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", "", "", "namecount", "none", "none",false,false); parameters.push_back(pname);
  19         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none",false,false); parameters.push_back(pcount);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  22                 CommandParameter pcutoff("cutoff", "Number", "", "100", "", "", "",false,false); parameters.push_back(pcutoff);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "ConsensusSeqsCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string ConsensusSeqsCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 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";
  40                 helpString += "The consensus.seqs command parameters are fasta, list, name, count, cutoff and label.\n";
  41                 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";
  42                 helpString += "The list parameter allows you to enter a your list file. \n";
  43                 helpString += "The name parameter allows you to enter a names file associated with the fasta file. \n";
  44                 helpString += "The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n";
  45                 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";
  46                 helpString += "The consensus.seqs command should be in the following format: \n";
  47                 helpString += "consensus.seqs(fasta=yourFastaFile, list=yourListFile) \n";
  48                 helpString += "Example: consensus.seqs(fasta=abrecovery.align, list=abrecovery.fn.list) \n";
  49                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  50                 return helpString;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "ConsensusSeqsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 string ConsensusSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  59         try {
  60         string outputFileName = "";
  61                 map<string, vector<string> >::iterator it;
  62
  63         //is this a type this command creates
  64         it = outputTypes.find(type);
  65         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  66         else {
  67             if (type == "fasta") {  outputFileName =  "cons.fasta"; }
  68             else if (type == "name") {  outputFileName =  "cons.names"; }
  69             else if (type == "count") {  outputFileName =  "cons.count_table"; }
  70             else if (type == "summary") {  outputFileName =  "cons.summary"; }
  71             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  72         }
  73         return outputFileName;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "ConsensusSeqsCommand", "getOutputFileNameTag");
  77                 exit(1);
  78         }
  79 }
  80
  81 //**********************************************************************************************************************
  82 ConsensusSeqsCommand::ConsensusSeqsCommand(){
  83         try {
  84                 abort = true; calledHelp = true;
  85                 setParameters();
  86                 vector<string> tempOutNames;
  87                 outputTypes["fasta"] = tempOutNames;
  88                 outputTypes["name"] = tempOutNames;
  89         outputTypes["count"] = tempOutNames;
  90                 outputTypes["summary"] = tempOutNames;
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
  94                 exit(1);
  95         }
  96 }
  97 //***************************************************************************************************************
  98 ConsensusSeqsCommand::ConsensusSeqsCommand(string option)  {
  99         try {
 100                 abort = false; calledHelp = false;
 101                 allLines = 1;
 102
 103                 //allow user to run help
 104                 if(option == "help") { help(); abort = true; calledHelp = true; }
 105                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 106
 107                 else {
 108
 109                         vector<string> myArray = setParameters();
 110
 111                         OptionParser parser(option);
 112                         map<string,string> parameters = parser.getParameters();
 113
 114                         ValidParameters validParameter;
 115                         map<string,string>::iterator it;
 116
 117                         //check to make sure all parameters are valid for command
 118                         for (it = parameters.begin(); it != parameters.end(); it++) {
 119                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 120                         }
 121
 122                         //initialize outputTypes
 123                         vector<string> tempOutNames;
 124                         outputTypes["fasta"] = tempOutNames;
 125                         outputTypes["name"] = tempOutNames;
 126             outputTypes["count"] = tempOutNames;
 127                         outputTypes["summary"] = tempOutNames;
 128
 129
 130                         //if the user changes the input directory command factory will send this info to us in the output parameter
 131                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 132                         if (inputDir == "not found"){   inputDir = "";          }
 133                         else {
 134                                 string path;
 135                                 it = parameters.find("fasta");
 136                                 //user has given a template file
 137                                 if(it != parameters.end()){
 138                                         path = m->hasPath(it->second);
 139                                         //if the user has not given a path then, add inputdir. else leave path alone.
 140                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 141                                 }
 142
 143                                 it = parameters.find("name");
 144                                 //user has given a template file
 145                                 if(it != parameters.end()){
 146                                         path = m->hasPath(it->second);
 147                                         //if the user has not given a path then, add inputdir. else leave path alone.
 148                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 149                                 }
 150
 151                                 it = parameters.find("list");
 152                                 //user has given a template file
 153                                 if(it != parameters.end()){
 154                                         path = m->hasPath(it->second);
 155                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 157                                 }
 158
 159                 it = parameters.find("count");
 160                                 //user has given a template file
 161                                 if(it != parameters.end()){
 162                                         path = m->hasPath(it->second);
 163                                         //if the user has not given a path then, add inputdir. else leave path alone.
