consensusseqscommand.cpp

   1 /*
   2  *  consensusseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/23/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "consensusseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "inputdata.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> ConsensusSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(plist);
  20                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  21                 CommandParameter pcutoff("cutoff", "Number", "", "100", "", "", "",false,false); parameters.push_back(pcutoff);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ConsensusSeqsCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ConsensusSeqsCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The consensus.seqs command can be used in 2 ways: create a consensus sequence from a fastafile, or with a listfile create a consensus sequence for each otu. Sequences must be aligned.\n";
  39                 helpString += "The consensus.seqs command parameters are fasta, list, name, cutoff and label.\n";
  40                 helpString += "The fasta parameter allows you to enter the fasta file containing your sequences, and is required, unless you have a valid current fasta file. \n";
  41                 helpString += "The list parameter allows you to enter a your list file. \n";
  42                 helpString += "The name parameter allows you to enter a names file associated with the fasta file. \n";
  43                 helpString += "The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n";
  44                 helpString += "The cutoff parameter allows you set a percentage of sequences that support the base. For example: cutoff=97 would only return a sequence that only showed ambiguities for bases that were not supported by at least 97% of sequences.\n";
  45                 helpString += "The consensus.seqs command should be in the following format: \n";
  46                 helpString += "consensus.seqs(fasta=yourFastaFile, list=yourListFile) \n";
  47                 helpString += "Example: consensus.seqs(fasta=abrecovery.align, list=abrecovery.fn.list) \n";
  48                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  49                 return helpString;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "ConsensusSeqsCommand", "getHelpString");
  53                 exit(1);
  54         }
  55 }
  56
  57 //**********************************************************************************************************************
  58 ConsensusSeqsCommand::ConsensusSeqsCommand(){
  59         try {
  60                 abort = true; calledHelp = true;
  61                 setParameters();
  62                 vector<string> tempOutNames;
  63                 outputTypes["fasta"] = tempOutNames;
  64                 outputTypes["name"] = tempOutNames;
  65                 outputTypes["summary"] = tempOutNames;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
  69                 exit(1);
  70         }
  71 }
  72 //***************************************************************************************************************
  73 ConsensusSeqsCommand::ConsensusSeqsCommand(string option)  {
  74         try {
  75                 abort = false; calledHelp = false;
  76                 allLines = 1;
  77
  78                 //allow user to run help
  79                 if(option == "help") { help(); abort = true; calledHelp = true; }
  80                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  81
  82                 else {
  83
  84                         vector<string> myArray = setParameters();
  85
  86                         OptionParser parser(option);
  87                         map<string,string> parameters = parser.getParameters();
  88
  89                         ValidParameters validParameter;
  90                         map<string,string>::iterator it;
  91
  92                         //check to make sure all parameters are valid for command
  93                         for (it = parameters.begin(); it != parameters.end(); it++) {
  94                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  95                         }
  96
  97                         //initialize outputTypes
  98                         vector<string> tempOutNames;
  99                         outputTypes["fasta"] = tempOutNames;
 100                         outputTypes["name"] = tempOutNames;
 101                         outputTypes["summary"] = tempOutNames;
 102
 103
 104                         //if the user changes the input directory command factory will send this info to us in the output parameter
 105                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 106                         if (inputDir == "not found"){   inputDir = "";          }
 107                         else {
 108                                 string path;
 109                                 it = parameters.find("fasta");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 115                                 }
 116
 117                                 it = parameters.find("name");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 123                                 }
 124
 125                                 it = parameters.find("list");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 131                                 }
 132                         }
 133
 134
 135                         //check for parameters
 136                         fastafile = validParameter.validFile(parameters, "fasta", true);
 137                         if (fastafile == "not open") { abort = true; }
 138                         else if (fastafile == "not found") {
 139                                 fastafile = m->getFastaFile();
 140                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 141                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 142                         }else { m->setFastaFile(fastafile); }
 143
 144                         namefile = validParameter.validFile(parameters, "name", true);
 145                         if (namefile == "not open") { abort = true; }
 146                         else if (namefile == "not found") { namefile = ""; }
 147                         else { m->setNameFile(namefile); }
 148
 149                         listfile = validParameter.validFile(parameters, "list", true);
 150                         if (listfile == "not open") { abort = true; }
 151                         else if (listfile == "not found") { listfile = "";  }
 152                         else { m->setListFile(listfile); }
 153
 154                         label = validParameter.validFile(parameters, "label", false);
