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 string ConsensusSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  58         try {
  59         string outputFileName = "";
  60                 map<string, vector<string> >::iterator it;
  61
  62         //is this a type this command creates
  63         it = outputTypes.find(type);
  64         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  65         else {
  66             if (type == "fasta") {  outputFileName =  "cons.fasta"; }
  67             else if (type == "name") {  outputFileName =  "cons.names"; }
  68             else if (type == "summary") {  outputFileName =  "cons.summary"; }
  69             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  70         }
  71         return outputFileName;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "ConsensusSeqsCommand", "getOutputFileNameTag");
  75                 exit(1);
  76         }
  77 }
  78
  79 //**********************************************************************************************************************
  80 ConsensusSeqsCommand::ConsensusSeqsCommand(){
  81         try {
  82                 abort = true; calledHelp = true;
  83                 setParameters();
  84                 vector<string> tempOutNames;
  85                 outputTypes["fasta"] = tempOutNames;
  86                 outputTypes["name"] = tempOutNames;
  87                 outputTypes["summary"] = tempOutNames;
  88         }
  89         catch(exception& e) {
  90                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
  91                 exit(1);
  92         }
  93 }
  94 //***************************************************************************************************************
  95 ConsensusSeqsCommand::ConsensusSeqsCommand(string option)  {
  96         try {
  97                 abort = false; calledHelp = false;
  98                 allLines = 1;
  99
 100                 //allow user to run help
 101                 if(option == "help") { help(); abort = true; calledHelp = true; }
 102                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 103
 104                 else {
 105
 106                         vector<string> myArray = setParameters();
 107
 108                         OptionParser parser(option);
 109                         map<string,string> parameters = parser.getParameters();
 110
 111                         ValidParameters validParameter;
 112                         map<string,string>::iterator it;
 113
 114                         //check to make sure all parameters are valid for command
 115                         for (it = parameters.begin(); it != parameters.end(); it++) {
 116                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 117                         }
 118
 119                         //initialize outputTypes
 120                         vector<string> tempOutNames;
 121                         outputTypes["fasta"] = tempOutNames;
 122                         outputTypes["name"] = tempOutNames;
 123                         outputTypes["summary"] = tempOutNames;
 124
 125
 126                         //if the user changes the input directory command factory will send this info to us in the output parameter
 127                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 128                         if (inputDir == "not found"){   inputDir = "";          }
 129                         else {
 130                                 string path;
 131                                 it = parameters.find("fasta");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 137                                 }
 138
 139                                 it = parameters.find("name");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 145                                 }
 146
 147                                 it = parameters.find("list");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 153                                 }
 154                         }
 155
 156
 157                         //check for parameters
 158                         fastafile = validParameter.validFile(parameters, "fasta", true);
 159                         if (fastafile == "not open") { abort = true; }
 160                         else if (fastafile == "not found") {
 161                                 fastafile = m->getFastaFile();
 162                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 163                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 164                         }else { m->setFastaFile(fastafile); }
 165
 166                         namefile = validParameter.validFile(parameters, "name", true);
 167                         if (namefile == "not open") { namefile = ""; abort = true; }
 168                         else if (namefile == "not found") { namefile = ""; }
 169                         else { m->setNameFile(namefile); }
 170
 171                         listfile = validParameter.validFile(parameters, "list", true);
 172                         if (listfile == "not open") { abort = true; }
 173                         else if (listfile == "not found") { listfile = "";  }
 174                         else { m->setListFile(listfile); }
 175
 176                         label = validParameter.validFile(parameters, "label", false);
 177                         if (label == "not found") { label = ""; }
 178                         else {
 179                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 180                                 else { allLines = 1;  }
 181                         }
 182
 183                         string temp = validParameter.validFile(parameters, "cutoff", false);  if (temp == "not found") { temp = "100"; }
 184                         m->mothurConvert(temp, cutoff);
 185
 186                         //if the user changes the output directory command factory will send this info to us in the output parameter
 187                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 188
 189                         if (namefile == ""){
 190                                 vector<string> files; files.push_back(fastafile);
 191                                 parser.getNameFile(files);
 192                         }
 193                 }
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "ConsensusSeqsCommand", "ConsensusSeqsCommand");
 197                 exit(1);
 198         }
 199 }
 200 //***************************************************************************************************************
