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;
  30                 //initialize outputTypes
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["name"] = tempOutNames;
  34                 outputTypes["summary"] = tempOutNames;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> ConsensusSeqsCommand::getRequiredParameters(){
  43         try {
  44                 string Array[] =  {"fasta", "list"};
  45                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "ConsensusSeqsCommand", "getRequiredParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> ConsensusSeqsCommand::getRequiredFiles(){
  55         try {
  56                 vector<string> myArray;
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "DegapSeqsCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 //***************************************************************************************************************
  65 ConsensusSeqsCommand::ConsensusSeqsCommand(string option)  {
  66         try {
  67                 abort = false;
  68                 allLines = 1;
  69
  70                 //allow user to run help
  71                 if(option == "help") { help(); abort = true; }
  72
  73                 else {
  74                         //valid paramters for this command
  75                         string Array[] =  {"fasta","list","name","label", "outputdir","inputdir"};
  76                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  77
  78                         OptionParser parser(option);
  79                         map<string,string> parameters = parser.getParameters();
  80
  81                         ValidParameters validParameter;
  82                         map<string,string>::iterator it;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["fasta"] = tempOutNames;
  92                         outputTypes["name"] = tempOutNames;
  93                         outputTypes["summary"] = tempOutNames;
  94
  95
  96                         //if the user changes the input directory command factory will send this info to us in the output parameter
  97                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  98                         if (inputDir == "not found"){   inputDir = "";          }
  99                         else {
 100                                 string path;
 101                                 it = parameters.find("fasta");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 107                                 }
 108
 109                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 115                                 }
 116
 117                                 it = parameters.find("list");
 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["list"] = inputDir + it->second;             }
 123                                 }
 124                         }
 125
 126
 127                         //check for parameters
 128                         fastafile = validParameter.validFile(parameters, "fasta", true);
 129                         if (fastafile == "not open") { abort = true; }
 130                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the consensus.seqs command."); m->mothurOutEndLine(); abort = true;  }
 131
 132                         namefile = validParameter.validFile(parameters, "name", true);
 133                         if (namefile == "not open") { abort = true; }
 134                         else if (namefile == "not found") { namefile = ""; }
 135
 136                         listfile = validParameter.validFile(parameters, "list", true);
 137                         if (listfile == "not open") { abort = true; }
 138                         else if (listfile == "not found") { listfile = "";  }
 139
 140                         label = validParameter.validFile(parameters, "label", false);
 141                         if (label == "not found") { label = ""; }
 142                         else {
 143                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 144                                 else { allLines = 1;  }
 145                         }
 146
 147                         //if the user changes the output directory command factory will send this info to us in the output parameter
 148                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 149
 150                 }
 151         }
 152         catch(exception& e) {
 153                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 154                 exit(1);
 155         }
 156 }
 157 //**********************************************************************************************************************
 158
 159 void ConsensusSeqsCommand::help(){
 160         try {
 161                 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");
 162                 m->mothurOut("The consensus.seqs command parameters are fasta, list, name and label.\n");
 163                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your sequences, and is required. \n");
 164                 m->mothurOut("The list parameter allows you to enter a your list file. \n");
 165                 m->mothurOut("The name parameter allows you to enter a names file associated with the fasta file. \n");
 166                 m->mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
 167                 m->mothurOut("The consensus.seqs command should be in the following format: \n");
 168                 m->mothurOut("consensus.seqs(fasta=yourFastaFile, list=yourListFile) \n");
 169                 m->mothurOut("Example: consensus.seqs(fasta=abrecovery.align, list=abrecovery.fn.list) \n");
 170                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "ConsensusSeqsCommand", "help");
 174                 exit(1);
 175         }
 176 }
 177
 178 //***************************************************************************************************************
 179
 180 ConsensusSeqsCommand::~ConsensusSeqsCommand(){  /*      do nothing      */      }
 181
 182 //***************************************************************************************************************
 183
 184 int ConsensusSeqsCommand::execute(){
 185         try{
 186
 187                 if (abort == true) { return 0; }
 188
 189                 readFasta();
 190
 191                 if (m->control_pressed) { return 0; }
 192
 193                 if (namefile != "") { readNames(); }
 194
 195                 if (m->control_pressed) { return 0; }
 196
 197
 198                 if (listfile == "") {
 199
 200                         ofstream outSummary;
 201                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.summary";
 202                         m->openOutputFile(outputSummaryFile, outSummary);
 203                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 204                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 205
 206                         ofstream outFasta;
 207                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.fasta";
 208                         m->openOutputFile(outputFastaFile, outFasta);
 209                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 210
 211                         vector<string> seqs;
 212                         int seqLength = 0;
 213                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 214
 215                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 216
 217                                 string seq = fastaMap[it->second];
 218                                 seqs.push_back(seq);