 155                         if (label == "not found") { label = ""; }
 156                         else {
 157                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 158                                 else { allLines = 1;  }
 159                         }
 160
 161                         string temp = validParameter.validFile(parameters, "cutoff", false);  if (temp == "not found") { temp = "100"; }
 162                         convert(temp, cutoff);
 163
 164                         //if the user changes the output directory command factory will send this info to us in the output parameter
 165                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 166
 167                 }
 168         }
 169         catch(exception& e) {
 170                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 171                 exit(1);
 172         }
 173 }
 174 //***************************************************************************************************************
 175
 176 int ConsensusSeqsCommand::execute(){
 177         try{
 178
 179                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 180
 181                 readFasta();
 182
 183                 if (m->control_pressed) { return 0; }
 184
 185                 if (namefile != "") { readNames(); }
 186
 187                 if (m->control_pressed) { return 0; }
 188
 189
 190                 if (listfile == "") {
 191
 192                         ofstream outSummary;
 193                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.summary";
 194                         m->openOutputFile(outputSummaryFile, outSummary);
 195                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 196                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 197
 198                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 199
 200                         ofstream outFasta;
 201                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.fasta";
 202                         m->openOutputFile(outputFastaFile, outFasta);
 203                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 204
 205                         vector<string> seqs;
 206                         int seqLength = 0;
 207                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 208
 209                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 210
 211                                 string seq = fastaMap[it->second];
 212                                 seqs.push_back(seq);
 213
 214                                 if (seqLength == 0) { seqLength = seq.length(); }
 215                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 216
 217                         }
 218
 219                         vector< vector<float> > percentages; percentages.resize(5);
 220                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 221
 222                         string consSeq = "";
 223                         //get counts
 224                         for (int j = 0; j < seqLength; j++) {
 225
 226                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 227
 228                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 229                                 int numDots = 0;
 230
 231                                 for (int i = 0; i < seqs.size(); i++) {
 232
 233                                         if (seqs[i][j] == '.') { numDots++; }
 234
 235                                         char base = toupper(seqs[i][j]);
 236                                         if (base == 'A') { counts[0]++; }
 237                                         else if (base == 'T') { counts[1]++; }
 238                                         else if (base == 'G') { counts[2]++; }
 239                                         else if (base == 'C') { counts[3]++; }
 240                                         else { counts[4]++; }
 241                                 }
 242
 243                                 char conBase = '.';
 244                                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 245
 246                                 consSeq += conBase;
 247
 248                                 percentages[0][j] = counts[0] / (float) seqs.size();
 249                                 percentages[1][j] = counts[1] / (float) seqs.size();
 250                                 percentages[2][j] = counts[2] / (float) seqs.size();
 251                                 percentages[3][j] = counts[3] / (float) seqs.size();
 252                                 percentages[4][j] = counts[4] / (float) seqs.size();
 253
 254                         }
 255
 256                         for (int j = 0; j < seqLength; j++) {
 257                                 outSummary << (j+1) << '\t' << percentages[0][j] << '\t'<< percentages[1][j] << '\t'<< percentages[2][j] << '\t' << percentages[3][j] << '\t' << percentages[4][j] << '\t' << seqs.size() << '\t' << consSeq[j] << endl;
 258                         }
 259
 260
 261                         outFasta << ">conseq" << endl << consSeq << endl;
 262
 263                         outSummary.close(); outFasta.close();
 264
 265
 266                 }else {
 267
 268
 269                         InputData* input = new InputData(listfile, "list");
 270                         ListVector* list = input->getListVector();
 271
 272                         string lastLabel = list->getLabel();
 273                         set<string> processedLabels;
 274                         set<string> userLabels = labels;
 275
 276                         //as long as you are not at the end of the file or done wih the lines you want
 277                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 278
 279                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } delete list; delete input;  return 0;  }
 280
 281                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 282
 283                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 284
 285                                         processList(list);
 286
 287                                         processedLabels.insert(list->getLabel());
 288                                         userLabels.erase(list->getLabel());
 289                                 }
 290
 291                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 292                                         string saveLabel = list->getLabel();
 293
 294                                         delete list;
 295
 296                                         list = input->getListVector(lastLabel);
 297                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 298
