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                         }
 143
 144                         namefile = validParameter.validFile(parameters, "name", true);
 145                         if (namefile == "not open") { abort = true; }
 146                         else if (namefile == "not found") { namefile = ""; }
 147
 148                         listfile = validParameter.validFile(parameters, "list", true);
 149                         if (listfile == "not open") { abort = true; }
 150                         else if (listfile == "not found") { listfile = "";  }
 151
 152                         label = validParameter.validFile(parameters, "label", false);
 153                         if (label == "not found") { label = ""; }
 154                         else {
 155                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 156                                 else { allLines = 1;  }
 157                         }
 158
 159                         string temp = validParameter.validFile(parameters, "cutoff", false);  if (temp == "not found") { temp = "100"; }
 160                         convert(temp, cutoff);
 161
 162                         //if the user changes the output directory command factory will send this info to us in the output parameter
 163                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 164
 165                 }
 166         }
 167         catch(exception& e) {
 168                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 169                 exit(1);
 170         }
 171 }
 172 //***************************************************************************************************************
 173
 174 int ConsensusSeqsCommand::execute(){
 175         try{
 176
 177                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 178
 179                 readFasta();
 180
 181                 if (m->control_pressed) { return 0; }
 182
 183                 if (namefile != "") { readNames(); }
 184
 185                 if (m->control_pressed) { return 0; }
 186
 187
 188                 if (listfile == "") {
 189
 190                         ofstream outSummary;
 191                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.summary";
 192                         m->openOutputFile(outputSummaryFile, outSummary);
 193                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 194                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 195
 196                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 197
 198                         ofstream outFasta;
 199                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "cons.fasta";
 200                         m->openOutputFile(outputFastaFile, outFasta);
 201                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 202
 203                         vector<string> seqs;
 204                         int seqLength = 0;
 205                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 206
 207                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 208
 209                                 string seq = fastaMap[it->second];
 210                                 seqs.push_back(seq);
 211
 212                                 if (seqLength == 0) { seqLength = seq.length(); }
 213                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 214
 215                         }
 216
 217                         vector< vector<float> > percentages; percentages.resize(5);
 218                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 219
 220                         string consSeq = "";
 221                         //get counts
 222                         for (int j = 0; j < seqLength; j++) {
 223
 224                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 225
 226                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 227                                 int numDots = 0;
 228
 229                                 for (int i = 0; i < seqs.size(); i++) {
 230
 231                                         if (seqs[i][j] == '.') { numDots++; }
 232
 233                                         char base = toupper(seqs[i][j]);
 234                                         if (base == 'A') { counts[0]++; }
 235                                         else if (base == 'T') { counts[1]++; }
 236                                         else if (base == 'G') { counts[2]++; }
 237                                         else if (base == 'C') { counts[3]++; }
 238                                         else { counts[4]++; }
 239                                 }
 240
 241                                 char conBase = '.';
 242                                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 243
 244                                 consSeq += conBase;
 245
 246                                 percentages[0][j] = counts[0] / (float) seqs.size();
 247                                 percentages[1][j] = counts[1] / (float) seqs.size();
 248                                 percentages[2][j] = counts[2] / (float) seqs.size();
 249                                 percentages[3][j] = counts[3] / (float) seqs.size();
 250                                 percentages[4][j] = counts[4] / (float) seqs.size();
 251
 252                         }
 253
 254                         for (int j = 0; j < seqLength; j++) {
 255                                 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;
 256                         }
 257
 258
 259                         outFasta << ">conseq" << endl << consSeq << endl;
 260
 261                         outSummary.close(); outFasta.close();
 262
 263
 264                 }else {
 265
 266
 267                         InputData* input = new InputData(listfile, "list");
 268                         ListVector* list = input->getListVector();
 269
 270                         string lastLabel = list->getLabel();
 271                         set<string> processedLabels;
 272                         set<string> userLabels = labels;
 273
 274                         //as long as you are not at the end of the file or done wih the lines you want
 275                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 276
 277                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } delete list; delete input;  return 0;  }
 278
 279                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 280
 281                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 282
 283                                         processList(list);
 284
 285                                         processedLabels.insert(list->getLabel());
 286                                         userLabels.erase(list->getLabel());
 287                                 }
 288
 289                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 290                                         string saveLabel = list->getLabel();
