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") { namefile = ""; 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                         m->mothurConvert(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                         if (namefile == ""){
 168                                 vector<string> files; files.push_back(fastafile);
 169                                 parser.getNameFile(files);
 170                         }
 171                 }
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 175                 exit(1);
 176         }
 177 }
 178 //***************************************************************************************************************
 179
 180 int ConsensusSeqsCommand::execute(){
 181         try{
 182
 183                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 184
 185                 readFasta();
 186
 187                 if (m->control_pressed) { return 0; }
 188
 189                 if (namefile != "") { readNames(); }
 190
 191                 if (m->control_pressed) { return 0; }
 192
 193
 194                 if (listfile == "") {
 195
 196                         ofstream outSummary;
 197                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.summary";
 198                         m->openOutputFile(outputSummaryFile, outSummary);
 199                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 200                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 201
 202                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 203
 204                         ofstream outFasta;
 205                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.fasta";
 206                         m->openOutputFile(outputFastaFile, outFasta);
 207                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 208
 209                         vector<string> seqs;
 210                         int seqLength = 0;
 211                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 212
 213                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 214
 215                                 string seq = fastaMap[it->second];
 216                                 seqs.push_back(seq);
 217
 218                                 if (seqLength == 0) { seqLength = seq.length(); }
 219                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 220
 221                         }
 222
 223                         vector< vector<float> > percentages; percentages.resize(5);
 224                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 225
 226                         string consSeq = "";
 227                         //get counts
 228                         for (int j = 0; j < seqLength; j++) {
 229
 230                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 231
 232                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 233                                 int numDots = 0;
 234
 235                                 for (int i = 0; i < seqs.size(); i++) {
 236
 237                                         if (seqs[i][j] == '.') { numDots++; }
 238
 239                                         char base = toupper(seqs[i][j]);
 240                                         if (base == 'A') { counts[0]++; }
 241                                         else if (base == 'T') { counts[1]++; }
 242                                         else if (base == 'G') { counts[2]++; }
 243                                         else if (base == 'C') { counts[3]++; }
 244                                         else { counts[4]++; }
 245                                 }
 246
 247                                 char conBase = '.';
 248                                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 249
 250                                 consSeq += conBase;
 251
 252                                 percentages[0][j] = counts[0] / (float) seqs.size();
 253                                 percentages[1][j] = counts[1] / (float) seqs.size();
 254                                 percentages[2][j] = counts[2] / (float) seqs.size();
 255                                 percentages[3][j] = counts[3] / (float) seqs.size();
 256                                 percentages[4][j] = counts[4] / (float) seqs.size();
 257
 258                         }
 259
 260                         for (int j = 0; j < seqLength; j++) {
 261                                 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;
 262                         }
 263
 264
 265                         outFasta << ">conseq" << endl << consSeq << endl;
 266
 267                         outSummary.close(); outFasta.close();
 268
 269
 270                 }else {
 271
 272
 273                         InputData* input = new InputData(listfile, "list");
 274                         ListVector* list = input->getListVector();
 275
 276                         string lastLabel = list->getLabel();
 277                         set<string> processedLabels;
 278                         set<string> userLabels = labels;
 279
 280                         //as long as you are not at the end of the file or done wih the lines you want
 281                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 282
 283                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } delete list; delete input;  return 0;  }
 284
 285                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 286
 287                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 288
 289                                         processList(list);
 290
 291                                         processedLabels.insert(list->getLabel());
 292                                         userLabels.erase(list->getLabel());
 293                                 }
 294
 295                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 296                                         string saveLabel = list->getLabel();
 297
 298                                         delete list;
 299
 300                                         list = input->getListVector(lastLabel);
 301                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 302
 303                                         processList(list);
 304
 305                                         processedLabels.insert(list->getLabel());
 306                                         userLabels.erase(list->getLabel());
 307
 308                                         //restore real lastlabel to save below
 309                                         list->setLabel(saveLabel);
 310                                 }
 311
 312                                 lastLabel = list->getLabel();
 313
 314                                 delete list; list = NULL;
 315
 316                                 //get next line to process
 317                                 list = input->getListVector();
 318                         }
 319
 320
 321                         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;  }
 322
 323                         //output error messages about any remaining user labels
 324                         set<string>::iterator it;
 325                         bool needToRun = false;
 326                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 327                                 m->mothurOut("Your file does not include the label " + *it);
 328                                 if (processedLabels.count(lastLabel) != 1) {
 329                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 330                                         needToRun = true;
 331                                 }else {
 332                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 333                                 }
 334                         }
 335
 336                         //run last label if you need to