 299                                         processList(list);
 300
 301                                         processedLabels.insert(list->getLabel());
 302                                         userLabels.erase(list->getLabel());
 303
 304                                         //restore real lastlabel to save below
 305                                         list->setLabel(saveLabel);
 306                                 }
 307
 308                                 lastLabel = list->getLabel();
 309
 310                                 delete list; list = NULL;
 311
 312                                 //get next line to process
 313                                 list = input->getListVector();
 314                         }
 315
 316
 317                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } if (list != NULL) { delete list; } delete input; return 0;  }
 318
 319                         //output error messages about any remaining user labels
 320                         set<string>::iterator it;
 321                         bool needToRun = false;
 322                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 323                                 m->mothurOut("Your file does not include the label " + *it);
 324                                 if (processedLabels.count(lastLabel) != 1) {
 325                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 326                                         needToRun = true;
 327                                 }else {
 328                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 329                                 }
 330                         }
 331
 332                         //run last label if you need to
 333                         if (needToRun == true)  {
 334                                 if (list != NULL) { delete list; }
 335
 336                                 list = input->getListVector(lastLabel);
 337
 338                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 339
 340                                 processList(list);
 341
 342                                 delete list; list = NULL;
 343                         }
 344
 345                         if (list != NULL) { delete list; }
 346                         delete input;
 347                 }
 348
 349                 m->mothurOutEndLine();
 350                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 351                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 352                 m->mothurOutEndLine();
 353
 354
 355                 return 0;
 356
 357         }
 358         catch(exception& e) {
 359                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 360                 exit(1);
 361         }
 362 }
 363 //***************************************************************************************************************
 364
 365 int ConsensusSeqsCommand::processList(ListVector*& list){
 366         try{
 367
 368                 ofstream outSummary;
 369                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.summary";
 370                 m->openOutputFile(outputSummaryFile, outSummary);
 371                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 372                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 373
 374                 ofstream outName;
 375                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.names";
 376                 m->openOutputFile(outputNameFile, outName);
 377                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 378
 379                 ofstream outFasta;
 380                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.fasta";
 381                 m->openOutputFile(outputFastaFile, outFasta);
 382                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 383
 384                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 385
 386                 for (int i = 0; i < list->getNumBins(); i++) {
 387
 388                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 389
 390                         string bin = list->get(i);
 391
 392                         string newName = "";
 393                         string consSeq = getConsSeq(bin, outSummary, newName, i);
 394
 395                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 396                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << newName << endl;
 397                 }
 398
 399                 outSummary.close(); outName.close(); outFasta.close();
 400
 401                 return 0;
 402
 403         }
 404         catch(exception& e) {
 405                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 406                 exit(1);
 407         }
 408 }
 409
 410 //***************************************************************************************************************
 411 //made this smart enough to owrk with unique or non unique list file
 412 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, string& name, int binNumber){
 413         try{
 414
 415                 string consSeq = "";
 416                 bool error = false;
 417
 418                 //the whole bin is the second column if no names file, otherwise build it
 419                 name = bin;
 420                 if (namefile != "") { name = ""; }
 421
 422                 vector<string> binNames;
 423                 m->splitAtComma(bin, binNames);
 424
 425                 //get sequence strings for each name in the bin
 426                 vector<string> seqs;
 427
 428                 set<string> addedAlready;
 429                 int seqLength = 0;
 430                 for (int i = 0; i < binNames.size(); i++) {
 431
 432                         map<string, string>::iterator it;
 433
 434                         it = nameMap.find(binNames[i]);
 435                         if (it == nameMap.end()) {
 436                                 if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 437                                 else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 438                                 break;
 439                         }else {
 440
 441                                 //add sequence string to seqs vector to process below
 442                                 string seq = fastaMap[it->second];
 443                                 seqs.push_back(seq);
 444
 445                                 if (seqLength == 0) { seqLength = seq.length(); }
 446                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); error = true; break; }
 447
 448                                 if (namefile != "") {
 449                                         //did we add this line from name file already?
