
1641
1642                     oligosPair newPair(foligo, roligo);
1643
1644                     group = "";
1645                     while (!in.eof())   {
1646                                                 char c = in.get();
1647                                                 if (c == 10 || c == 13){        break;  }
1648                                                 else if (c == 32 || c == 9){;} //space or tab
1649                                                 else {  group += c;  }
1650                                         }
1651
1652                     if (m->debug) { m->mothurOut("[DEBUG]: barcode pair " + newPair.forward + " " + newPair.reverse + ", and group = " + group + ".\n"); }
1653
1654                     //check for repeat barcodes
1655                     string tempPair = foligo+roligo;
1656                     if (uniqueBarcodes.count(tempPair) != 0) { m->mothurOut("barcode pair " + newPair.forward + " " + newPair.reverse +  " is in your oligos file already, disregarding."); m->mothurOutEndLine();  }
1657                     else { uniqueBarcodes.insert(tempPair); }
1658
1659                     barcodes[indexBarcode]=newPair; indexBarcode++;
1660                                         barcodeNameVector.push_back(group);
1661                                 }else if(type == "LINKER"){
1662                                         linker.push_back(foligo);
1663                     m->mothurOut("[WARNING]: make.contigs is not setup to remove linkers, ignoring.\n");
1664                                 }else if(type == "SPACER"){
1665                                         spacer.push_back(foligo);
1666                     m->mothurOut("[WARNING]: make.contigs is not setup to remove spacers, ignoring.\n");
1667                                 }
1668                                 else{   m->mothurOut("[WARNING]: " + type + " is not recognized as a valid type. Choices are primer, barcode, linker and spacer. Ignoring " + foligo + "."); m->mothurOutEndLine(); }
1669                         }
1670                         m->gobble(in);
1671                 }
1672                 in.close();
1673
1674                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
1675
1676                 //add in potential combos
1677                 if(barcodeNameVector.size() == 0){
1678             oligosPair temp("", "");
1679                         barcodes[0] = temp;
1680                         barcodeNameVector.push_back("");
1681                 }
1682
1683                 if(primerNameVector.size() == 0){
1684             oligosPair temp("", "");
1685                         primers[0] = temp;
1686                         primerNameVector.push_back("");
1687                 }
1688
1689                 fastaFileNames.resize(barcodeNameVector.size());
1690                 for(int i=0;i<fastaFileNames.size();i++){
1691                         fastaFileNames[i].assign(primerNameVector.size(), "");
1692                 }
1693
1694                 if(allFiles){
1695                         set<string> uniqueNames; //used to cleanup outputFileNames
1696                         for(map<int, oligosPair>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
1697                                 for(map<int, oligosPair>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
1698
1699                                         string primerName = primerNameVector[itPrimer->first];
1700                                         string barcodeName = barcodeNameVector[itBar->first];
1701
1702                     if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
1703                                         else {
1704                         string comboGroupName = "";
1705                         string fastaFileName = "";
1706                         string qualFileName = "";
1707                         string nameFileName = "";
1708                         string countFileName = "";
1709
1710                         if(primerName == ""){
1711                             comboGroupName = barcodeNameVector[itBar->first];
1712                         }
1713                         else{
1714                             if(barcodeName == ""){
1715                                 comboGroupName = primerNameVector[itPrimer->first];
1716                             }
1717                             else{
1718                                 comboGroupName = barcodeNameVector[itBar->first] + "." + primerNameVector[itPrimer->first];
1719                             }
1720                         }
1721
1722
1723                         ofstream temp;
1724                         fastaFileName = rootname + comboGroupName + ".fasta";
1725                         if (uniqueNames.count(fastaFileName) == 0) {
1726                             outputNames.push_back(fastaFileName);
