seqerrorcommand.cpp

   1 /*
   2  *  seqerrorcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 7/15/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "seqerrorcommand.h"
  11 #include "reportfile.h"
  12 #include "qualityscores.h"
  13 #include "refchimeratest.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> SeqErrorCommand::getValidParameters(){
  17         try {
  18                 string Array[] =  {"query", "reference", "name", "qfile", "report", "threshold", "ignorechimeras", "inputdir", "outputdir"};
  19                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "SeqErrorCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 SeqErrorCommand::SeqErrorCommand(){
  29         try {
  30                 abort = true; calledHelp = true;
  31                 vector<string> tempOutNames;
  32                 outputTypes["error.summary"] = tempOutNames;
  33                 outputTypes["error.seq"] = tempOutNames;
  34                 outputTypes["error.quality"] = tempOutNames;
  35                 outputTypes["error.qual.forward"] = tempOutNames;
  36                 outputTypes["error.qual.reverse"] = tempOutNames;
  37                 outputTypes["error.forward"] = tempOutNames;
  38                 outputTypes["error.reverse"] = tempOutNames;
  39                 outputTypes["error.count"] = tempOutNames;
  40                 outputTypes["error.matrix"] = tempOutNames;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 vector<string> SeqErrorCommand::getRequiredParameters(){
  49         try {
  50                 string Array[] =  {"query","reference"};
  51                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  52                 return myArray;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "SeqErrorCommand", "getRequiredParameters");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 vector<string> SeqErrorCommand::getRequiredFiles(){
  61         try {
  62                 vector<string> myArray;
  63                 return myArray;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "SeqErrorCommand", "getRequiredFiles");
  67                 exit(1);
  68         }
  69 }
  70 //***************************************************************************************************************
  71
  72 SeqErrorCommand::SeqErrorCommand(string option)  {
  73         try {
  74
  75                 abort = false; calledHelp = false;
  76
  77                 //allow user to run help
  78                 if(option == "help") { help(); abort = true; calledHelp = true; }
  79
  80                 else {
  81                         string temp;
  82
  83                         //valid paramters for this command
  84                         string AlignArray[] =  {"query", "reference", "name", "qfile", "report", "threshold", "inputdir", "ignorechimeras", "outputdir"};
  85
  86                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  87
  88                         OptionParser parser(option);
  89                         map<string,string> parameters = parser.getParameters();
  90
  91                         ValidParameters validParameter;
  92                         map<string,string>::iterator it;
  93
  94                         //check to make sure all parameters are valid for command
  95                         for (it = parameters.begin(); it != parameters.end(); it++) {
  96                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  97                         }
  98
  99                         //initialize outputTypes
 100                         vector<string> tempOutNames;
 101                         outputTypes["error.summary"] = tempOutNames;
 102                         outputTypes["error.seq"] = tempOutNames;
 103                         outputTypes["error.quality"] = tempOutNames;
 104                         outputTypes["error.qual.forward"] = tempOutNames;
 105                         outputTypes["error.qual.reverse"] = tempOutNames;
 106                         outputTypes["error.forward"] = tempOutNames;
 107                         outputTypes["error.reverse"] = tempOutNames;
 108                         outputTypes["error.count"] = tempOutNames;
 109                         outputTypes["error.matrix"] = tempOutNames;
 110
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("query");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["query"] = inputDir + it->second;            }
 123                                 }
 124
 125                                 it = parameters.find("reference");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 131                                 }
 132
 133                                 it = parameters.find("name");
 134                                 //user has given a names file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 139                                 }
 140
 141                                 it = parameters.find("qfile");
 142                                 //user has given a quality score file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 147                                 }
 148
 149                                 it = parameters.find("report");
 150                                 //user has given a alignment report file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["report"] = inputDir + it->second;           }
 155                                 }
 156
 157                         }
 158                         //check for required parameters
 159                         queryFileName = validParameter.validFile(parameters, "query", true);
 160                         if (queryFileName == "not found") { m->mothurOut("query is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
 161                         else if (queryFileName == "not open") { abort = true; }
 162
 163                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 164                         if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
 165                         else if (referenceFileName == "not open") { abort = true; }
 166
 167
 168                         //check for optional parameters
 169                         namesFileName = validParameter.validFile(parameters, "name", true);
 170                         if(namesFileName == "not found"){       namesFileName = "";     }
 171
 172                         qualFileName = validParameter.validFile(parameters, "qfile", true);
 173                         if(qualFileName == "not found"){        qualFileName = "";      }
 174
 175                         reportFileName = validParameter.validFile(parameters, "report", true);
 176                         if(reportFileName == "not found"){      reportFileName = "";    }
 177
 178                         if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
 179                                 m->mothurOut("if you use either a qual file or a report file, you have to have both.");
 180                                 m->mothurOutEndLine();
 181                                 abort = true;
 182                         }
 183
 184                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 185                         if (outputDir == "not found"){
 186                                 outputDir = "";
 187                                 outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
 188                         }
 189
 190                         //check for optional parameter and set defaults
 191                         // ...at some point should added some additional type checking...