 219
 220                                 if (seqLength == 0) { seqLength = seq.length(); }
 221                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 222
 223                         }
 224
 225                         vector< vector<float> > percentages; percentages.resize(5);
 226                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 227
 228                         string consSeq = "";
 229                         //get counts
 230                         for (int j = 0; j < seqLength; j++) {
 231
 232                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 233
 234                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 235                                 int numDots = 0;
 236
 237                                 for (int i = 0; i < seqs.size(); i++) {
 238
 239                                         if (seqs[i][j] == '.') { numDots++; }
 240
 241                                         char base = toupper(seqs[i][j]);
 242                                         if (base == 'A') { counts[0]++; }
 243                                         else if (base == 'T') { counts[1]++; }
 244                                         else if (base == 'G') { counts[2]++; }
 245                                         else if (base == 'C') { counts[3]++; }
 246                                         else { counts[4]++; }
 247                                 }
 248
 249                                 char conBase = '.';
 250                                 if (numDots != seqs.size()) { conBase = getBase(counts); }
 251
 252                                 consSeq += conBase;
 253
 254                                 percentages[0][j] = counts[0] / (float) seqs.size();
 255                                 percentages[1][j] = counts[1] / (float) seqs.size();
 256                                 percentages[2][j] = counts[2] / (float) seqs.size();
 257                                 percentages[3][j] = counts[3] / (float) seqs.size();
 258                                 percentages[4][j] = counts[4] / (float) seqs.size();
 259
 260                         }
 261
 262                         outSummary << "A" << '\t';
 263                         for (int j = 0; j < seqLength; j++) { outSummary << percentages[0][j] << '\t'; }
 264                         outSummary << endl;
 265                         outSummary << "T" << '\t';
 266                         for (int j = 0; j < seqLength; j++) { outSummary << percentages[1][j] << '\t'; }
 267                         outSummary << endl;
 268                         outSummary << "G" << '\t';
 269                         for (int j = 0; j < seqLength; j++) { outSummary << percentages[2][j] << '\t'; }
 270                         outSummary << endl;
 271                         outSummary << "C" << '\t';
 272                         for (int j = 0; j < seqLength; j++) { outSummary << percentages[3][j] << '\t'; }
 273                         outSummary << endl;
 274                         outSummary << "Gap" << '\t';
 275                         for (int j = 0; j < seqLength; j++) { outSummary << percentages[4][j] << '\t'; }
 276                         outSummary << endl;
 277
 278
 279                         outFasta << ">conseq" << endl << consSeq << endl;
 280
 281                         outSummary.close(); outFasta.close();
 282
 283
 284                 }else {
 285
 286
 287                         InputData* input = new InputData(listfile, "list");
 288                         ListVector* list = input->getListVector();
 289
 290                         string lastLabel = list->getLabel();
 291                         set<string> processedLabels;
 292                         set<string> userLabels = labels;
 293
 294                         //as long as you are not at the end of the file or done wih the lines you want
 295                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 296
 297                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete list; delete input;  return 0;  }
 298
 299                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 300
 301                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 302
 303                                         processList(list);
 304
 305                                         processedLabels.insert(list->getLabel());
 306                                         userLabels.erase(list->getLabel());
 307                                 }
 308
 309                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 310                                         string saveLabel = list->getLabel();
 311
 312                                         delete list;
 313
 314                                         list = input->getListVector(lastLabel);
 315                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 316
 317                                         processList(list);
 318
 319                                         processedLabels.insert(list->getLabel());
 320                                         userLabels.erase(list->getLabel());
 321
 322                                         //restore real lastlabel to save below
 323                                         list->setLabel(saveLabel);
 324                                 }
 325
 326                                 lastLabel = list->getLabel();
 327
 328                                 delete list; list = NULL;
 329
 330                                 //get next line to process
 331                                 list = input->getListVector();
 332                         }
 333
 334
 335                         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;  }
 336
 337                         //output error messages about any remaining user labels
 338                         set<string>::iterator it;
 339                         bool needToRun = false;
 340                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 341                                 m->mothurOut("Your file does not include the label " + *it);
 342                                 if (processedLabels.count(lastLabel) != 1) {
 343                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 344                                         needToRun = true;
 345                                 }else {
 346                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 347                                 }
 348                         }
 349
 350                         //run last label if you need to
 351                         if (needToRun == true)  {
 352                                 if (list != NULL) { delete list; }
 353
 354                                 list = input->getListVector(lastLabel);
 355
 356                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 357
 358                                 processList(list);
 359
 360                                 delete list; list = NULL;
 361                         }
 362
 363                         if (list != NULL) { delete list; }
 364                         delete input;
 365                 }
 366
 367                 m->mothurOutEndLine();
 368                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 369                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 370                 m->mothurOutEndLine();
 371
 372
 373                 return 0;
 374
 375         }
 376         catch(exception& e) {