 164                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 165                                 }
 166                         }
 167
 168
 169                         //check for parameters
 170                         fastafile = validParameter.validFile(parameters, "fasta", true);
 171                         if (fastafile == "not open") { abort = true; }
 172                         else if (fastafile == "not found") {
 173                                 fastafile = m->getFastaFile();
 174                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 175                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 176                         }else { m->setFastaFile(fastafile); }
 177
 178                         namefile = validParameter.validFile(parameters, "name", true);
 179                         if (namefile == "not open") { namefile = ""; abort = true; }
 180                         else if (namefile == "not found") { namefile = ""; }
 181                         else { m->setNameFile(namefile); }
 182
 183             countfile = validParameter.validFile(parameters, "count", true);
 184                         if (countfile == "not open") { abort = true; countfile = ""; }
 185                         else if (countfile == "not found") { countfile = ""; }
 186                         else { m->setCountTableFile(countfile); }
 187
 188             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 189
 190                         listfile = validParameter.validFile(parameters, "list", true);
 191                         if (listfile == "not open") { abort = true; }
 192                         else if (listfile == "not found") { listfile = "";  }
 193                         else { m->setListFile(listfile); }
 194
 195                         label = validParameter.validFile(parameters, "label", false);
 196                         if (label == "not found") { label = ""; }
 197                         else {
 198                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 199                                 else { allLines = 1;  }
 200                         }
 201
 202                         string temp = validParameter.validFile(parameters, "cutoff", false);  if (temp == "not found") { temp = "100"; }
 203                         m->mothurConvert(temp, cutoff);
 204
 205                         //if the user changes the output directory command factory will send this info to us in the output parameter
 206                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 207
 208             if (countfile == "") {
 209                 if (namefile == ""){
 210                     vector<string> files; files.push_back(fastafile);
 211                     parser.getNameFile(files);
 212                 }
 213             }
 214                 }
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 218                 exit(1);
 219         }
 220 }
 221 //***************************************************************************************************************
 222
 223 int ConsensusSeqsCommand::execute(){
 224         try{
 225
 226                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 227
 228                 readFasta();
 229
 230                 if (m->control_pressed) { return 0; }
 231
 232                 if (namefile != "") { readNames(); }
 233         if (countfile != "") { ct.readTable(countfile);  }
 234
 235                 if (m->control_pressed) { return 0; }
 236
 237
 238                 if (listfile == "") {
 239
 240                         ofstream outSummary;
 241                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("summary");
 242                         m->openOutputFile(outputSummaryFile, outSummary);
 243                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 244                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 245
 246                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 247
 248                         ofstream outFasta;
 249                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta");
 250                         m->openOutputFile(outputFastaFile, outFasta);
 251                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 252
 253                         vector< vector<float> > percentages; percentages.resize(5);
 254                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 255
 256                         string consSeq = "";
 257             int thisCount;
 258                         //get counts
 259                         for (int j = 0; j < seqLength; j++) {
 260
 261                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 262
 263                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 264                                 int numDots = 0;
 265                                 thisCount = 0;
 266                                 for (map<string, string>::iterator it = fastaMap.begin(); it != fastaMap.end(); it++) {
 267
 268                     string thisSeq = it->second;
 269                     int size = 0;
 270
 271                     if (countfile != "") { size = ct.getNumSeqs(it->first); }
 272                     else {
 273                         map<string, int>::iterator itCount = nameFileMap.find(it->first);
 274                         if (itCount != nameFileMap.end()) {
 275                             size = itCount->second;
 276                         }else { m->mothurOut("[ERROR]: file mismatch, aborting.\n"); m->control_pressed = true; break; }
 277                     }
 278
 279                     for (int k = 0; k < size; k++) {
 280                         if (thisSeq[j] == '.') { numDots++; }
 281
 282                         char base = toupper(thisSeq[j]);
 283                         if (base == 'A') { counts[0]++; }
 284                         else if (base == 'T') { counts[1]++; }
 285                         else if (base == 'G') { counts[2]++; }
 286                         else if (base == 'C') { counts[3]++; }
 287                         else { counts[4]++; }
 288                         thisCount++;
 289                     }
 290                                 }
 291
 292                                 char conBase = '.';
 293                                 if (numDots != thisCount) { conBase = getBase(counts, thisCount); }
 294
 295                                 consSeq += conBase;
 296
 297                                 percentages[0][j] = counts[0] / (float) thisCount;
 298                                 percentages[1][j] = counts[1] / (float) thisCount;
 299                                 percentages[2][j] = counts[2] / (float) thisCount;
 300                                 percentages[3][j] = counts[3] / (float) thisCount;