 192                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
 193                         convert(temp, threshold);
 194
 195                         temp = validParameter.validFile(parameters, "ignorechimeras", false);   if (temp == "not found") { temp = "1"; }
 196                         convert(temp, ignoreChimeras);
 197
 198                         substitutionMatrix.resize(6);
 199                         for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
 200                 }
 201         }
 202         catch(exception& e) {
 203                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 204                 exit(1);
 205         }
 206 }
 207
 208 //**********************************************************************************************************************
 209
 210 void SeqErrorCommand::help(){
 211         try {
 212                 m->mothurOut("The seq.error command reads a query alignment file and a reference alignment file and creates .....\n");
 213                 m->mothurOut("Example seq.error(...).\n");
 214                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
 215                 m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n\n");
 216         }
 217         catch(exception& e) {
 218                 m->errorOut(e, "SeqErrorCommand", "help");
 219                 exit(1);
 220         }
 221 }
 222
 223 //***************************************************************************************************************
 224
 225 SeqErrorCommand::~SeqErrorCommand(){    /*      void    */      }
 226
 227 //***************************************************************************************************************
 228
 229 int SeqErrorCommand::execute(){
 230         try{
 231                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 232
 233                 maxLength = 2000;
 234
 235                 string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
 236                 m->openOutputFile(errorSummaryFileName, errorSummaryFile);
 237                 outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
 238                 printErrorHeader();
 239
 240                 string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
 241                 m->openOutputFile(errorSeqFileName, errorSeqFile);
 242                 outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
 243
 244                 getReferences();        //read in reference sequences - make sure there's no ambiguous bases
 245
 246                 map<string, int> weights;
 247                 if(namesFileName != ""){        weights = getWeights(); }
 248
 249                 ifstream queryFile;
 250                 m->openInputFile(queryFileName, queryFile);
 251
 252                 ifstream reportFile;
 253                 ifstream qualFile;
 254
 255                 ReportFile report;
 256                 QualityScores quality;
 257                 vector<vector<int> > qualForwardMap;
 258                 vector<vector<int> > qualReverseMap;
 259
 260                 if(qualFileName != "" && reportFileName != ""){
 261                         m->openInputFile(qualFileName, qualFile);
 262                         report = ReportFile(reportFile, reportFileName);
 263
 264                         qualForwardMap.resize(maxLength);
 265                         qualReverseMap.resize(maxLength);
 266                         for(int i=0;i<maxLength;i++){
 267                                 qualForwardMap[i].assign(41,0);
 268                                 qualReverseMap[i].assign(41,0);
 269                         }
 270                 }
 271
 272                 int totalBases = 0;
 273                 int totalMatches = 0;
 274
 275                 vector<int> misMatchCounts(11, 0);
 276                 int maxMismatch = 0;
 277                 int numSeqs = 0;
 278
 279                 map<string, int>::iterator it;
 280                 map<char, vector<int> > qScoreErrorMap;
 281                 qScoreErrorMap['m'].assign(41, 0);
 282                 qScoreErrorMap['s'].assign(41, 0);
 283                 qScoreErrorMap['i'].assign(41, 0);
 284                 qScoreErrorMap['a'].assign(41, 0);
 285
 286                 map<char, vector<int> > errorForward;
 287                 errorForward['m'].assign(maxLength,0);
 288                 errorForward['s'].assign(maxLength,0);
 289                 errorForward['i'].assign(maxLength,0);
 290                 errorForward['d'].assign(maxLength,0);
 291                 errorForward['a'].assign(maxLength,0);
 292
 293                 map<char, vector<int> > errorReverse;
 294                 errorReverse['m'].assign(maxLength,0);
 295                 errorReverse['s'].assign(maxLength,0);
 296                 errorReverse['i'].assign(maxLength,0);
 297                 errorReverse['d'].assign(maxLength,0);
 298                 errorReverse['a'].assign(maxLength,0);
 299
 300
 301                 string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
 302                 RefChimeraTest chimeraTest(referenceSeqs, errorChimeraFileName);
 303                 outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
 304
 305                 vector<string> megaAlignVector(numRefs, "");
 306
 307                 int index = 0;
 308                 bool ignoreSeq = 0;
 309
 310                 while(queryFile){
 311
 312                         if (m->control_pressed) { errorSummaryFile.close();     errorSeqFile.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 313