 201
 202 int ConsensusSeqsCommand::execute(){
 203         try{
 204
 205                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 206
 207                 readFasta();
 208
 209                 if (m->control_pressed) { return 0; }
 210
 211                 if (namefile != "") { readNames(); }
 212
 213                 if (m->control_pressed) { return 0; }
 214
 215
 216                 if (listfile == "") {
 217
 218                         ofstream outSummary;
 219                         string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("summary");
 220                         m->openOutputFile(outputSummaryFile, outSummary);
 221                         outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 222                         outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 223
 224                         outSummary << "PositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 225
 226                         ofstream outFasta;
 227                         string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta");
 228                         m->openOutputFile(outputFastaFile, outFasta);
 229                         outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 230
 231                         vector<string> seqs;
 232                         int seqLength = 0;
 233                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 234
 235                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 236
 237                                 string seq = fastaMap[it->second];
 238                                 seqs.push_back(seq);
 239
 240                                 if (seqLength == 0) { seqLength = seq.length(); }
 241                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 242
 243                         }
 244
 245                         vector< vector<float> > percentages; percentages.resize(5);
 246                         for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 247
 248                         string consSeq = "";
 249                         //get counts
 250                         for (int j = 0; j < seqLength; j++) {
 251
 252                                 if (m->control_pressed) { outSummary.close(); outFasta.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } return 0; }
 253
 254                                 vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 255                                 int numDots = 0;
 256
 257                                 for (int i = 0; i < seqs.size(); i++) {
 258
 259                                         if (seqs[i][j] == '.') { numDots++; }
 260
 261                                         char base = toupper(seqs[i][j]);
 262                                         if (base == 'A') { counts[0]++; }
 263                                         else if (base == 'T') { counts[1]++; }
 264                                         else if (base == 'G') { counts[2]++; }
 265                                         else if (base == 'C') { counts[3]++; }
 266                                         else { counts[4]++; }
 267                                 }
 268
 269                                 char conBase = '.';
 270                                 if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 271
 272                                 consSeq += conBase;
 273
 274                                 percentages[0][j] = counts[0] / (float) seqs.size();
 275                                 percentages[1][j] = counts[1] / (float) seqs.size();
 276                                 percentages[2][j] = counts[2] / (float) seqs.size();
 277                                 percentages[3][j] = counts[3] / (float) seqs.size();
 278                                 percentages[4][j] = counts[4] / (float) seqs.size();
 279
 280                         }
 281
 282                         for (int j = 0; j < seqLength; j++) {
 283                                 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;
 284                         }
 285
 286
 287                         outFasta << ">conseq" << endl << consSeq << endl;
 288
 289                         outSummary.close(); outFasta.close();
 290
 291
 292                 }else {
 293
 294
 295                         InputData* input = new InputData(listfile, "list");
 296                         ListVector* list = input->getListVector();
 297
 298                         string lastLabel = list->getLabel();
 299                         set<string> processedLabels;
 300                         set<string> userLabels = labels;
 301
 302                         //as long as you are not at the end of the file or done wih the lines you want
 303                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 304
 305                                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } delete list; delete input;  return 0;  }
 306
 307                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 308
 309                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 310
 311                                         processList(list);
 312
 313                                         processedLabels.insert(list->getLabel());
 314                                         userLabels.erase(list->getLabel());
 315                                 }
 316
 317                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 318                                         string saveLabel = list->getLabel();
 319
 320                                         delete list;
 321
 322                                         list = input->getListVector(lastLabel);
 323                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 324
 325                                         processList(list);
 326
 327                                         processedLabels.insert(list->getLabel());
 328                                         userLabels.erase(list->getLabel());
 329
 330                                         //restore real lastlabel to save below
 331                                         list->setLabel(saveLabel);
 332                                 }
 333
 334                                 lastLabel = list->getLabel();
 335
 336                                 delete list; list = NULL;
 337
 338                                 //get next line to process
 339                                 list = input->getListVector();
 340                         }
 341