 301                                 percentages[4][j] = counts[4] / (float) thisCount;
 302                         }
 303
 304                         for (int j = 0; j < seqLength; j++) {
 305                                 outSummary << (j+1) << '\t' << percentages[0][j] << '\t'<< percentages[1][j] << '\t'<< percentages[2][j] << '\t' << percentages[3][j] << '\t' << percentages[4][j] << '\t' << thisCount << '\t' << consSeq[j] << endl;
 306                         }
 307
 308
 309                         outFasta << ">conseq" << endl << consSeq << endl;
 310
 311                         outSummary.close(); outFasta.close();
 312
 313                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 314
 315                 }else {
 316
 317
 318                         InputData* input = new InputData(listfile, "list");
 319                         ListVector* list = input->getListVector();
 320
 321                         string lastLabel = list->getLabel();
 322                         set<string> processedLabels;
 323                         set<string> userLabels = labels;
 324
 325                         //as long as you are not at the end of the file or done wih the lines you want
 326                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 327
 328                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } delete list; delete input;  return 0;  }
 329
 330                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 331
 332                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 333
 334                                         processList(list);
 335
 336                                         processedLabels.insert(list->getLabel());
 337                                         userLabels.erase(list->getLabel());
 338                                 }
 339
 340                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 341                                         string saveLabel = list->getLabel();
 342
 343                                         delete list;
 344
 345                                         list = input->getListVector(lastLabel);
 346                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 347
 348                                         processList(list);
 349
 350                                         processedLabels.insert(list->getLabel());
 351                                         userLabels.erase(list->getLabel());
 352
 353                                         //restore real lastlabel to save below
 354                                         list->setLabel(saveLabel);
 355                                 }
 356
 357                                 lastLabel = list->getLabel();
 358
 359                                 delete list; list = NULL;
 360
 361                                 //get next line to process
 362                                 list = input->getListVector();
 363                         }
 364
 365
 366                         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;  }
 367
 368                         //output error messages about any remaining user labels
 369                         set<string>::iterator it;
 370                         bool needToRun = false;
 371                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 372                                 m->mothurOut("Your file does not include the label " + *it);
 373                                 if (processedLabels.count(lastLabel) != 1) {
 374                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 375                                         needToRun = true;
 376                                 }else {
 377                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 378                                 }
 379                         }
 380
 381                         //run last label if you need to
 382                         if (needToRun == true)  {
 383                                 if (list != NULL) { delete list; }
 384
 385                                 list = input->getListVector(lastLabel);
 386
 387                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 388
 389                                 processList(list);
 390
 391                                 delete list; list = NULL;
 392                         }
 393
 394                         if (list != NULL) { delete list; }
 395                         delete input;
 396                 }
 397
 398                 m->mothurOutEndLine();
 399                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 400                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 401                 m->mothurOutEndLine();
 402
 403
 404                 return 0;
 405
 406         }
 407         catch(exception& e) {
 408                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 409                 exit(1);
 410         }
 411 }
 412 //***************************************************************************************************************
 413
 414 int ConsensusSeqsCommand::processList(ListVector*& list){
 415         try{
 416
 417                 ofstream outSummary;
 418                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("summary");
 419                 m->openOutputFile(outputSummaryFile, outSummary);
 420                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 421                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 422
 423                 ofstream outName;
 424                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("name");
 425                 m->openOutputFile(outputNameFile, outName);
 426                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 427
 428                 ofstream outFasta;
 429                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("fasta");
 430                 m->openOutputFile(outputFastaFile, outFasta);
 431                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 432
 433                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 434
 435                 for (int i = 0; i < list->getNumBins(); i++) {
 436
 437                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 438
 439                         string bin = list->get(i);
 440                         string consSeq = getConsSeq(bin, outSummary, i);
 441
 442                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 443                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << bin << endl;
 444                 }
 445
 446                 outSummary.close(); outName.close(); outFasta.close();
 447
 448                 return 0;
 449
 450         }
 451         catch(exception& e) {
 452                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 453                 exit(1);
 454         }
 455 }
 456