 314                         Sequence query(queryFile);
 315
 316                         int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned());
 317                         int closestRefIndex = chimeraTest.getClosestRefIndex();
 318
 319                         if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
 320                         else                                                                                    {       ignoreSeq = 0;  }
 321
 322                         Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
 323
 324                         if(namesFileName != ""){
 325                                 it = weights.find(query.getName());
 326                                 minCompare.weight = it->second;
 327                         }
 328                         else{   minCompare.weight = 1;  }
 329
 330                         printErrorData(minCompare, numParentSeqs);
 331
 332                         if(!ignoreSeq){
 333
 334                                 for(int i=0;i<minCompare.total;i++){
 335                                         char letter = minCompare.sequence[i];
 336
 337                                         errorForward[letter][i] += minCompare.weight;
 338                                         errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
 339                                 }
 340                         }
 341
 342                         if(qualFileName != "" && reportFileName != ""){
 343                                 report = ReportFile(reportFile);
 344
 345 //                              int origLength = report.getQueryLength();
 346                                 int startBase = report.getQueryStart();
 347                                 int endBase = report.getQueryEnd();
 348
 349                                 quality = QualityScores(qualFile);
 350
 351                                 if(!ignoreSeq){
 352                                         quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
 353                                         quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
 354                                         quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
 355                                 }
 356                         }
 357
 358                         if(minCompare.errorRate < threshold && !ignoreSeq){
 359                                 totalBases += (minCompare.total * minCompare.weight);
 360                                 totalMatches += minCompare.matches * minCompare.weight;
 361                                 if(minCompare.mismatches > maxMismatch){
 362                                         maxMismatch = minCompare.mismatches;
 363                                         misMatchCounts.resize(maxMismatch + 1, 0);
 364                                 }
 365                                 misMatchCounts[minCompare.mismatches] += minCompare.weight;
 366                                 numSeqs++;
 367
 368                                 megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
 369                         }
 370
 371                         index++;
 372
 373                         if(index % 1000 == 0){  m->mothurOut(toString(index) + '\n');   }
 374                 }
 375                 queryFile.close();
 376                 errorSummaryFile.close();
 377                 errorSeqFile.close();
 378
 379                 if(qualFileName != "" && reportFileName != ""){
 380                         printErrorQuality(qScoreErrorMap);
 381                         printQualityFR(qualForwardMap, qualReverseMap);
 382                 }
 383
 384                 printErrorFRFile(errorForward, errorReverse);
 385
 386                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 387
 388                 string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
 389                 ofstream errorCountFile;
 390                 m->openOutputFile(errorCountFileName, errorCountFile);
 391                 outputNames.push_back(errorCountFileName);  outputTypes["error.count"].push_back(errorCountFileName);
 392                 m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n\n");
 393                 m->mothurOut("Errors\tSequences\n");
 394                 errorCountFile << "Errors\tSequences\n";
 395                 for(int i=0;i<misMatchCounts.size();i++){
 396                         m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
 397                         errorCountFile << i << '\t' << misMatchCounts[i] << endl;
 398                 }
 399                 errorCountFile.close();
 400
 401                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 402
 403                 printSubMatrix();
 404
 405                 string megAlignmentFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.ref-query";
 406                 ofstream megAlignmentFile;
 407                 m->openOutputFile(megAlignmentFileName, megAlignmentFile);
 408                 outputNames.push_back(megAlignmentFileName);  outputTypes["error.ref-query"].push_back(megAlignmentFileName);
 409
 410                 for(int i=0;i<numRefs;i++){
 411                         megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
 412                         megAlignmentFile << megaAlignVector[i] << endl;
 413                 }
 414
 415
 416                 m->mothurOutEndLine();
 417                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 418                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 419                 m->mothurOutEndLine();
 420