 337                         if (needToRun == true)  {
 338                                 if (list != NULL) { delete list; }
 339
 340                                 list = input->getListVector(lastLabel);
 341
 342                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 343
 344                                 processList(list);
 345
 346                                 delete list; list = NULL;
 347                         }
 348
 349                         if (list != NULL) { delete list; }
 350                         delete input;
 351                 }
 352
 353                 m->mothurOutEndLine();
 354                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 355                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 356                 m->mothurOutEndLine();
 357
 358
 359                 return 0;
 360
 361         }
 362         catch(exception& e) {
 363                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 364                 exit(1);
 365         }
 366 }
 367 //***************************************************************************************************************
 368
 369 int ConsensusSeqsCommand::processList(ListVector*& list){
 370         try{
 371
 372                 ofstream outSummary;
 373                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.summary";
 374                 m->openOutputFile(outputSummaryFile, outSummary);
 375                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 376                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 377
 378                 ofstream outName;
 379                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.names";
 380                 m->openOutputFile(outputNameFile, outName);
 381                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 382
 383                 ofstream outFasta;
 384                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.fasta";
 385                 m->openOutputFile(outputFastaFile, outFasta);
 386                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 387
 388                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 389
 390                 for (int i = 0; i < list->getNumBins(); i++) {
 391
 392                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 393
 394                         string bin = list->get(i);
 395
 396                         string newName = "";
 397                         string consSeq = getConsSeq(bin, outSummary, newName, i);
 398
 399                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 400                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << newName << endl;
 401                 }
 402
 403                 outSummary.close(); outName.close(); outFasta.close();
 404
 405                 return 0;
 406
 407         }
 408         catch(exception& e) {
 409                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 410                 exit(1);
 411         }
 412 }
 413
 414 //***************************************************************************************************************
 415 //made this smart enough to owrk with unique or non unique list file
 416 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, string& name, int binNumber){
 417         try{
 418
 419                 string consSeq = "";
 420                 bool error = false;
 421
 422                 //the whole bin is the second column if no names file, otherwise build it
 423                 name = bin;
 424                 if (namefile != "") { name = ""; }
 425
 426                 vector<string> binNames;
 427                 m->splitAtComma(bin, binNames);
 428
 429                 //get sequence strings for each name in the bin
 430                 vector<string> seqs;
 431
 432                 set<string> addedAlready;
 433                 int seqLength = 0;
 434                 for (int i = 0; i < binNames.size(); i++) {
 435
 436                         map<string, string>::iterator it;
 437
 438                         it = nameMap.find(binNames[i]);
 439                         if (it == nameMap.end()) {
 440                                 if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 441                                 else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 442                                 break;
 443                         }else {
 444
 445                                 //add sequence string to seqs vector to process below
 446                                 string seq = fastaMap[it->second];
 447                                 seqs.push_back(seq);
 448
 449                                 if (seqLength == 0) { seqLength = seq.length(); }
 450                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); error = true; break; }
 451
 452                                 if (namefile != "") {
 453                                         //did we add this line from name file already?
 454                                         if (addedAlready.count(it->second) == 0) {
 455                                                 name += "," + nameFileMap[it->second];
 456                                                 addedAlready.insert(it->second);
 457                                         }
 458                                 }
 459
 460                         }
 461                 }
 462
 463                 if (error) { m->control_pressed = true; return consSeq; }
 464
 465                 if (namefile != "") { name = name.substr(1); }
 466
 467                 vector< vector<float> > percentages; percentages.resize(5);
 468                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 469
 470                 //get counts
 471                 for (int j = 0; j < seqLength; j++) {
 472
 473                         if (m->control_pressed) { return consSeq; }
 474
 475                         vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 476                         int numDots = 0;
 477
 478                         for (int i = 0; i < seqs.size(); i++) {
 479
 480                                 if (seqs[i][j] == '.') { numDots++; }
 481
 482                                 char base = toupper(seqs[i][j]);
 483                                 if (base == 'A') { counts[0]++; }
 484                                 else if (base == 'T') { counts[1]++; }
 485                                 else if (base == 'G') { counts[2]++; }
 486                                 else if (base == 'C') { counts[3]++; }
 487                                 else { counts[4]++; }
 488                         }
 489
 490                         char conBase = '.';
 491                         if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 492
 493                         consSeq += conBase;
 494
 495                         percentages[0][j] = counts[0] / (float) seqs.size();
 496                         percentages[1][j] = counts[1] / (float) seqs.size();
 497                         percentages[2][j] = counts[2] / (float) seqs.size();
 498                         percentages[3][j] = counts[3] / (float) seqs.size();
 499                         percentages[4][j] = counts[4] / (float) seqs.size();
 500
 501                 }
 502
 503                 for (int j = 0; j < seqLength; j++) {
 504                         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;
 505                 }
 506
 507                 return consSeq;
 508
 509         }
 510         catch(exception& e) {
 511                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 512                 exit(1);
 513         }
 514 }
 515 //***************************************************************************************************************
 516
 517 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 518         try{