 450                                         if (addedAlready.count(it->second) == 0) {
 451                                                 name += "," + nameFileMap[it->second];
 452                                                 addedAlready.insert(it->second);
 453                                         }
 454                                 }
 455
 456                         }
 457                 }
 458
 459                 if (error) { m->control_pressed = true; return consSeq; }
 460
 461                 if (namefile != "") { name = name.substr(1); }
 462
 463                 vector< vector<float> > percentages; percentages.resize(5);
 464                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 465
 466                 //get counts
 467                 for (int j = 0; j < seqLength; j++) {
 468
 469                         if (m->control_pressed) { return consSeq; }
 470
 471                         vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 472                         int numDots = 0;
 473
 474                         for (int i = 0; i < seqs.size(); i++) {
 475
 476                                 if (seqs[i][j] == '.') { numDots++; }
 477
 478                                 char base = toupper(seqs[i][j]);
 479                                 if (base == 'A') { counts[0]++; }
 480                                 else if (base == 'T') { counts[1]++; }
 481                                 else if (base == 'G') { counts[2]++; }
 482                                 else if (base == 'C') { counts[3]++; }
 483                                 else { counts[4]++; }
 484                         }
 485
 486                         char conBase = '.';
 487                         if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 488
 489                         consSeq += conBase;
 490
 491                         percentages[0][j] = counts[0] / (float) seqs.size();
 492                         percentages[1][j] = counts[1] / (float) seqs.size();
 493                         percentages[2][j] = counts[2] / (float) seqs.size();
 494                         percentages[3][j] = counts[3] / (float) seqs.size();
 495                         percentages[4][j] = counts[4] / (float) seqs.size();
 496
 497                 }
 498
 499                 for (int j = 0; j < seqLength; j++) {
 500                         outSummary << (binNumber + 1) << '\t' << (j+1) << '\t' << percentages[0][j] << '\t'<< percentages[1][j] << '\t'<< percentages[2][j] << '\t' << percentages[3][j] << '\t' << percentages[4][j] << '\t' << seqs.size() << '\t' << consSeq[j] << endl;
 501                 }
 502
 503                 return consSeq;
 504
 505         }
 506         catch(exception& e) {
 507                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 508                 exit(1);
 509         }
 510 }
 511 //***************************************************************************************************************
 512
 513 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 514         try{
 515                 /* A = adenine
 516                 * C = cytosine
 517                 * G = guanine
 518                 * T = thymine
 519                 * R = G A (purine)
 520                 * Y = T C (pyrimidine)
 521                 * K = G T (keto)
 522                 * M = A C (amino)
 523                 * S = G C (strong bonds)
 524                 * W = A T (weak bonds)
 525                 * B = G T C (all but A)
 526                 * D = G A T (all but C)
 527                 * H = A C T (all but G)
 528                 * V = G C A (all but T)
 529                 * N = A G C T (any) */
 530
 531                 char conBase = 'N';
 532
 533                 //zero out counts that don't make the cutoff
 534                 float percentage = (100.0 - cutoff) / 100.0;
 535                 int zeroCutoff = percentage * size;
 536
 537                 for (int i = 0; i < counts.size(); i++) {
 538                         if (counts[i] < zeroCutoff) { counts[i] = 0; }
 539                 }
 540
 541                 //any
 542                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 543                 //any no gap
 544                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 545                 //all but T
 546                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 547                 //all but T no gap
 548                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 549                 //all but G
 550                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 551                 //all but G no gap
 552                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 553                 //all but C
 554                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 555                 //all but C no gap
 556                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 557                 //all but A
 558                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 559                 //all but A no gap
 560                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 561                 //W = A T (weak bonds)
 562                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 563                 //W = A T (weak bonds) no gap
 564                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 565                 //S = G C (strong bonds)