 421                 return 0;
 422         }
 423         catch(exception& e) {
 424                 m->errorOut(e, "SeqErrorCommand", "execute");
 425                 exit(1);
 426         }
 427 }
 428
 429 //***************************************************************************************************************
 430
 431 void SeqErrorCommand::getReferences(){
 432         try {
 433
 434                 ifstream referenceFile;
 435                 m->openInputFile(referenceFileName, referenceFile);
 436
 437                 int numAmbigSeqs = 0;
 438
 439                 int maxStartPos = 0;
 440                 int minEndPos = 100000;
 441
 442                 while(referenceFile){
 443                         Sequence currentSeq(referenceFile);
 444                         int numAmbigs = currentSeq.getAmbigBases();
 445                         if(numAmbigs > 0){      numAmbigSeqs++; }
 446
 447                         int startPos = currentSeq.getStartPos();
 448                         if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 449
 450                         int endPos = currentSeq.getEndPos();
 451                         if(endPos < minEndPos)          {       minEndPos = endPos;             }
 452                         referenceSeqs.push_back(currentSeq);
 453                         m->gobble(referenceFile);
 454                 }
 455                 referenceFile.close();
 456                 numRefs = referenceSeqs.size();
 457
 458
 459                 for(int i=0;i<numRefs;i++){
 460                         referenceSeqs[i].padToPos(maxStartPos);
 461                         referenceSeqs[i].padFromPos(minEndPos);
 462                 }
 463
 464                 if(numAmbigSeqs != 0){
 465                         m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
 466                 }
 467
 468         }
 469         catch(exception& e) {
 470                 m->errorOut(e, "SeqErrorCommand", "getReferences");
 471                 exit(1);
 472         }
 473 }
 474
 475 //***************************************************************************************************************
 476
 477 Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
 478         try {
 479                 if(query.getAlignLength() != reference.getAlignLength()){
 480                         m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
 481                 }
 482                 int alignLength = query.getAlignLength();
 483
 484                 string q = query.getAligned();
 485                 string r = reference.getAligned();
 486
 487                 int started = 0;
 488                 Compare errors;
 489
 490                 for(int i=0;i<alignLength;i++){
 491                         if(r[i] != 'N' && q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){                  //      no missing data and no double gaps
 492                                 started = 1;
 493
 494                                 if(q[i] == 'A'){
 495                                         if(r[i] == 'A'){        errors.AA++;    errors.matches++;       errors.sequence += 'm'; }
 496                                         if(r[i] == 'T'){        errors.AT++;    errors.sequence += 's'; }
 497                                         if(r[i] == 'G'){        errors.AG++;    errors.sequence += 's'; }
 498                                         if(r[i] == 'C'){        errors.AC++;    errors.sequence += 's'; }
 499                                         if(r[i] == '-'){        errors.Ai++;    errors.sequence += 'i'; }
 500                                 }
 501                                 else if(q[i] == 'T'){
 502                                         if(r[i] == 'A'){        errors.TA++;    errors.sequence += 's'; }
 503                                         if(r[i] == 'T'){        errors.TT++;    errors.matches++;       errors.sequence += 'm'; }
 504                                         if(r[i] == 'G'){        errors.TG++;    errors.sequence += 's'; }
 505                                         if(r[i] == 'C'){        errors.TC++;    errors.sequence += 's'; }
 506                                         if(r[i] == '-'){        errors.Ti++;    errors.sequence += 'i'; }
 507                                 }
 508                                 else if(q[i] == 'G'){
 509                                         if(r[i] == 'A'){        errors.GA++;    errors.sequence += 's'; }
 510                                         if(r[i] == 'T'){        errors.GT++;    errors.sequence += 's'; }
 511                                         if(r[i] == 'G'){        errors.GG++;    errors.matches++;       errors.sequence += 'm'; }
 512                                         if(r[i] == 'C'){        errors.GC++;    errors.sequence += 's'; }
 513                                         if(r[i] == '-'){        errors.Gi++;    errors.sequence += 'i'; }
 514                                 }
 515                                 else if(q[i] == 'C'){
 516                                         if(r[i] == 'A'){        errors.CA++;    errors.sequence += 's'; }
 517                                         if(r[i] == 'T'){        errors.CT++;    errors.sequence += 's'; }
 518                                         if(r[i] == 'G'){        errors.CG++;    errors.sequence += 's'; }
 519                                         if(r[i] == 'C'){        errors.CC++;    errors.matches++;       errors.sequence += 'm'; }
 520                                         if(r[i] == '-'){        errors.Ci++;    errors.sequence += 'i'; }