 519                 /* A = adenine
 520                 * C = cytosine
 521                 * G = guanine
 522                 * T = thymine
 523                 * R = G A (purine)
 524                 * Y = T C (pyrimidine)
 525                 * K = G T (keto)
 526                 * M = A C (amino)
 527                 * S = G C (strong bonds)
 528                 * W = A T (weak bonds)
 529                 * B = G T C (all but A)
 530                 * D = G A T (all but C)
 531                 * H = A C T (all but G)
 532                 * V = G C A (all but T)
 533                 * N = A G C T (any) */
 534
 535                 char conBase = 'N';
 536
 537                 //zero out counts that don't make the cutoff
 538                 float percentage = (100.0 - cutoff) / 100.0;
 539                 int zeroCutoff = percentage * size;
 540
 541                 for (int i = 0; i < counts.size(); i++) {
 542                         if (counts[i] < zeroCutoff) { counts[i] = 0; }
 543                 }
 544
 545                 //any
 546                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 547                 //any no gap
 548                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 549                 //all but T
 550                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 551                 //all but T no gap
 552                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 553                 //all but G
 554                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 555                 //all but G no gap
 556                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 557                 //all but C
 558                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 559                 //all but C no gap
 560                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 561                 //all but A
 562                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 563                 //all but A no gap
 564                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 565                 //W = A T (weak bonds)
 566                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 567                 //W = A T (weak bonds) no gap
 568                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 569                 //S = G C (strong bonds)
 570                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 571                 //S = G C (strong bonds) no gap
 572                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 573                 //M = A C (amino)
 574                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 575                 //M = A C (amino) no gap
 576                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 577                 //K = G T (keto)
 578                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 579                 //K = G T (keto) no gap
 580                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 581                 //Y = T C (pyrimidine)
 582                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 583                 //Y = T C (pyrimidine) no gap
 584                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 585                 //R = G A (purine)
 586                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 587                 //R = G A (purine) no gap
 588                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 589                 //only A
 590                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 591                 //only A no gap
 592                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 593                 //only T
 594                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 595                 //only T no gap
 596                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 597                 //only G
 598                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 599                 //only G no gap
 600                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 601                 //only C
 602                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 603                 //only C no gap
 604                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 605                 //only gap
 606                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 607                 //cutoff removed all counts
 608                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'N'; }
 609                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 610
 611                 return conBase;
 612
 613         }
 614         catch(exception& e) {
 615                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 616                 exit(1);
 617         }
 618 }
 619
 620 //***************************************************************************************************************
 621
 622 int ConsensusSeqsCommand::readFasta(){
 623         try{
 624
 625                 ifstream in;
 626                 m->openInputFile(fastafile, in);
 627
 628                 while (!in.eof()) {
 629
 630                         if (m->control_pressed) { break; }
 631
 632                         Sequence seq(in); m->gobble(in);
 633                         string name = seq.getName();
 634
 635                         if (name != "") {
 636                                 fastaMap[name] = seq.getAligned();
 637                                 nameMap[name] = name; //set nameMap incase no names file
 638                                 nameFileMap[name] = name;
 639                         }
 640                 }
 641
 642                 in.close();
 643
 644                 return 0;
 645
 646         }
 647         catch(exception& e) {
 648                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 649                 exit(1);
 650         }
 651 }
 652 //***************************************************************************************************************
 653
 654 int ConsensusSeqsCommand::readNames(){
 655          try{
 656
 657                  ifstream in;
 658                  m->openInputFile(namefile, in);
 659
 660                  string thisname, repnames;
 661                  map<string, string>::iterator it;
 662
 663                  bool error = false;
 664
 665                  while(!in.eof()){
 666
 667                          if (m->control_pressed) { break; }
 668
 669                          in >> thisname;                m->gobble(in);          //read from first column
 670                          in >> repnames;                        //read from second column
 671
 672                          it = nameMap.find(thisname);
 673                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 674
 675                                  vector<string> splitRepNames;
 676                                  m->splitAtComma(repnames, splitRepNames);
 677
 678                                  nameFileMap[thisname] = repnames;      //for later when outputting the new namesFile if the list file is unique
 679                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 680
 681                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 682
 683                          m->gobble(in);
 684                  }
 685
 686                  in.close();
 687
 688                  if (error) { m->control_pressed = true; }
 689
 690                  return 0;
 691
 692         }
 693          catch(exception& e) {
 694                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 695                  exit(1);
 696          }
 697  }
 698
 699 //***************************************************************************************************************
 700
 701