 377                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 378                 exit(1);
 379         }
 380 }
 381 //***************************************************************************************************************
 382
 383 int ConsensusSeqsCommand::processList(ListVector*& list){
 384         try{
 385
 386                 ofstream outSummary;
 387                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.summary";
 388                 m->openOutputFile(outputSummaryFile, outSummary);
 389                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 390                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 391
 392                 ofstream outName;
 393                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.names";
 394                 m->openOutputFile(outputNameFile, outName);
 395                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 396
 397                 ofstream outFasta;
 398                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.fasta";
 399                 m->openOutputFile(outputFastaFile, outFasta);
 400                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 401
 402                 for (int i = 0; i < list->getNumBins(); i++) {
 403
 404                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 405
 406                         string bin = list->get(i);
 407
 408                         string newName = "";
 409                         string consSeq = getConsSeq(bin, outSummary, newName, i);
 410
 411                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 412                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << newName << endl;
 413                 }
 414
 415                 outSummary.close(); outName.close(); outFasta.close();
 416
 417                 return 0;
 418
 419         }
 420         catch(exception& e) {
 421                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 422                 exit(1);
 423         }
 424 }
 425
 426 //***************************************************************************************************************
 427 //made this smart enough to owrk with unique or non unique list file
 428 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, string& name, int binNumber){
 429         try{
 430
 431                 string consSeq = "";
 432                 bool error = false;
 433
 434                 //the whole bin is the second column if no names file, otherwise build it
 435                 name = bin;
 436                 if (namefile != "") { name = ""; }
 437
 438                 vector<string> binNames;
 439                 m->splitAtComma(bin, binNames);
 440
 441                 //get sequence strings for each name in the bin
 442                 vector<string> seqs;
 443
 444                 set<string> addedAlready;
 445                 int seqLength = 0;
 446                 for (int i = 0; i < binNames.size(); i++) {
 447
 448                         map<string, string>::iterator it;
 449
 450                         it = nameMap.find(binNames[i]);
 451                         if (it == nameMap.end()) {
 452                                 if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 453                                 else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 454                                 break;
 455                         }else {
 456
 457                                 //add sequence string to seqs vector to process below
 458                                 string seq = fastaMap[it->second];
 459                                 seqs.push_back(seq);
 460
 461                                 if (seqLength == 0) { seqLength = seq.length(); }
 462                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); error = true; break; }
 463
 464                                 if (namefile != "") {
 465                                         //did we add this line from name file already?
 466                                         if (addedAlready.count(it->second) == 0) {
 467                                                 name += "," + nameFileMap[it->second];
 468                                                 addedAlready.insert(it->second);
 469                                         }
 470                                 }
 471
 472                         }
 473                 }
 474
 475                 if (error) { m->control_pressed = true; return consSeq; }
 476
 477                 if (namefile != "") { name = name.substr(1); }
 478
 479                 vector< vector<float> > percentages; percentages.resize(5);
 480                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 481
 482                 //get counts
 483                 for (int j = 0; j < seqLength; j++) {
 484
 485                         if (m->control_pressed) { return consSeq; }
 486
 487                         vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 488                         int numDots = 0;
 489
 490                         for (int i = 0; i < seqs.size(); i++) {
 491
 492                                 if (seqs[i][j] == '.') { numDots++; }
 493
 494                                 char base = toupper(seqs[i][j]);
 495                                 if (base == 'A') { counts[0]++; }
 496                                 else if (base == 'T') { counts[1]++; }
 497                                 else if (base == 'G') { counts[2]++; }
 498                                 else if (base == 'C') { counts[3]++; }
 499                                 else { counts[4]++; }
 500                         }
 501
 502                         char conBase = '.';
 503                         if (numDots != seqs.size()) { conBase = getBase(counts); }
 504
 505                         consSeq += conBase;
 506
 507                         percentages[0][j] = counts[0] / (float) seqs.size();
 508                         percentages[1][j] = counts[1] / (float) seqs.size();
 509                         percentages[2][j] = counts[2] / (float) seqs.size();
 510                         percentages[3][j] = counts[3] / (float) seqs.size();
 511                         percentages[4][j] = counts[4] / (float) seqs.size();
 512
 513                 }
 514
 515                 outSummary << ">seq" << (binNumber + 1) << endl;
 516                 outSummary << "A" << '\t';
 517                 for (int j = 0; j < seqLength; j++) { outSummary << percentages[0][j] << '\t'; }
 518                 outSummary << endl;
 519                 outSummary << "T" << '\t';
 520                 for (int j = 0; j < seqLength; j++) { outSummary << percentages[1][j] << '\t'; }
 521                 outSummary << endl;
 522                 outSummary << "G" << '\t';
 523                 for (int j = 0; j < seqLength; j++) { outSummary << percentages[2][j] << '\t'; }
 524                 outSummary << endl;
 525                 outSummary << "C" << '\t';
 526                 for (int j = 0; j < seqLength; j++) { outSummary << percentages[3][j] << '\t'; }
 527                 outSummary << endl;
 528                 outSummary << "Gap" << '\t';
 529                 for (int j = 0; j < seqLength; j++) { outSummary << percentages[4][j] << '\t'; }
 530                 outSummary << endl;
 531
 532
 533                 return consSeq;
 534
 535         }
 536         catch(exception& e) {
 537                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 538                 exit(1);
 539         }
 540 }