 521                                 }
 522                                 else if(q[i] == 'N'){
 523                                         if(r[i] == 'A'){        errors.NA++;    errors.sequence += 'a'; }
 524                                         if(r[i] == 'T'){        errors.NT++;    errors.sequence += 'a'; }
 525                                         if(r[i] == 'G'){        errors.NG++;    errors.sequence += 'a'; }
 526                                         if(r[i] == 'C'){        errors.NC++;    errors.sequence += 'a'; }
 527                                         if(r[i] == '-'){        errors.Ni++;    errors.sequence += 'a'; }
 528                                 }
 529                                 else if(q[i] == '-' && r[i] != '-'){
 530                                         if(r[i] == 'A'){        errors.dA++;    errors.sequence += 'd'; }
 531                                         if(r[i] == 'T'){        errors.dT++;    errors.sequence += 'd'; }
 532                                         if(r[i] == 'G'){        errors.dG++;    errors.sequence += 'd'; }
 533                                         if(r[i] == 'C'){        errors.dC++;    errors.sequence += 'd'; }
 534                                 }
 535                                 errors.total++;
 536
 537                         }
 538                         else if(q[i] == '.' && r[i] != '.'){            //      reference extends beyond query
 539                                 if(started == 1){       break;  }
 540                         }
 541                         else if(q[i] != '.' && r[i] == '.'){            //      query extends beyond reference
 542                                 if(started == 1){       break;  }
 543                         }
 544                         else if(q[i] == '.' && r[i] == '.'){            //      both are missing data
 545                                 if(started == 1){       break;  }
 546                         }
 547
 548                 }
 549                 errors.mismatches = errors.total-errors.matches;
 550                 errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
 551                 errors.queryName = query.getName();
 552                 errors.refName = reference.getName();
 553
 554                 return errors;
 555         }
 556         catch(exception& e) {
 557                 m->errorOut(e, "SeqErrorCommand", "getErrors");
 558                 exit(1);
 559         }
 560 }
 561
 562 //***************************************************************************************************************
 563
 564 map<string, int> SeqErrorCommand::getWeights(){
 565         ifstream nameFile;
 566         m->openInputFile(namesFileName, nameFile);
 567
 568         string seqName;
 569         string redundantSeqs;
 570         map<string, int> nameCountMap;
 571
 572         while(nameFile){
 573                 nameFile >> seqName >> redundantSeqs;
 574                 nameCountMap[seqName] = m->getNumNames(redundantSeqs);
 575                 m->gobble(nameFile);
 576         }
 577         return nameCountMap;
 578 }
 579
 580
 581 //***************************************************************************************************************
 582
 583 void SeqErrorCommand::printErrorHeader(){
 584         try {
 585                 errorSummaryFile << "query\treference\tweight\t";
 586                 errorSummaryFile << "AA\tAT\tAG\tAC\tTA\tTT\tTG\tTC\tGA\tGT\tGG\tGC\tCA\tCT\tCG\tCC\tNA\tNT\tNG\tNC\tAi\tTi\tGi\tCi\tNi\tdA\tdT\tdG\tdC\t";
 587                 errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
 588
 589                 errorSummaryFile << setprecision(6);
 590                 errorSummaryFile.setf(ios::fixed);
 591         }
 592         catch(exception& e) {
 593                 m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
 594                 exit(1);
 595         }
 596 }
 597
 598 //***************************************************************************************************************
 599
 600 void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs){
 601         try {
 602
 603                 errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
 604                 errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
 605                 errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
 606                 errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
 607                 errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
 608                 errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
 609                 errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
 610                 errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
 611
 612                 errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t';                  //insertions
 613                 errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t';                  //deletions
 614                 errorSummaryFile << error.mismatches - (error.Ai + error.Ti + error.Gi + error.Ci) - (error.dA + error.dT + error.dG + error.dC) - (error.NA + error.NT + error.NG + error.NC + error.Ni) << '\t';      //substitutions
 615                 errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t';       //ambiguities
 616                 errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
 617
 618                 errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
 619
 620                 int a=0;                int t=1;                int g=2;                int c=3;