 342
 343                         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;  }
 344
 345                         //output error messages about any remaining user labels
 346                         set<string>::iterator it;
 347                         bool needToRun = false;
 348                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 349                                 m->mothurOut("Your file does not include the label " + *it);
 350                                 if (processedLabels.count(lastLabel) != 1) {
 351                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 352                                         needToRun = true;
 353                                 }else {
 354                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 355                                 }
 356                         }
 357
 358                         //run last label if you need to
 359                         if (needToRun == true)  {
 360                                 if (list != NULL) { delete list; }
 361
 362                                 list = input->getListVector(lastLabel);
 363
 364                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 365
 366                                 processList(list);
 367
 368                                 delete list; list = NULL;
 369                         }
 370
 371                         if (list != NULL) { delete list; }
 372                         delete input;
 373                 }
 374
 375                 m->mothurOutEndLine();
 376                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 377                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 378                 m->mothurOutEndLine();
 379
 380
 381                 return 0;
 382
 383         }
 384         catch(exception& e) {
 385                 m->errorOut(e, "ConsensusSeqsCommand", "execute");
 386                 exit(1);
 387         }
 388 }
 389 //***************************************************************************************************************
 390
 391 int ConsensusSeqsCommand::processList(ListVector*& list){
 392         try{
 393
 394                 ofstream outSummary;
 395                 string outputSummaryFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("summary");
 396                 m->openOutputFile(outputSummaryFile, outSummary);
 397                 outSummary.setf(ios::fixed, ios::floatfield); outSummary.setf(ios::showpoint);
 398                 outputNames.push_back(outputSummaryFile); outputTypes["summary"].push_back(outputSummaryFile);
 399
 400                 ofstream outName;
 401                 string outputNameFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("name");
 402                 m->openOutputFile(outputNameFile, outName);
 403                 outputNames.push_back(outputNameFile); outputTypes["name"].push_back(outputNameFile);
 404
 405                 ofstream outFasta;
 406                 string outputFastaFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + list->getLabel() + getOutputFileNameTag("fasta");
 407                 m->openOutputFile(outputFastaFile, outFasta);
 408                 outputNames.push_back(outputFastaFile); outputTypes["fasta"].push_back(outputFastaFile);
 409
 410                 outSummary << "OTU#\tPositioninAlignment\tA\tT\tG\tC\tGap\tNumberofSeqs\tConsensusBase" << endl;
 411
 412                 for (int i = 0; i < list->getNumBins(); i++) {
 413
 414                         if (m->control_pressed) { outSummary.close(); outName.close(); outFasta.close(); return 0; }
 415
 416                         string bin = list->get(i);
 417
 418                         string newName = "";
 419                         string consSeq = getConsSeq(bin, outSummary, newName, i);
 420
 421                         outFasta << ">seq" << (i+1) << endl << consSeq << endl;
 422                         outName << "seq" << (i+1) << '\t' << "seq" << (i+1) << "," << newName << endl;
 423                 }
 424
 425                 outSummary.close(); outName.close(); outFasta.close();
 426
 427                 return 0;
 428
 429         }
 430         catch(exception& e) {
 431                 m->errorOut(e, "ConsensusSeqsCommand", "processList");
 432                 exit(1);
 433         }
 434 }
 435
 436 //***************************************************************************************************************
 437 //made this smart enough to owrk with unique or non unique list file
 438 string ConsensusSeqsCommand::getConsSeq(string bin, ofstream& outSummary, string& name, int binNumber){
 439         try{
 440
 441                 string consSeq = "";
 442                 bool error = false;
 443
 444                 //the whole bin is the second column if no names file, otherwise build it
 445                 name = bin;
 446                 if (namefile != "") { name = ""; }
 447
 448                 vector<string> binNames;
 449                 m->splitAtComma(bin, binNames);
 450
 451                 //get sequence strings for each name in the bin
 452                 vector<string> seqs;
 453
 454                 set<string> addedAlready;
 455                 int seqLength = 0;
 456                 for (int i = 0; i < binNames.size(); i++) {
 457
 458                         map<string, string>::iterator it;
 459
 460                         it = nameMap.find(binNames[i]);
 461                         if (it == nameMap.end()) {
 462                                 if (namefile == "") { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 463                                 else { m->mothurOut("[ERROR]: " + binNames[i] + " is not in your fasta or name file, please correct."); m->mothurOutEndLine(); error = true; }
 464                                 break;
 465                         }else {
 466
 467                                 //add sequence string to seqs vector to process below
 468                                 string seq = fastaMap[it->second];
 469                                 seqs.push_back(seq);
 470
 471                                 if (seqLength == 0) { seqLength = seq.length(); }
 472                                 else if (seqLength != seq.length()) { m->mothurOut("[ERROR]: sequence are not the same length, please correct."); m->mothurOutEndLine(); error = true; break; }
 473
 474                                 if (namefile != "") {
 475                                         //did we add this line from name file already?