 291
 292                                         delete list;
 293
 294                                         list = input->getListVector(lastLabel);
 295                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 296
 297                                         processList(list);
 298
 299                                         processedLabels.insert(list->getLabel());
 300                                         userLabels.erase(list->getLabel());
 301
 302                                         //restore real lastlabel to save below
 303                                         list->setLabel(saveLabel);
 304                                 }
 305
 306                                 lastLabel = list->getLabel();
 307
 308                                 delete list; list = NULL;
 309
 310                                 //get next line to process
 311                                 list = input->getListVector();
 312                         }
 313
 314
 315                         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;  }
 316
 317                         //output error messages about any remaining user labels
 318                         set<string>::iterator it;
 319                         bool needToRun = false;
 320                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 321                                 m->mothurOut("Your file does not include the label " + *it);
 322                                 if (processedLabels.count(lastLabel) != 1) {
 323                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 324                                         needToRun = true;
 325                                 }else {
 326                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 327                                 }
 328                         }
 329
 330                         //run last label if you need to
 331                         if (needToRun == true)  {
 332                                 if (list != NULL) { delete list; }
 333
 334                                 list = input->getListVector(lastLabel);
 335
 336                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 337
 338                                 processList(list);
 339
 340                                 delete list; list = NULL;
 341                         }
 342
 343                         if (list != NULL) { delete list; }
 344                         delete input;
 345                 }
 346
 347                 m->mothurOutEndLine();
 348                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 349                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 350                 m->mothurOutEndLine();
 351
 352
 353                 return 0;
 354
 355         }
 356         catch(exception& e) {
 357                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 358                 exit(1);
 359         }
 360 }
 361 //***************************************************************************************************************
 362
 363 int ConsensusSeqsCommand::processList(ListVector*& list){
 364         try{
 365
 366                 ofstream outSummary;
 367                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.summary";
 368                 m->openOutputFile(outputSummaryFile, outSummary);
 369                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 370                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 371
 372                 ofstream outName;
 373                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.names";
 374                 m->openOutputFile(outputNameFile, outName);
 375                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 376
 377                 ofstream outFasta;
 378                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + ".cons.fasta";
 379                 m->openOutputFile(outputFastaFile, outFasta);
 380                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 381
 382                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 383
 384                 for (int i = 0; i < list->getNumBins(); i++) {
 385
 386                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 387
 388                         string bin = list->get(i);
 389
 390                         string newName = "";
 391                         string consSeq = getConsSeq(bin, outSummary, newName, i);
 392
 393                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 394                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << newName << endl;
 395                 }
 396
 397                 outSummary.close(); outName.close(); outFasta.close();
 398
 399                 return 0;
 400
 401         }
 402         catch(exception& e) {
 403                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 404                 exit(1);
 405         }
 406 }
 407
 408 //***************************************************************************************************************
 409 //made this smart enough to owrk with unique or non unique list file
 410 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, string& name, int binNumber){
 411         try{
 412
 413                 string consSeq = "";
 414                 bool error = false;
 415
 416                 //the whole bin is the second column if no names file, otherwise build it
 417                 name = bin;
 418                 if (namefile != "") { name = ""; }
 419
 420                 vector<string> binNames;
 421                 m->splitAtComma(bin, binNames);
 422
 423                 //get sequence strings for each name in the bin
 424                 vector<string> seqs;
 425
 426                 set<string> addedAlready;
 427                 int seqLength = 0;
 428                 for (int i = 0; i < binNames.size(); i++) {
 429
 430                         map<string, string>::iterator it;
 431
 432                         it = nameMap.find(binNames[i]);
 433                         if (it == nameMap.end()) {
 434                                 if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 435                                 else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 436                                 break;
 437                         }else {
 438
 439                                 //add sequence string to seqs vector to process below
 440                                 string seq = fastaMap[it->second];
 441                                 seqs.push_back(seq);
 442
 443                                 if (seqLength == 0) { seqLength = seq.length(); }
 444                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); error = true; break; }
 445
 446                                 if (namefile != "") {
 447                                         //did we add this line from name file already?
