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 #include "filterseqscommand.h"
  15
  16
  17 //**********************************************************************************************************************
  18 vector<string> SeqErrorCommand::setParameters(){
  19         try {
  20                 CommandParameter pquery("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pquery);
  21                 CommandParameter preference("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(preference);
  22                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(pqfile);
  23                 CommandParameter preport("report", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(preport);
  24                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  25                 CommandParameter pignorechimeras("ignorechimeras", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pignorechimeras);
  26                 CommandParameter pthreshold("threshold", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pthreshold);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  28                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  29                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  30                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "SeqErrorCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string SeqErrorCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The seq.error command reads a query alignment file and a reference alignment file and creates .....\n";
  46                 helpString += "The fasta parameter...\n";
  47                 helpString += "The reference parameter...\n";
  48                 helpString += "The qfile parameter...\n";
  49                 helpString += "The report parameter...\n";
  50                 helpString += "The name parameter...\n";
  51                 helpString += "The ignorechimeras parameter...\n";
  52                 helpString += "The threshold parameter...\n";
  53                 helpString += "The processors parameter...\n";
  54                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  55                 helpString += "Example seq.error(...).\n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  57                 helpString += "For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n";
  58                 return helpString;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "SeqErrorCommand", "getHelpString");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 SeqErrorCommand::SeqErrorCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["error.summary"] = tempOutNames;
  72                 outputTypes["error.seq"] = tempOutNames;
  73                 outputTypes["error.quality"] = tempOutNames;
  74                 outputTypes["error.qual.forward"] = tempOutNames;
  75                 outputTypes["error.qual.reverse"] = tempOutNames;
  76                 outputTypes["error.forward"] = tempOutNames;
  77                 outputTypes["error.reverse"] = tempOutNames;
  78                 outputTypes["error.count"] = tempOutNames;
  79                 outputTypes["error.matrix"] = tempOutNames;
  80         }
  81         catch(exception& e) {
  82                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
  83                 exit(1);
  84         }
  85 }
  86 //***************************************************************************************************************
  87
  88 SeqErrorCommand::SeqErrorCommand(string option)  {
  89         try {
  90
  91                 abort = false; calledHelp = false;
  92                 rdb = ReferenceDB::getInstance();
  93
  94                 //allow user to run help
  95                 if(option == "help") { help(); abort = true; calledHelp = true; }
  96                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  97
  98                 else {
  99                         string temp;
 100                         vector<string> myArray = setParameters();
 101
 102                         OptionParser parser(option);
 103                         map<string,string> parameters = parser.getParameters();
 104
 105                         ValidParameters validParameter;
 106                         map<string,string>::iterator it;
 107
 108                         //check to make sure all parameters are valid for command
 109                         for (it = parameters.begin(); it != parameters.end(); it++) {
 110                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 111                         }
 112
 113                         //initialize outputTypes
 114                         vector<string> tempOutNames;
 115                         outputTypes["error.summary"] = tempOutNames;
 116                         outputTypes["error.seq"] = tempOutNames;
 117                         outputTypes["error.quality"] = tempOutNames;
 118                         outputTypes["error.qual.forward"] = tempOutNames;
 119                         outputTypes["error.qual.reverse"] = tempOutNames;
 120                         outputTypes["error.forward"] = tempOutNames;
 121                         outputTypes["error.reverse"] = tempOutNames;
 122                         outputTypes["error.count"] = tempOutNames;
 123                         outputTypes["error.matrix"] = 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("reference");
 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["reference"] = inputDir + it->second;                }
 145                                 }
 146
 147                                 it = parameters.find("name");
 148                                 //user has given a names 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["name"] = inputDir + it->second;             }
 153                                 }
 154
 155                                 it = parameters.find("qfile");
 156                                 //user has given a quality score file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("report");
 164                                 //user has given a alignment report file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["report"] = inputDir + it->second;           }
 169                                 }
 170
 171                         }
 172                         //check for required parameters