 621                 int gap=4;              int n=5;
 622
 623                 if(numParentSeqs == 1 || ignoreChimeras == 0){
 624                         substitutionMatrix[a][a] += error.weight * error.AA;
 625                         substitutionMatrix[a][t] += error.weight * error.TA;
 626                         substitutionMatrix[a][g] += error.weight * error.GA;
 627                         substitutionMatrix[a][c] += error.weight * error.CA;
 628                         substitutionMatrix[a][gap] += error.weight * error.dA;
 629                         substitutionMatrix[a][n] += error.weight * error.NA;
 630
 631                         substitutionMatrix[t][a] += error.weight * error.AT;
 632                         substitutionMatrix[t][t] += error.weight * error.TT;
 633                         substitutionMatrix[t][g] += error.weight * error.GT;
 634                         substitutionMatrix[t][c] += error.weight * error.CT;
 635                         substitutionMatrix[t][gap] += error.weight * error.dT;
 636                         substitutionMatrix[t][n] += error.weight * error.NT;
 637
 638                         substitutionMatrix[g][a] += error.weight * error.AG;
 639                         substitutionMatrix[g][t] += error.weight * error.TG;
 640                         substitutionMatrix[g][g] += error.weight * error.GG;
 641                         substitutionMatrix[g][c] += error.weight * error.CG;
 642                         substitutionMatrix[g][gap] += error.weight * error.dG;
 643                         substitutionMatrix[g][n] += error.weight * error.NG;
 644
 645                         substitutionMatrix[c][a] += error.weight * error.AC;
 646                         substitutionMatrix[c][t] += error.weight * error.TC;
 647                         substitutionMatrix[c][g] += error.weight * error.GC;
 648                         substitutionMatrix[c][c] += error.weight * error.CC;
 649                         substitutionMatrix[c][gap] += error.weight * error.dC;
 650                         substitutionMatrix[c][n] += error.weight * error.NC;
 651
 652                         substitutionMatrix[gap][a] += error.weight * error.Ai;
 653                         substitutionMatrix[gap][t] += error.weight * error.Ti;
 654                         substitutionMatrix[gap][g] += error.weight * error.Gi;
 655                         substitutionMatrix[gap][c] += error.weight * error.Ci;
 656                         substitutionMatrix[gap][n] += error.weight * error.Ni;
 657                 }
 658         }
 659         catch(exception& e) {
 660                 m->errorOut(e, "SeqErrorCommand", "printErrorData");
 661                 exit(1);
 662         }
 663 }
 664
 665 //***************************************************************************************************************
 666
 667 void SeqErrorCommand::printSubMatrix(){
 668         try {
 669                 string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
 670                 ofstream subMatrixFile;
 671                 m->openOutputFile(subMatrixFileName, subMatrixFile);
 672                 outputNames.push_back(subMatrixFileName);  outputTypes["error.matrix"].push_back(subMatrixFileName);
 673                 vector<string> bases(6);
 674                 bases[0] = "A";
 675                 bases[1] = "T";
 676                 bases[2] = "G";
 677                 bases[3] = "C";
 678                 bases[4] = "Gap";
 679                 bases[5] = "N";
 680                 vector<int> refSums(5,1);
 681
 682                 for(int i=0;i<5;i++){
 683                         subMatrixFile << "\tr" << bases[i];
 684
 685                         for(int j=0;j<6;j++){
 686                                 refSums[i] += substitutionMatrix[i][j];
 687                         }
 688                 }
 689                 subMatrixFile << endl;
 690
 691                 for(int i=0;i<6;i++){
 692                         subMatrixFile << 'q' << bases[i];
 693                         for(int j=0;j<5;j++){
 694                                 subMatrixFile << '\t' << substitutionMatrix[j][i];
 695                         }
 696                         subMatrixFile << endl;
 697                 }
 698
 699                 subMatrixFile << "total";
 700                 for(int i=0;i<5;i++){
 701                         subMatrixFile << '\t' << refSums[i];
 702                 }
 703                 subMatrixFile << endl;
 704                 subMatrixFile.close();
 705         }
 706         catch(exception& e) {
 707                 m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
 708                 exit(1);
 709         }
 710 }
 711 //***************************************************************************************************************
 712
 713 void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
 714         try{
 715                 string errorForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.forward";
 716                 ofstream errorForwardFile;
 717                 m->openOutputFile(errorForwardFileName, errorForwardFile);
 718                 outputNames.push_back(errorForwardFileName);  outputTypes["error.forward"].push_back(errorForwardFileName);
 719
 720                 errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
 721                 for(int i=0;i<maxLength;i++){
 722                         float match = (float)errorForward['m'][i];
 723                         float subst = (float)errorForward['s'][i];
 724                         float insert = (float)errorForward['i'][i];
