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","fasta-name",false,true,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "namecount", "none", "none","name",false,false,true); parameters.push_back(pname);
  19         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none","count",false,false,true); parameters.push_back(pcount);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none","fasta-name",false,false,true); 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::getOutputPattern(string type) {
  59     try {
  60         string pattern = "";
  61
  62         if (type == "fasta") {  pattern = "[filename],cons.fasta-[filename],[tag],cons.fasta"; }
  63         else if (type == "name") {  pattern = "[filename],cons.names-[filename],[tag],cons.names"; }
  64         else if (type == "count") {  pattern = "[filename],cons.count_table-[filename],[tag],cons.count_table"; }
  65         else if (type == "summary") {  pattern = "[filename],cons.summary-[filename],[tag],cons.summary"; }
  66         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  67
  68         return pattern;
  69     }
  70     catch(exception& e) {
  71         m->errorOut(e, "ConsensusSeqsCommand", "getOutputPattern");
  72         exit(1);
  73     }
  74 }
  75 //**********************************************************************************************************************
  76 ConsensusSeqsCommand::ConsensusSeqsCommand(){
  77         try {
  78                 abort = true; calledHelp = true;
  79                 setParameters();
  80                 vector<string> tempOutNames;
  81                 outputTypes["fasta"] = tempOutNames;
  82                 outputTypes["name"] = tempOutNames;
  83         outputTypes["count"] = tempOutNames;
  84                 outputTypes["summary"] = tempOutNames;
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
  88                 exit(1);
  89         }
  90 }
  91 //***************************************************************************************************************
  92 ConsensusSeqsCommand::ConsensusSeqsCommand(string option)  {
  93         try {
  94                 abort = false; calledHelp = false;
  95                 allLines = 1;
  96
  97                 //allow user to run help
  98                 if(option == "help") { help(); abort = true; calledHelp = true; }
  99                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 100
 101                 else {
 102
 103                         vector<string> myArray = setParameters();
 104
 105                         OptionParser parser(option);
 106                         map<string,string> parameters = parser.getParameters();
 107
 108                         ValidParameters validParameter;
 109                         map<string,string>::iterator it;
 110
 111                         //check to make sure all parameters are valid for command
 112                         for (it = parameters.begin(); it != parameters.end(); it++) {
 113                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 114                         }
 115
 116                         //initialize outputTypes
 117                         vector<string> tempOutNames;
 118                         outputTypes["fasta"] = tempOutNames;
 119                         outputTypes["name"] = tempOutNames;
 120             outputTypes["count"] = tempOutNames;
 121                         outputTypes["summary"] = tempOutNames;
 122
 123
 124                         //if the user changes the input directory command factory will send this info to us in the output parameter
 125                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 126                         if (inputDir == "not found"){   inputDir = "";          }
 127                         else {
 128                                 string path;
 129                                 it = parameters.find("fasta");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 135                                 }
 136
 137                                 it = parameters.find("name");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 143                                 }
 144
 145                                 it = parameters.find("list");
 146                                 //user has given a template file
 147                                 if(it != parameters.end()){
 148                                         path = m->hasPath(it->second);
 149                                         //if the user has not given a path then, add inputdir. else leave path alone.
 150                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 151                                 }
 152
 153                 it = parameters.find("count");
 154                                 //user has given a template file
 155                                 if(it != parameters.end()){
 156                                         path = m->hasPath(it->second);
 157                                         //if the user has not given a path then, add inputdir. else leave path alone.