 173                         queryFileName = validParameter.validFile(parameters, "fasta", true);
 174                         if (queryFileName == "not found") {
 175                                 queryFileName = m->getFastaFile();
 176                                 if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 177                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 178                         }
 179                         else if (queryFileName == "not open") { abort = true; }
 180                         else { m->setFastaFile(queryFileName); }
 181
 182                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 183                         if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
 184                         else if (referenceFileName == "not open") { abort = true; }
 185
 186                         //check for optional parameters
 187                         namesFileName = validParameter.validFile(parameters, "name", true);
 188                         if(namesFileName == "not found"){       namesFileName = "";     }
 189                         else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
 190                         else { m->setNameFile(namesFileName); }
 191
 192                         qualFileName = validParameter.validFile(parameters, "qfile", true);
 193                         if(qualFileName == "not found"){        qualFileName = "";      }
 194                         else if (qualFileName == "not open") { qualFileName = ""; abort = true; }
 195                         else { m->setQualFile(qualFileName); }
 196
 197                         reportFileName = validParameter.validFile(parameters, "report", true);
 198                         if(reportFileName == "not found"){      reportFileName = "";    }
 199                         else if (reportFileName == "not open") { reportFileName = ""; abort = true; }
 200
 201                         if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
 202                                 m->mothurOut("if you use either a qual file or a report file, you have to have both.");
 203                                 m->mothurOutEndLine();
 204                                 abort = true;
 205                         }
 206
 207                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 208                         if (outputDir == "not found"){
 209                                 outputDir = "";
 210                                 outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
 211                         }
 212
 213                         //check for optional parameter and set defaults
 214                         // ...at some point should added some additional type checking...
 215                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
 216                         m->mothurConvert(temp, threshold);
 217
 218                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 219                         save = m->isTrue(temp);
 220                         rdb->save = save;
 221                         if (save) { //clear out old references
 222                                 rdb->clearMemory();
 223                         }
 224
 225                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 226                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 227                         if (referenceFileName == "not found") {
 228                                 //check for saved reference sequences
 229                                 if (rdb->referenceSeqs.size() != 0) {
 230                                         referenceFileName = "saved";
 231                                 }else {
 232                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 233                                         m->mothurOutEndLine();
 234                                         abort = true;
 235                                 }
 236                         }else if (referenceFileName == "not open") { abort = true; }
 237                         else {  if (save) {     rdb->setSavedReference(referenceFileName);      }       }
 238
 239
 240                         temp = validParameter.validFile(parameters, "ignorechimeras", false);   if (temp == "not found") { temp = "T"; }
 241                         ignoreChimeras = m->isTrue(temp);
 242
 243                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 244                         m->setProcessors(temp);
 245                         m->mothurConvert(temp, processors);
 246
 247                         substitutionMatrix.resize(6);
 248                         for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
 249                 }
 250         }
 251         catch(exception& e) {
 252                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 253                 exit(1);
 254         }
 255 }
 256 //***************************************************************************************************************
 257
 258 int SeqErrorCommand::execute(){
 259         try{
 260                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 261
 262                 int start = time(NULL);
 263                 maxLength = 2000;
 264                 totalBases = 0;
 265                 totalMatches = 0;
 266
 267                 string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
 268                 outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
 269
 270                 string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
 271                 outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
 272
 273                 string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
 274                 outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
 275
 276                 getReferences();        //read in reference sequences - make sure there's no ambiguous bases
 277
 278                 if(namesFileName != ""){        weights = getWeights(); }
 279
 280                 vector<unsigned long long> fastaFilePos;
 281                 vector<unsigned long long> qFilePos;
 282                 vector<unsigned long long> reportFilePos;
 283
 284                 setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
 285
 286                 if (m->control_pressed) { return 0; }
 287