 457 //***************************************************************************************************************
 458 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, int binNumber){
 459         try{
 460
 461                 string consSeq = "";
 462                 bool error = false;
 463         int totalSize=0;
 464
 465                 vector<string> binNames;
 466                 m->splitAtComma(bin, binNames);
 467
 468         vector< vector<float> > percentages; percentages.resize(5);
 469                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 470
 471         if (countfile != "") {
 472             //get counts
 473             for (int j = 0; j < seqLength; j++) {
 474
 475                 if (m->control_pressed) { return consSeq; }
 476
 477                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 478                 int numDots = 0;
 479                 totalSize = 0;
 480                  for (int i = 0; i < binNames.size(); i++) {
 481                      if (m->control_pressed) { return consSeq; }
 482
 483                      string thisSeq = "";
 484                      map<string, string>::iterator itFasta = fastaMap.find(binNames[i]);
 485                      if (itFasta != fastaMap.end()) {
 486                          thisSeq = itFasta->second;
 487                      }else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 488
 489                      int size = ct.getNumSeqs(binNames[i]);
 490                      if (size != 0) {
 491                          for (int k = 0; k < size; k++) {
 492                              if (thisSeq[j] == '.') { numDots++; }
 493
 494                              char base = toupper(thisSeq[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                              totalSize++;
 501                          }
 502                      }else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your count file, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 503                  }
 504                 char conBase = '.';
 505                 if (numDots != totalSize) { conBase = getBase(counts, totalSize); }
 506
 507                 consSeq += conBase;
 508
 509                 percentages[0][j] = counts[0] / (float) totalSize;
 510                 percentages[1][j] = counts[1] / (float) totalSize;
 511                 percentages[2][j] = counts[2] / (float) totalSize;
 512                 percentages[3][j] = counts[3] / (float) totalSize;
 513                 percentages[4][j] = counts[4] / (float) totalSize;
 514             }
 515
 516         }else {
 517
 518             //get sequence strings for each name in the bin
 519             vector<string> seqs;
 520             for (int i = 0; i < binNames.size(); i++) {
 521
 522                 map<string, string>::iterator it;
 523                 it = nameMap.find(binNames[i]);
 524                 if (it == nameMap.end()) {
 525                     if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 526                     else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 527                     break;
 528                 }else {
 529                     //add sequence string to seqs vector to process below
 530                     map<string, string>::iterator itFasta = fastaMap.find(it->second);
 531
 532                     if (itFasta != fastaMap.end()) {
 533                         string seq = itFasta->second;
 534                         seqs.push_back(seq);
 535                     }else { m->mothurOut("[ERROR]: file mismatch, aborting. \n"); }
 536                 }
 537             }
 538
 539             if (error) { m->control_pressed = true; return consSeq; }
 540             totalSize = seqs.size();
 541             //get counts
 542             for (int j = 0; j < seqLength; j++) {
 543
 544                 if (m->control_pressed) { return consSeq; }
 545
 546                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 547                 int numDots = 0;
 548
 549                 for (int i = 0; i < seqs.size(); i++) {
 550
 551                     if (seqs[i][j] == '.') { numDots++; }
 552
 553                     char base = toupper(seqs[i][j]);
 554                     if (base == 'A') { counts[0]++; }
 555                     else if (base == 'T') { counts[1]++; }
 556                     else if (base == 'G') { counts[2]++; }
 557                     else if (base == 'C') { counts[3]++; }
 558                     else { counts[4]++; }
 559                 }
 560
 561                 char conBase = '.';
 562                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 563
 564                 consSeq += conBase;
 565
 566                 percentages[0][j] = counts[0] / (float) seqs.size();
 567                 percentages[1][j] = counts[1] / (float) seqs.size();
 568                 percentages[2][j] = counts[2] / (float) seqs.size();
 569                 percentages[3][j] = counts[3] / (float) seqs.size();
 570                 percentages[4][j] = counts[4] / (float) seqs.size();
 571
 572             }
 573                 }
 574
 575
 576
 577                 for (int j = 0; j < seqLength; j++) {
 578                         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' << totalSize << '\t' << consSeq[j] << endl;
 579                 }
 580
 581                 return consSeq;
 582
 583         }
 584         catch(exception& e) {
 585                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 586                 exit(1);
 587         }
 588 }
 589 //***************************************************************************************************************
 590
 591 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 592         try{
 593                 /* A = adenine
 594                 * C = cytosine
 595                 * G = guanine
 596                 * T = thymine
 597                 * R = G A (purine)
 598                 * Y = T C (pyrimidine)
 599                 * K = G T (keto)
 600                 * M = A C (amino)
 601                 * S = G C (strong bonds)
 602                 * W = A T (weak bonds)
 603                 * B = G T C (all but A)
 604                 * D = G A T (all but C)
 605                 * H = A C T (all but G)
 606                 * V = G C A (all but T)
 607                 * N = A G C T (any) */
 608
 609                 char conBase = 'N';
 610
 611                 //zero out counts that don't make the cutoff
 612                 float percentage = (100.0 - cutoff) / 100.0;
 613                 int zeroCutoff = percentage * size;
 614
 615                 for (int i = 0; i < counts.size(); i++) {
 616                         if (counts[i] < zeroCutoff) { counts[i] = 0; }
 617                 }
 618
 619                 //any