 541 //***************************************************************************************************************
 542
 543 char ConsensusSeqsCommand::getBase(vector<int> counts){  //A,T,G,C,Gap
 544         try{
 545                 /* A = adenine
 546                 * C = cytosine
 547                 * G = guanine
 548                 * T = thymine
 549                 * R = G A (purine)
 550                 * Y = T C (pyrimidine)
 551                 * K = G T (keto)
 552                 * M = A C (amino)
 553                 * S = G C (strong bonds)
 554                 * W = A T (weak bonds)
 555                 * B = G T C (all but A)
 556                 * D = G A T (all but C)
 557                 * H = A C T (all but G)
 558                 * V = G C A (all but T)
 559                 * N = A G C T (any) */
 560
 561                 char conBase = 'N';
 562
 563                 //any
 564                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 565                 //any no gap
 566                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 567                 //all but T
 568                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 569                 //all but T no gap
 570                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 571                 //all but G
 572                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 573                 //all but G no gap
 574                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 575                 //all but C
 576                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 577                 //all but C no gap
 578                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 579                 //all but A
 580                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 581                 //all but A no gap
 582                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 583                 //W = A T (weak bonds)
 584                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 585                 //W = A T (weak bonds) no gap
 586                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 587                 //S = G C (strong bonds)
 588                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 589                 //S = G C (strong bonds) no gap
 590                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 591                 //M = A C (amino)
 592                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 593                 //M = A C (amino) no gap
 594                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 595                 //K = G T (keto)
 596                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 597                 //K = G T (keto) no gap
 598                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 599                 //Y = T C (pyrimidine)
 600                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 601                 //Y = T C (pyrimidine) no gap
 602                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 603                 //R = G A (purine)
 604                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 605                 //R = G A (purine) no gap
 606                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 607                 //only A
 608                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 609                 //only A no gap
 610                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 611                 //only T
 612                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 613                 //only T no gap
 614                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 615                 //only G
 616                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 617                 //only G no gap
 618                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 619                 //only C
 620                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 621                 //only C no gap
 622                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 623                 //only gap
 624                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 625                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 626
 627                 return conBase;
 628
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 632                 exit(1);
 633         }
 634 }
 635
 636 //***************************************************************************************************************
 637
 638 int ConsensusSeqsCommand::readFasta(){
 639         try{
 640
 641                 ifstream in;
 642                 m->openInputFile(fastafile, in);
 643
 644                 while (!in.eof()) {
 645
 646                         if (m->control_pressed) { break; }
 647
 648                         Sequence seq(in); m->gobble(in);
 649                         string name = seq.getName();
 650
 651                         if (name != "") {
 652                                 fastaMap[name] = seq.getAligned();
 653                                 nameMap[name] = name; //set nameMap incase no names file
 654                                 nameFileMap[name] = name;
 655                         }
 656                 }
 657
 658                 in.close();
 659
 660                 return 0;
 661
 662         }
 663         catch(exception& e) {
 664                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 665                 exit(1);
 666         }
 667 }
 668 //***************************************************************************************************************
 669
 670 int ConsensusSeqsCommand::readNames(){
 671          try{
 672
 673                  ifstream in;
 674                  m->openInputFile(namefile, in);
 675
 676                  string thisname, repnames;
 677                  map<string, string>::iterator it;
 678
 679                  bool error = false;
 680
 681                  while(!in.eof()){
 682
 683                          if (m->control_pressed) { break; }
 684
 685                          in >> thisname;                m->gobble(in);          //read from first column
 686                          in >> repnames;                        //read from second column
 687
 688                          it = nameMap.find(thisname);
 689                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 690
 691                                  vector<string> splitRepNames;
 692                                  m->splitAtComma(repnames, splitRepNames);
 693
 694                                  nameFileMap[thisname] = repnames;      //for later when outputting the new namesFile if the list file is unique
 695                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 696
 697                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 698
 699                          m->gobble(in);
 700                  }
 701
 702                  in.close();
 703
 704                  if (error) { m->control_pressed = true; }
 705
 706                  return 0;
 707
 708         }
 709          catch(exception& e) {
 710                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 711                  exit(1);
 712          }
 713  }
 714
 715 //***************************************************************************************************************
 716
 717