 158                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 159                                 }
 160                         }
 161
 162
 163                         //check for parameters
 164                         fastafile = validParameter.validFile(parameters, "fasta", true);
 165                         if (fastafile == "not open") { abort = true; }
 166                         else if (fastafile == "not found") {
 167                                 fastafile = m->getFastaFile();
 168                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 169                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 170                         }else { m->setFastaFile(fastafile); }
 171
 172                         namefile = validParameter.validFile(parameters, "name", true);
 173                         if (namefile == "not open") { namefile = ""; abort = true; }
 174                         else if (namefile == "not found") { namefile = ""; }
 175                         else { m->setNameFile(namefile); }
 176
 177             countfile = validParameter.validFile(parameters, "count", true);
 178                         if (countfile == "not open") { abort = true; countfile = ""; }
 179                         else if (countfile == "not found") { countfile = ""; }
 180                         else { m->setCountTableFile(countfile); }
 181
 182             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 183
 184                         listfile = validParameter.validFile(parameters, "list", true);
 185                         if (listfile == "not open") { abort = true; }
 186                         else if (listfile == "not found") { listfile = "";  }
 187                         else { m->setListFile(listfile); }
 188
 189                         label = validParameter.validFile(parameters, "label", false);
 190                         if (label == "not found") { label = ""; }
 191                         else {
 192                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 193                                 else { allLines = 1;  }
 194                         }
 195
 196                         string temp = validParameter.validFile(parameters, "cutoff", false);  if (temp == "not found") { temp = "100"; }
 197                         m->mothurConvert(temp, cutoff);
 198
 199                         //if the user changes the output directory command factory will send this info to us in the output parameter
 200                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 201
 202             if (countfile == "") {
 203                 if (namefile == ""){
 204                     vector<string> files; files.push_back(fastafile);
 205                     parser.getNameFile(files);
 206                 }
 207             }
 208                 }
 209         }
 210         catch(exception& e) {
 211                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 212                 exit(1);
 213         }
 214 }
 215 //***************************************************************************************************************
 216
 217 int ConsensusSeqsCommand::execute(){
 218         try{
 219
 220                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 221
 222         int start = time(NULL);
 223
 224                 readFasta();
 225
 226                 if (m->control_pressed) { return 0; }
 227
 228                 if (namefile != "") { readNames(); }
 229         if (countfile != "") { ct.readTable(countfile, true, false);  }
 230
 231                 if (m->control_pressed) { return 0; }
 232
 233
 234                 if (listfile == "") {
 235
 236                         ofstream outSummary;
 237             map<string, string> variables;
 238             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 239                         string outputSummaryFile = getOutputFileName("summary", variables);
 240                         m->openOutputFile(outputSummaryFile, outSummary);
 241                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 242                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 243
 244                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 245
 246                         ofstream outFasta;
 247                         string outputFastaFile = getOutputFileName("fasta", variables);
 248                         m->openOutputFile(outputFastaFile, outFasta);
 249                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 250
 251                         vector< vector<float> > percentages; percentages.resize(5);
 252                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 253
 254                         string consSeq = "";
 255             int thisCount;
 256                         //get counts
 257                         for (int j = 0; j < seqLength; j++) {
 258
 259                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 260
 261                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 262                                 int numDots = 0;
 263                                 thisCount = 0;
 264                                 for (map<string, string>::iterator it = fastaMap.begin(); it != fastaMap.end(); it++) {
 265
 266                     string thisSeq = it->second;
 267                     int size = 0;
 268
 269                     if (countfile != "") { size = ct.getNumSeqs(it->first); }
 270                     else {
 271                         map<string, int>::iterator itCount = nameFileMap.find(it->first);
 272                         if (itCount != nameFileMap.end()) {
 273                             size = itCount->second;
 274                         }else { m->mothurOut("[ERROR]: file mismatch, aborting.\n"); m->control_pressed = true; break; }
 275                     }
 276
 277                     for (int k = 0; k < size; k++) {
 278                         if (thisSeq[j] == '.') { numDots++; }
 279
 280                         char base = toupper(thisSeq[j]);
 281                         if (base == 'A') { counts[0]++; }
 282                         else if (base == 'T') { counts[1]++; }
 283                         else if (base == 'G') { counts[2]++; }
 284                         else if (base == 'C') { counts[3]++; }
 285                         else { counts[4]++; }
 286                         thisCount++;
 287                     }
 288                                 }
 289
 290                                 char conBase = '.';
 291                                 if (numDots != thisCount) { conBase = getBase(counts, thisCount); }
 292
 293                                 consSeq += conBase;
 294
 295                                 percentages[0][j] = counts[0] / (float) thisCount;
 296                                 percentages[1][j] = counts[1] / (float) thisCount;