 476                                         if (addedAlready.count(it->second) == 0) {
 477                                                 name += "," + nameFileMap[it->second];
 478                                                 addedAlready.insert(it->second);
 479                                         }
 480                                 }
 481
 482                         }
 483                 }
 484
 485                 if (error) { m->control_pressed = true; return consSeq; }
 486
 487                 if (namefile != "") { name = name.substr(1); }
 488
 489                 vector< vector<float> > percentages; percentages.resize(5);
 490                 for (int j = 0; j < percentages.size(); j++) { percentages[j].resize(seqLength, 0.0); }
 491
 492                 //get counts
 493                 for (int j = 0; j < seqLength; j++) {
 494
 495                         if (m->control_pressed) { return consSeq; }
 496
 497                         vector<int> counts; counts.resize(5, 0); //A,T,G,C,Gap
 498                         int numDots = 0;
 499
 500                         for (int i = 0; i < seqs.size(); i++) {
 501
 502                                 if (seqs[i][j] == '.') { numDots++; }
 503
 504                                 char base = toupper(seqs[i][j]);
 505                                 if (base == 'A') { counts[0]++; }
 506                                 else if (base == 'T') { counts[1]++; }
 507                                 else if (base == 'G') { counts[2]++; }
 508                                 else if (base == 'C') { counts[3]++; }
 509                                 else { counts[4]++; }
 510                         }
 511
 512                         char conBase = '.';
 513                         if (numDots != seqs.size()) { conBase = getBase(counts, seqs.size()); }
 514
 515                         consSeq += conBase;
 516
 517                         percentages[0][j] = counts[0] / (float) seqs.size();
 518                         percentages[1][j] = counts[1] / (float) seqs.size();
 519                         percentages[2][j] = counts[2] / (float) seqs.size();
 520                         percentages[3][j] = counts[3] / (float) seqs.size();
 521                         percentages[4][j] = counts[4] / (float) seqs.size();
 522
 523                 }
 524
 525                 for (int j = 0; j < seqLength; j++) {
 526                         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;
 527                 }
 528
 529                 return consSeq;
 530
 531         }
 532         catch(exception& e) {
 533                 m->errorOut(e, "ConsensusSeqsCommand", "getConsSeq");
 534                 exit(1);
 535         }
 536 }
 537 //***************************************************************************************************************
 538
 539 char ConsensusSeqsCommand::getBase(vector<int> counts, int size){  //A,T,G,C,Gap
 540         try{
 541                 /* A = adenine
 542                 * C = cytosine
 543                 * G = guanine
 544                 * T = thymine
 545                 * R = G A (purine)
 546                 * Y = T C (pyrimidine)
 547                 * K = G T (keto)
 548                 * M = A C (amino)
 549                 * S = G C (strong bonds)
 550                 * W = A T (weak bonds)
 551                 * B = G T C (all but A)
 552                 * D = G A T (all but C)
 553                 * H = A C T (all but G)
 554                 * V = G C A (all but T)
 555                 * N = A G C T (any) */
 556
 557                 char conBase = 'N';
 558
 559                 //zero out counts that don't make the cutoff
 560                 float percentage = (100.0 - cutoff) / 100.0;
 561                 int zeroCutoff = percentage * size;
 562
 563                 for (int i = 0; i < counts.size(); i++) {
 564                         if (counts[i] < zeroCutoff) { counts[i] = 0; }
 565                 }
 566
 567                 //any
 568                 if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'n'; }
 569                 //any no gap
 570                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'N'; }
 571                 //all but T
 572                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'v'; }
 573                 //all but T no gap
 574                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'V'; }
 575                 //all but G
 576                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'h'; }
 577                 //all but G no gap
 578                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'H'; }
 579                 //all but C
 580                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'd'; }
 581                 //all but C no gap
 582                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'D'; }
 583                 //all but A
 584                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'b'; }
 585                 //all but A no gap
 586                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'B'; }
 587                 //W = A T (weak bonds)
 588                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'w'; }
 589                 //W = A T (weak bonds) no gap
 590                 else if ((counts[0] != 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'W'; }
 591                 //S = G C (strong bonds)
 592                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 's'; }
 593                 //S = G C (strong bonds) no gap
 594                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'S'; }