 725                         float del = (float)errorForward['d'][i];
 726                         float amb = (float)errorForward['a'][i];
 727                         float total = match + subst + insert + del + amb;
 728                         if(total == 0){ break;  }
 729                         errorForwardFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
 730                 }
 731                 errorForwardFile.close();
 732
 733                 string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
 734                 ofstream errorReverseFile;
 735                 m->openOutputFile(errorReverseFileName, errorReverseFile);
 736                 outputNames.push_back(errorReverseFileName);  outputTypes["error.reverse"].push_back(errorReverseFileName);
 737
 738                 errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
 739                 for(int i=0;i<maxLength;i++){
 740                         float match = (float)errorReverse['m'][i];
 741                         float subst = (float)errorReverse['s'][i];
 742                         float insert = (float)errorReverse['i'][i];
 743                         float del = (float)errorReverse['d'][i];
 744                         float amb = (float)errorReverse['a'][i];
 745                         float total = match + subst + insert + del + amb;
 746                         if(total == 0){ break;  }
 747                         errorReverseFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
 748                 }
 749                 errorReverseFile.close();
 750         }
 751         catch(exception& e) {
 752                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
 753                 exit(1);
 754         }
 755 }
 756
 757 //***************************************************************************************************************
 758
 759 void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
 760         try{
 761
 762                 string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
 763                 ofstream errorQualityFile;
 764                 m->openOutputFile(errorQualityFileName, errorQualityFile);
 765                 outputNames.push_back(errorQualityFileName);  outputTypes["error.quality"].push_back(errorQualityFileName);
 766
 767                 errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
 768                 for(int i=0;i<41;i++){
 769                         errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
 770                 }
 771                 errorQualityFile.close();
 772         }
 773         catch(exception& e) {
 774                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
 775                 exit(1);
 776         }
 777 }
 778
 779
 780 //***************************************************************************************************************
 781
 782 void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
 783
 784         try{
 785                 int numRows = 0;
 786                 int numColumns = qualForwardMap[0].size();
 787
 788                 for(int i=0;i<qualForwardMap.size();i++){
 789                         for(int j=0;j<numColumns;j++){
 790                                 if(qualForwardMap[i][j] != 0){
 791                                         if(numRows < i)         {       numRows = i+20;         }
 792                                 }
 793                         }
 794                 }
 795
 796                 string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
 797                 ofstream qualityForwardFile;
 798                 m->openOutputFile(qualityForwardFileName, qualityForwardFile);
 799                 outputNames.push_back(qualityForwardFileName);  outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
 800
 801                 for(int i=0;i<numColumns;i++){  qualityForwardFile << '\t' << i;        }       qualityForwardFile << endl;
 802
 803                 for(int i=0;i<numRows;i++){
 804                         qualityForwardFile << i+1;
 805                         for(int j=0;j<numColumns;j++){
 806                                 qualityForwardFile << '\t' << qualForwardMap[i][j];
 807                         }
 808
 809                         qualityForwardFile << endl;
 810                 }
 811                 qualityForwardFile.close();
 812
 813
 814                 string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
 815                 ofstream qualityReverseFile;
 816                 m->openOutputFile(qualityReverseFileName, qualityReverseFile);
 817                 outputNames.push_back(qualityReverseFileName);  outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
 818
 819                 for(int i=0;i<numColumns;i++){  qualityReverseFile << '\t' << i;        }       qualityReverseFile << endl;
 820                 for(int i=0;i<numRows;i++){
 821
 822                         qualityReverseFile << i+1;
 823                         for(int j=0;j<numColumns;j++){
 824                                 qualityReverseFile << '\t' << qualReverseMap[i][j];
 825                         }
 826                         qualityReverseFile << endl;
 827                 }
 828                 qualityReverseFile.close();
 829         }
 830         catch(exception& e) {
 831                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
 832                 exit(1);
 833         }
 834
 835 }
 836
 837 //***************************************************************************************************************