 297                                 percentages[2][j] = counts[2] / (float) thisCount;
 298                                 percentages[3][j] = counts[3] / (float) thisCount;
 299                                 percentages[4][j] = counts[4] / (float) thisCount;
 300                         }
 301
 302                         for (int j = 0; j < seqLength; j++) {
 303                                 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;
 304                         }
 305
 306
 307                         outFasta << ">conseq" << endl << consSeq << endl;
 308
 309                         outSummary.close(); outFasta.close();
 310
 311                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 312
 313                 }else {
 314
 315
 316                         InputData* input = new InputData(listfile, "list");
 317                         ListVector* list = input->getListVector();
 318
 319                         string lastLabel = list->getLabel();
 320                         set<string> processedLabels;
 321                         set<string> userLabels = labels;
 322
 323                         //as long as you are not at the end of the file or done wih the lines you want
 324                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 325
 326                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } delete list; delete input;  return 0;  }
 327
 328                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 329
 330                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 331
 332                                         processList(list);
 333
 334                                         processedLabels.insert(list->getLabel());
 335                                         userLabels.erase(list->getLabel());
 336                                 }
 337
 338                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 339                                         string saveLabel = list->getLabel();
 340
 341                                         delete list;
 342
 343                                         list = input->getListVector(lastLabel);
 344                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 345
 346                                         processList(list);
 347
 348                                         processedLabels.insert(list->getLabel());
 349                                         userLabels.erase(list->getLabel());
 350
 351                                         //restore real lastlabel to save below
 352                                         list->setLabel(saveLabel);
 353                                 }
 354
 355                                 lastLabel = list->getLabel();
 356
 357                                 delete list; list = NULL;
 358
 359                                 //get next line to process
 360                                 list = input->getListVector();
 361                         }
 362
 363
 364                         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;  }
 365
 366                         //output error messages about any remaining user labels
 367                         set<string>::iterator it;
 368                         bool needToRun = false;
 369                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 370                                 m->mothurOut("Your file does not include the label " + *it);
 371                                 if (processedLabels.count(lastLabel) != 1) {
 372                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 373                                         needToRun = true;
 374                                 }else {
 375                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 376                                 }
 377                         }
 378
 379                         //run last label if you need to
 380                         if (needToRun == true)  {
 381                                 if (list != NULL) { delete list; }
 382
 383                                 list = input->getListVector(lastLabel);
 384
 385                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 386
 387                                 processList(list);
 388
 389                                 delete list; list = NULL;
 390                         }
 391
 392                         if (list != NULL) { delete list; }
 393                         delete input;
 394                 }
 395
 396         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to find the consensus sequences.");
 397
 398                 m->mothurOutEndLine();
 399                 m->mothurOut("Output File Names: "); 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         map<string, string> variables;
 419         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 420         variables["[tag]"] = list->getLabel();
 421                 string outputSummaryFile = getOutputFileName("summary", variables);
 422                 m->openOutputFile(outputSummaryFile, outSummary);
 423                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 424                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 425
 426                 ofstream outName;
 427                 string outputNameFile = getOutputFileName("name",variables);
 428                 m->openOutputFile(outputNameFile, outName);
 429                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 430
 431                 ofstream outFasta;
 432                 string outputFastaFile = getOutputFileName("fasta",variables);
 433                 m->openOutputFile(outputFastaFile, outFasta);
 434                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 435
 436                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 437
 438         string snumBins = toString(list->getNumBins());
 439         vector<string> binLabels = list->getLabels();
 440                 for (int i = 0; i < list->getNumBins(); i++) {
 441
 442                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 443
 444                         string bin = list->get(i);
 445                         string consSeq = getConsSeq(bin, outSummary, i);
 446
 447                         outFasta << ">" << binLabels[i] << endl << consSeq << endl;
 448                         outName << binLabels[i] << '\t' << binLabels[i] << "," << bin << endl;
 449                 }
 450
 451                 outSummary.close(); outName.close(); outFasta.close();
 452
 453                 return 0;
 454
 455         }
 456         catch(exception& e) {
 457                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 458                 exit(1);
 459         }
 460 }
 461
 462 //***************************************************************************************************************