 566                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 567                 //S = G C (strong bonds) no gap
 568                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 569                 //M = A C (amino)
 570                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 571                 //M = A C (amino) no gap
 572                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 573                 //K = G T (keto)
 574                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 575                 //K = G T (keto) no gap
 576                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 577                 //Y = T C (pyrimidine)
 578                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 579                 //Y = T C (pyrimidine) no gap
 580                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 581                 //R = G A (purine)
 582                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 583                 //R = G A (purine) no gap
 584                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 585                 //only A
 586                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 587                 //only A no gap
 588                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 589                 //only T
 590                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 591                 //only T no gap
 592                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 593                 //only G
 594                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 595                 //only G no gap
 596                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 597                 //only C
 598                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 599                 //only C no gap
 600                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 601                 //only gap
 602                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 603                 //cutoff removed all counts
 604                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'N'; }
 605                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 606
 607                 return conBase;
 608
 609         }
 610         catch(exception& e) {
 611                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 612                 exit(1);
 613         }
 614 }
 615
 616 //***************************************************************************************************************
 617
 618 int ConsensusSeqsCommand::readFasta(){
 619         try{
 620
 621                 ifstream in;
 622                 m->openInputFile(fastafile, in);
 623
 624                 while (!in.eof()) {
 625
 626                         if (m->control_pressed) { break; }
 627
 628                         Sequence seq(in); m->gobble(in);
 629                         string name = seq.getName();
 630
 631                         if (name != "") {
 632                                 fastaMap[name] = seq.getAligned();
 633                                 nameMap[name] = name; //set nameMap incase no names file
 634                                 nameFileMap[name] = name;
 635                         }
 636                 }
 637
 638                 in.close();
 639
 640                 return 0;
 641
 642         }
 643         catch(exception& e) {
 644                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 645                 exit(1);
 646         }
 647 }
 648 //***************************************************************************************************************
 649
 650 int ConsensusSeqsCommand::readNames(){
 651          try{
 652
 653                  ifstream in;
 654                  m->openInputFile(namefile, in);
 655
 656                  string thisname, repnames;
 657                  map<string, string>::iterator it;
 658
 659                  bool error = false;
 660
 661                  while(!in.eof()){
 662
 663                          if (m->control_pressed) { break; }
 664
 665                          in >> thisname;                m->gobble(in);          //read from first column
 666                          in >> repnames;                        //read from second column
 667
 668                          it = nameMap.find(thisname);
 669                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 670
 671                                  vector<string> splitRepNames;
 672                                  m->splitAtComma(repnames, splitRepNames);
 673
 674                                  nameFileMap[thisname] = repnames;      //for later when outputting the new namesFile if the list file is unique
 675                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 676
 677                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 678
 679                          m->gobble(in);
 680                  }
 681
 682                  in.close();
 683
 684                  if (error) { m->control_pressed = true; }
 685
 686                  return 0;
 687
 688         }
 689          catch(exception& e) {
 690                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 691                  exit(1);
 692          }
 693  }
 694
 695 //***************************************************************************************************************
 696
 697