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