 595                 //M = A C (amino)
 596                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'm'; }
 597                 //M = A C (amino) no gap
 598                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'M'; }
 599                 //K = G T (keto)
 600                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'k'; }
 601                 //K = G T (keto) no gap
 602                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'K'; }
 603                 //Y = T C (pyrimidine)
 604                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'y'; }
 605                 //Y = T C (pyrimidine) no gap
 606                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'Y'; }
 607                 //R = G A (purine)
 608                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'r'; }
 609                 //R = G A (purine) no gap
 610                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'R'; }
 611                 //only A
 612                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'a'; }
 613                 //only A no gap
 614                 else if ((counts[0] != 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'A'; }
 615                 //only T
 616                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 't'; }
 617                 //only T no gap
 618                 else if ((counts[0] == 0) && (counts[1] != 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'T'; }
 619                 //only G
 620                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = 'g'; }
 621                 //only G no gap
 622                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] != 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'G'; }
 623                 //only C
 624                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] != 0)) {  conBase = 'c'; }
 625                 //only C no gap
 626                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] != 0) && (counts[4] == 0)) {  conBase = 'C'; }
 627                 //only gap
 628                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] != 0)) {  conBase = '-'; }
 629                 //cutoff removed all counts
 630                 else if ((counts[0] == 0) && (counts[1] == 0) && (counts[2] == 0) && (counts[3] == 0) && (counts[4] == 0)) {  conBase = 'N'; }
 631                 else{ m->mothurOut("[ERROR]: cannot find consensus base."); m->mothurOutEndLine(); }
 632
 633                 return conBase;
 634
 635         }
 636         catch(exception& e) {
 637                 m->errorOut(e, "ConsensusSeqsCommand", "getBase");
 638                 exit(1);
 639         }
 640 }
 641
 642 //***************************************************************************************************************
 643
 644 int ConsensusSeqsCommand::readFasta(){
 645         try{
 646
 647                 ifstream in;
 648                 m->openInputFile(fastafile, in);
 649
 650                 while (!in.eof()) {
 651
 652                         if (m->control_pressed) { break; }
 653
 654                         Sequence seq(in); m->gobble(in);
 655                         string name = seq.getName();
 656
 657                         if (name != "") {
 658                                 fastaMap[name] = seq.getAligned();
 659                                 nameMap[name] = name; //set nameMap incase no names file
 660                                 nameFileMap[name] = name;
 661                         }
 662                 }
 663
 664                 in.close();
 665
 666                 return 0;
 667
 668         }
 669         catch(exception& e) {
 670                 m->errorOut(e, "ConsensusSeqsCommand", "readFasta");
 671                 exit(1);
 672         }
 673 }
 674 //***************************************************************************************************************
 675
 676 int ConsensusSeqsCommand::readNames(){
 677          try{
 678          map<string, string> temp;
 679          map<string, string>::iterator it;
 680          bool error = false;
 681
 682          m->readNames(namefile, temp); //use central buffered read
 683
 684          for (map<string, string>::iterator itTemp = temp.begin(); itTemp != temp.end(); itTemp++) {
 685              string thisname, repnames;
 686              thisname = itTemp->first;
 687              repnames = itTemp->second;
 688
 689              it = nameMap.find(thisname);
 690                          if (it != nameMap.end()) { //then this sequence was in the fastafile
 691                                  nameFileMap[thisname] = repnames;      //for later when outputting the new namesFile if the list file is unique
 692
 693                                  vector<string> splitRepNames;
 694                                  m->splitAtComma(repnames, splitRepNames);
 695
 696                                  for (int i = 0; i < splitRepNames.size(); i++) { nameMap[splitRepNames[i]] = thisname; }
 697
 698                          }else{ m->mothurOut("[ERROR]: " + thisname + " is not in the fasta file, please correct."); m->mothurOutEndLine(); error = true; }
 699          }
 700
 701                  if (error) { m->control_pressed = true; }
 702
 703                  return 0;
 704
 705         }
 706          catch(exception& e) {
 707                  m->errorOut(e, "ConsensusSeqsCommand", "readNames");
 708                  exit(1);
 709          }
 710  }
 711
 712 //***************************************************************************************************************
 713
 714