 620                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 621                 //any no gap
 622                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 623                 //all but T
 624                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 625                 //all but T no gap
 626                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 627                 //all but G
 628                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 629                 //all but G no gap
 630                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 631                 //all but C
 632                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 633                 //all but C no gap
 634                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 635                 //all but A
 636                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 637                 //all but A no gap
 638                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 639                 //W = A T (weak bonds)
 640                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 641                 //W = A T (weak bonds) no gap
 642                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 643                 //S = G C (strong bonds)
 644                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 645                 //S = G C (strong bonds) no gap
 646                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 647                 //M = A C (amino)
 648                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 649                 //M = A C (amino) no gap
 650                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 651                 //K = G T (keto)
 652                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 653                 //K = G T (keto) no gap
 654                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 655                 //Y = T C (pyrimidine)
 656                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 657                 //Y = T C (pyrimidine) no gap
 658                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 659                 //R = G A (purine)
 660                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 661                 //R = G A (purine) no gap
 662                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 663                 //only A
 664                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 665                 //only A no gap
 666                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 667                 //only T
 668                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 669                 //only T no gap
 670                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 671                 //only G
 672                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 673                 //only G no gap
 674                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 675                 //only C
 676                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 677                 //only C no gap
 678                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 679                 //only gap
 680                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 681                 //cutoff removed all counts
 682                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'N'; }
 683                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 684
 685                 return conBase;
 686
 687         }
 688         catch(exception& e) {
 689                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 690                 exit(1);
 691         }
 692 }
 693
 694 //***************************************************************************************************************
 695
 696 int ConsensusSeqsCommand::readFasta(){
 697         try{
 698
 699                 ifstream in;
 700                 m->openInputFile(fastafile, in);
 701                 seqLength = 0;
 702
 703                 while (!in.eof()) {
 704
 705                         if (m->control_pressed) { break; }
 706
 707                         Sequence seq(in); m->gobble(in);
 708                         string name = seq.getName();
 709
 710                         if (name != "") {
 711                                 fastaMap[name] = seq.getAligned();
 712                                 nameMap[name] = name; //set nameMap incase no names file
 713                                 nameFileMap[name] = 1;
 714
 715                 if (seqLength == 0) { seqLength = seq.getAligned().length(); }
 716                                 else if (seqLength != seq.getAligned().length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 717                         }
 718                 }
 719
 720                 in.close();
 721
 722                 return 0;
 723
 724         }
 725         catch(exception& e) {
 726                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 727                 exit(1);
 728         }
 729 }
 730 //***************************************************************************************************************
 731
 732 int ConsensusSeqsCommand::readNames(){
 733          try{
 734          map<string, string> temp;
 735          map<string, string>::iterator it;
 736          bool error = false;
 737
 738          m->readNames(namefile, temp); //use central buffered read
 739
 740          for (map<string, string>::iterator itTemp = temp.begin(); itTemp != temp.end(); itTemp++) {
 741              string thisname, repnames;
 742              thisname = itTemp->first;
 743              repnames = itTemp->second;
 744
 745              it = nameMap.find(thisname);
 746                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 747                                  nameFileMap[thisname] = m->getNumNames(repnames);      //for later when outputting the new namesFile if the list file is unique
 748
 749                                  vector<string> splitRepNames;
 750                                  m->splitAtComma(repnames, splitRepNames);
 751
 752                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 753
 754                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 755          }
 756
 757                  if (error) { m->control_pressed = true; }
 758
 759                  return 0;
 760
 761         }
 762          catch(exception& e) {
 763                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 764                  exit(1);
 765          }
 766  }
 767
 768 //***************************************************************************************************************
 769
 770