 463 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, int binNumber){
 464         try{
 465
 466                 string consSeq = "";
 467                 bool error = false;
 468         int totalSize=0;
 469
 470                 vector<string> binNames;
 471                 m->splitAtComma(bin, binNames);
 472
 473         vector< vector<float> > percentages; percentages.resize(5);
 474                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 475
 476         if (countfile != "") {
 477             //get counts
 478             for (int j = 0; j < seqLength; j++) {
 479
 480                 if (m->control_pressed) { return consSeq; }
 481
 482                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 483                 int numDots = 0;
 484                 totalSize = 0;
 485                  for (int i = 0; i < binNames.size(); i++) {
 486                      if (m->control_pressed) { return consSeq; }
 487
 488                      string thisSeq = "";
 489                      map<string, string>::iterator itFasta = fastaMap.find(binNames[i]);
 490                      if (itFasta != fastaMap.end()) {
 491                          thisSeq = itFasta->second;
 492                      }else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 493
 494                      int size = ct.getNumSeqs(binNames[i]);
 495                      if (size != 0) {
 496                          for (int k = 0; k < size; k++) {
 497                              if (thisSeq[j] == '.') { numDots++; }
 498
 499                              char base = toupper(thisSeq[j]);
 500                              if (base == 'A') { counts[0]++; }
 501                              else if (base == 'T') { counts[1]++; }
 502                              else if (base == 'G') { counts[2]++; }
 503                              else if (base == 'C') { counts[3]++; }
 504                              else { counts[4]++; }
 505                              totalSize++;
 506                          }
 507                      }else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your count file, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 508                  }
 509                 char conBase = '.';
 510                 if (numDots != totalSize) { conBase = getBase(counts, totalSize); }
 511
 512                 consSeq += conBase;
 513
 514                 percentages[0][j] = counts[0] / (float) totalSize;
 515                 percentages[1][j] = counts[1] / (float) totalSize;
 516                 percentages[2][j] = counts[2] / (float) totalSize;
 517                 percentages[3][j] = counts[3] / (float) totalSize;
 518                 percentages[4][j] = counts[4] / (float) totalSize;
 519             }
 520
 521         }else {
 522
 523             //get sequence strings for each name in the bin
 524             vector<string> seqs;
 525             for (int i = 0; i < binNames.size(); i++) {
 526
 527                 map<string, string>::iterator it;
 528                 it = nameMap.find(binNames[i]);
 529                 if (it == nameMap.end()) {
 530                     if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 531                     else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 532                     break;
 533                 }else {
 534                     //add sequence string to seqs vector to process below
 535                     map<string, string>::iterator itFasta = fastaMap.find(it->second);
 536
 537                     if (itFasta != fastaMap.end()) {
 538                         string seq = itFasta->second;
 539                         seqs.push_back(seq);
 540                     }else { m->mothurOut("[ERROR]: file mismatch, aborting. \n"); }
 541                 }
 542             }
 543
 544             if (error) { m->control_pressed = true; return consSeq; }
 545             totalSize = seqs.size();
 546             //get counts
 547             for (int j = 0; j < seqLength; j++) {
 548
 549                 if (m->control_pressed) { return consSeq; }
 550
 551                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 552                 int numDots = 0;
 553
 554                 for (int i = 0; i < seqs.size(); i++) {
 555
 556                     if (seqs[i][j] == '.') { numDots++; }
 557
 558                     char base = toupper(seqs[i][j]);
 559                     if (base == 'A') { counts[0]++; }
 560                     else if (base == 'T') { counts[1]++; }
 561                     else if (base == 'G') { counts[2]++; }
 562                     else if (base == 'C') { counts[3]++; }
 563                     else { counts[4]++; }
 564                 }
 565
 566                 char conBase = '.';
 567                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 568
 569                 consSeq += conBase;
 570
 571                 percentages[0][j] = counts[0] / (float) seqs.size();
 572                 percentages[1][j] = counts[1] / (float) seqs.size();
 573                 percentages[2][j] = counts[2] / (float) seqs.size();
 574                 percentages[3][j] = counts[3] / (float) seqs.size();
 575                 percentages[4][j] = counts[4] / (float) seqs.size();
 576
 577             }
 578                 }
 579
 580
 581
 582                 for (int j = 0; j < seqLength; j++) {
 583                         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;
 584                 }
 585
 586                 return consSeq;
 587
 588         }
 589         catch(exception& e) {
 590                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 591                 exit(1);
 592         }
 593 }
 594 //***************************************************************************************************************
 595
 596 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 597         try{
 598                 /* A = adenine
 599                 * C = cytosine
 600                 * G = guanine
 601                 * T = thymine
 602                 * R = G A (purine)
 603                 * Y = T C (pyrimidine)
 604                 * K = G T (keto)
 605                 * M = A C (amino)
 606                 * S = G C (strong bonds)
 607                 * W = A T (weak bonds)
 608                 * B = G T C (all but A)
 609                 * D = G A T (all but C)
 610                 * H = A C T (all but G)
 611                 * V = G C A (all but T)
 612                 * N = A G C T (any) */
 613
 614                 char conBase = 'N';
 615
 616                 //zero out counts that don't make the cutoff
 617                 float percentage = (100.0 - cutoff) / 100.0;
 618
 619                 for (int i = 0; i < counts.size(); i++) {
 620             float countPercentage = counts[i] / (float) size;
 621                         if (countPercentage < percentage) { counts[i] = 0; }
 622                 }