1727                             outputTypes["fasta"].push_back(fastaFileName);
1728                             uniqueNames.insert(fastaFileName);
1729                         }
1730
1731                         fastaFileNames[itBar->first][itPrimer->first] = fastaFileName;
1732                         m->openOutputFile(fastaFileName, temp);         temp.close();
1733                     }
1734                                 }
1735                         }
1736                 }
1737
1738                 bool allBlank = true;
1739                 for (int i = 0; i < barcodeNameVector.size(); i++) {
1740                         if (barcodeNameVector[i] != "") {
1741                                 allBlank = false;
1742                                 break;
1743                         }
1744                 }
1745                 for (int i = 0; i < primerNameVector.size(); i++) {
1746                         if (primerNameVector[i] != "") {
1747                                 allBlank = false;
1748                                 break;
1749                         }
1750                 }
1751
1752                 if (allBlank) {
1753                         m->mothurOut("[WARNING]: your oligos file does not contain any group names.  mothur will not create a groupfile."); m->mothurOutEndLine();
1754                         allFiles = false;
1755                         return false;
1756                 }
1757
1758                 return true;
1759
1760         }
1761         catch(exception& e) {
1762                 m->errorOut(e, "MakeContigsCommand", "getOligos");
1763                 exit(1);
1764         }
1765 }
1766 //********************************************************************/
1767 string MakeContigsCommand::reverseOligo(string oligo){
1768         try {
1769         string reverse = "";
1770
1771         for(int i=oligo.length()-1;i>=0;i--){
1772
1773             if(oligo[i] == 'A')         {       reverse += 'T'; }
1774             else if(oligo[i] == 'T'){   reverse += 'A'; }
1775             else if(oligo[i] == 'U'){   reverse += 'A'; }
1776
1777             else if(oligo[i] == 'G'){   reverse += 'C'; }
1778             else if(oligo[i] == 'C'){   reverse += 'G'; }
1779
1780             else if(oligo[i] == 'R'){   reverse += 'Y'; }
1781             else if(oligo[i] == 'Y'){   reverse += 'R'; }
1782
1783             else if(oligo[i] == 'M'){   reverse += 'K'; }
1784             else if(oligo[i] == 'K'){   reverse += 'M'; }
1785
1786             else if(oligo[i] == 'W'){   reverse += 'W'; }
1787             else if(oligo[i] == 'S'){   reverse += 'S'; }
1788
1789             else if(oligo[i] == 'B'){   reverse += 'V'; }
1790             else if(oligo[i] == 'V'){   reverse += 'B'; }
1791
1792             else if(oligo[i] == 'D'){   reverse += 'H'; }
1793             else if(oligo[i] == 'H'){   reverse += 'D'; }
1794
1795             else                                                {       reverse += 'N'; }
1796         }
1797
1798
1799         return reverse;
1800     }
1801         catch(exception& e) {
1802                 m->errorOut(e, "MakeContigsCommand", "reverseOligo");
1803                 exit(1);
1804         }
1805 }
1806 //**********************************************************************************************************************
1807 vector<int> MakeContigsCommand::convertQual(string qual) {
1808         try {
1809                 vector<int> qualScores;
1810         bool negativeScores = false;
1811
1812                 for (int i = 0; i < qual.length(); i++) {
1813
1814             int temp = 0;
1815             temp = int(qual[i]);
1816             if (format == "illumina") {
1817                 temp -= 64; //char '@'
1818             }else if (format == "illumina1.8+") {
1819                     temp -= int('!'); //char '!'
1820             }else if (format == "solexa") {
1821                 temp = int(convertTable[temp]); //convert to sanger
1822                 temp -= int('!'); //char '!'
1823             }else {
1824                 temp -= int('!'); //char '!'
1825             }
1826
1827             if (temp < -5) { negativeScores = true; }
1828                         qualScores.push_back(temp);
1829                 }
1830
1831         if (negativeScores) { m->mothurOut("[ERROR]: finding negative quality scores, do you have the right format selected? http://en.wikipedia.org/wiki/FASTQ_format#Encoding \n");  m->control_pressed = true;  }
1832
1833                 return qualScores;
1834         }
1835         catch(exception& e) {
1836                 m->errorOut(e, "MakeContigsCommand", "convertQual");
1837                 exit(1);
1838         }
1839 }
1840
1841 //**********************************************************************************************************************
1842
1843
1844
1845