 448                                         if (addedAlready.count(it->second) == 0) {
 449                                                 name += "," + nameFileMap[it->second];
 450                                                 addedAlready.insert(it->second);
 451                                         }
 452                                 }
 453
 454                         }
 455                 }
 456
 457                 if (error) { m->control_pressed = true; return consSeq; }
 458
 459                 if (namefile != "") { name = name.substr(1); }
 460
 461                 vector< vector<float> > percentages; percentages.resize(5);
 462                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 463
 464                 //get counts
 465                 for (int j = 0; j < seqLength; j++) {
 466
 467                         if (m->control_pressed) { return consSeq; }
 468
 469                         vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 470                         int numDots = 0;
 471
 472                         for (int i = 0; i < seqs.size(); i++) {
 473
 474                                 if (seqs[i][j] == '.') { numDots++; }
 475
 476                                 char base = toupper(seqs[i][j]);
 477                                 if (base == 'A') { counts[0]++; }
 478                                 else if (base == 'T') { counts[1]++; }
 479                                 else if (base == 'G') { counts[2]++; }
 480                                 else if (base == 'C') { counts[3]++; }
 481                                 else { counts[4]++; }
 482                         }
 483
 484                         char conBase = '.';
 485                         if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 486
 487                         consSeq += conBase;
 488
 489                         percentages[0][j] = counts[0] / (float) seqs.size();
 490                         percentages[1][j] = counts[1] / (float) seqs.size();
 491                         percentages[2][j] = counts[2] / (float) seqs.size();
 492                         percentages[3][j] = counts[3] / (float) seqs.size();
 493                         percentages[4][j] = counts[4] / (float) seqs.size();
 494
 495                 }
 496
 497                 for (int j = 0; j < seqLength; j++) {
 498                         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;
 499                 }
 500
 501                 return consSeq;
 502
 503         }
 504         catch(exception& e) {
 505                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 506                 exit(1);
 507         }
 508 }
 509 //***************************************************************************************************************
 510
 511 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 512         try{
 513                 /* A = adenine
 514                 * C = cytosine
 515                 * G = guanine
 516                 * T = thymine
 517                 * R = G A (purine)
 518                 * Y = T C (pyrimidine)
 519                 * K = G T (keto)
 520                 * M = A C (amino)
 521                 * S = G C (strong bonds)
 522                 * W = A T (weak bonds)
 523                 * B = G T C (all but A)
 524                 * D = G A T (all but C)
 525                 * H = A C T (all but G)
 526                 * V = G C A (all but T)
 527                 * N = A G C T (any) */
 528
 529                 char conBase = 'N';
 530
 531                 //zero out counts that don't make the cutoff
 532                 float percentage = (100.0 - cutoff) / 100.0;
 533                 int zeroCutoff = percentage * size;
 534
 535                 for (int i = 0; i < counts.size(); i++) {
 536                         if (counts[i] < zeroCutoff) { counts[i] = 0; }
 537                 }
 538
 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                 //cutoff removed all counts
 603                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  m->mothurOut("cutoff ...."); m->mothurOutEndLine(); }
 604                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 605
 606                 return conBase;
 607
 608         }
 609         catch(exception& e) {
 610                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 611                 exit(1);
 612         }
 613 }
 614
 615 //***************************************************************************************************************
 616
 617 int ConsensusSeqsCommand::readFasta(){
 618         try{
 619
 620                 ifstream in;
 621                 m->openInputFile(fastafile, in);
 622
 623                 while (!in.eof()) {
 624
 625                         if (m->control_pressed) { break; }
 626
 627                         Sequence seq(in); m->gobble(in);
 628                         string name = seq.getName();
 629
 630                         if (name != "") {
 631                                 fastaMap[name] = seq.getAligned();
 632                                 nameMap[name] = name; //set nameMap incase no names file
 633                                 nameFileMap[name] = name;
 634                         }
 635                 }
 636
 637                 in.close();
 638
 639                 return 0;
 640
 641         }
 642         catch(exception& e) {
 643                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 644                 exit(1);
 645         }
 646 }
 647 //***************************************************************************************************************
 648
 649 int ConsensusSeqsCommand::readNames(){
 650          try{
 651
 652                  ifstream in;
 653                  m->openInputFile(namefile, in);
 654
 655                  string thisname, repnames;
 656                  map<string, string>::iterator it;
 657
 658                  bool error = false;
 659
 660                  while(!in.eof()){
 661
 662                          if (m->control_pressed) { break; }
 663
 664                          in >> thisname;                m->gobble(in);          //read from first column
 665                          in >> repnames;                        //read from second column
 666
 667                          it = nameMap.find(thisname);
 668                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 669
 670                                  vector<string> splitRepNames;
 671                                  m->splitAtComma(repnames, splitRepNames);
 672
 673                                  nameFileMap[thisname] = repnames;      //for later when outputting the new namesFile if the list file is unique
 674                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 675
 676                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 677
 678                          m->gobble(in);
 679                  }
 680
 681                  in.close();
 682
 683                  if (error) { m->control_pressed = true; }
 684
 685                  return 0;
 686
 687         }
 688          catch(exception& e) {
 689                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 690                  exit(1);
 691          }
 692  }
 693
 694 //***************************************************************************************************************
 695
 696