 623
 624                 //any
 625                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 626                 //any no gap
 627                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 628                 //all but T
 629                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 630                 //all but T no gap
 631                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 632                 //all but G
 633                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 634                 //all but G no gap
 635                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 636                 //all but C
 637                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 638                 //all but C no gap
 639                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 640                 //all but A
 641                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 642                 //all but A no gap
 643                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 644                 //W = A T (weak bonds)
 645                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 646                 //W = A T (weak bonds) no gap
 647                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 648                 //S = G C (strong bonds)
 649                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 650                 //S = G C (strong bonds) no gap
 651                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 652                 //M = A C (amino)
 653                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 654                 //M = A C (amino) no gap
 655                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 656                 //K = G T (keto)
 657                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 658                 //K = G T (keto) no gap
 659                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 660                 //Y = T C (pyrimidine)
 661                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 662                 //Y = T C (pyrimidine) no gap
 663                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 664                 //R = G A (purine)
 665                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 666                 //R = G A (purine) no gap
 667                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 668                 //only A
 669                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 670                 //only A no gap
 671                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 672                 //only T
 673                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 674                 //only T no gap
 675                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 676                 //only G
 677                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 678                 //only G no gap
 679                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 680                 //only C
 681                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 682                 //only C no gap
 683                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 684                 //only gap
 685                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 686                 //cutoff removed all counts
 687                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'N'; }
 688                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 689
 690                 return conBase;
 691
 692         }
 693         catch(exception& e) {
 694                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 695                 exit(1);
 696         }
 697 }
 698
 699 //***************************************************************************************************************
 700
 701 int ConsensusSeqsCommand::readFasta(){
 702         try{
 703
 704                 ifstream in;
 705                 m->openInputFile(fastafile, in);
 706                 seqLength = 0;
 707
 708                 while (!in.eof()) {
 709
 710                         if (m->control_pressed) { break; }
 711
 712                         Sequence seq(in); m->gobble(in);
 713                         string name = seq.getName();
 714
 715                         if (name != "") {
 716                                 fastaMap[name] = seq.getAligned();
 717                                 nameMap[name] = name; //set nameMap incase no names file
 718                                 nameFileMap[name] = 1;
 719
 720                 if (seqLength == 0) { seqLength = seq.getAligned().length(); }
 721                                 else if (seqLength != seq.getAligned().length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 722                         }
 723                 }
 724
 725                 in.close();
 726
 727                 return 0;
 728
 729         }
 730         catch(exception& e) {
 731                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 732                 exit(1);
 733         }
 734 }
 735 //***************************************************************************************************************
 736
 737 int ConsensusSeqsCommand::readNames(){
 738          try{
 739          map<string, string> temp;
 740          map<string, string>::iterator it;
 741          bool error = false;
 742
 743          m->readNames(namefile, temp); //use central buffered read
 744
 745          for (map<string, string>::iterator itTemp = temp.begin(); itTemp != temp.end(); itTemp++) {
 746              string thisname, repnames;
 747              thisname = itTemp->first;
 748              repnames = itTemp->second;
 749
 750              it = nameMap.find(thisname);
 751                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 752                                  nameFileMap[thisname] = m->getNumNames(repnames);      //for later when outputting the new namesFile if the list file is unique
 753
 754                                  vector<string> splitRepNames;
 755                                  m->splitAtComma(repnames, splitRepNames);
 756
 757                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 758
 759                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 760          }
 761
 762                  if (error) { m->control_pressed = true; }
 763
 764                  return 0;
 765
 766         }
 767          catch(exception& e) {
 768                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 769                  exit(1);
 770          }
 771  }
 772
 773 //***************************************************************************************************************
 774
 775