 288                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 289                         lines.push_back(linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 290                         if (qualFileName != "") {  qLines.push_back(linePair(qFilePos[i], qFilePos[(i+1)]));  }
 291                         if (reportFileName != "") {  rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)]));  }
 292                 }
 293                 if(qualFileName == "")  {       qLines = lines; rLines = lines; } //fills with duds
 294
 295                 int numSeqs = 0;
 296 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 297                 if(processors == 1){
 298                         numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 299                 }else{
 300                         numSeqs = createProcesses(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName);
 301                 }
 302 #else
 303                 numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 304 #endif
 305
 306                 if(qualFileName != "" && reportFileName != ""){
 307                         printErrorQuality(qScoreErrorMap);
 308                         printQualityFR(qualForwardMap, qualReverseMap);
 309                 }
 310
 311                 printErrorFRFile(errorForward, errorReverse);
 312
 313                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 314
 315                 string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
 316                 ofstream errorCountFile;
 317                 m->openOutputFile(errorCountFileName, errorCountFile);
 318                 outputNames.push_back(errorCountFileName);  outputTypes["error.count"].push_back(errorCountFileName);
 319                 m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n");
 320                 m->mothurOut("Errors\tSequences\n");
 321                 errorCountFile << "Errors\tSequences\n";
 322                 for(int i=0;i<misMatchCounts.size();i++){
 323                         m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
 324                         errorCountFile << i << '\t' << misMatchCounts[i] << endl;
 325                 }
 326                 errorCountFile.close();
 327
 328                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 329
 330                 printSubMatrix();
 331
 332                 string megAlignmentFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.ref-query";
 333                 ofstream megAlignmentFile;
 334                 m->openOutputFile(megAlignmentFileName, megAlignmentFile);
 335                 outputNames.push_back(megAlignmentFileName);  outputTypes["error.ref-query"].push_back(megAlignmentFileName);
 336
 337                 for(int i=0;i<numRefs;i++){
 338                         megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
 339                         megAlignmentFile << megaAlignVector[i] << endl;
 340                 }
 341
 342                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
 343                 m->mothurOutEndLine();
 344
 345                 m->mothurOutEndLine();
 346                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 347                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 348                 m->mothurOutEndLine();
 349
 350                 return 0;
 351         }
 352         catch(exception& e) {
 353                 m->errorOut(e, "SeqErrorCommand", "execute");
 354                 exit(1);
 355         }
 356 }
 357 //**********************************************************************************************************************
 358 int SeqErrorCommand::createProcesses(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName) {
 359         try {
 360                 int process = 1;
 361                 processIDS.clear();
 362                 map<char, vector<int> >::iterator it;
 363                 int num = 0;
 364 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 365
 366                 //loop through and create all the processes you want
 367                 while (process != processors) {
 368                         int pid = fork();
 369
 370                         if (pid > 0) {
 371                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 372                                 process++;
 373                         }else if (pid == 0){
 374
 375                                 num = driver(filename, qFileName, rFileName, summaryFileName + toString(getpid()) + ".temp", errorOutputFileName+ toString(getpid()) + ".temp", chimeraOutputFileName + toString(getpid()) + ".temp", lines[process], qLines[process], rLines[process]);
 376
 377                                 //pass groupCounts to parent
 378                                 ofstream out;
 379                                 string tempFile = filename + toString(getpid()) + ".info.temp";
 380                                 m->openOutputFile(tempFile, out);
 381
 382                                 //output totalBases and totalMatches
 383                                 out << num << '\t' << totalBases << '\t' << totalMatches << endl << endl;
 384
 385                                 //output substitutionMatrix
 386                                 for(int i = 0; i < substitutionMatrix.size(); i++) {
 387                                         for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 388                                                 out << substitutionMatrix[i][j] << '\t';
 389                                         }
 390                                         out << endl;
 391                                 }
 392                                 out << endl;
 393
 394                                 //output qScoreErrorMap
 395                                 for (it = qScoreErrorMap.begin(); it != qScoreErrorMap.end(); it++) {
 396                                         vector<int> thisScoreErrorMap = it->second;
 397                                         out << it->first << '\t';
 398                                         for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 399                                                 out << thisScoreErrorMap[i] << '\t';
 400                                         }
 401                                         out << endl;
 402                                 }
 403                                 out << endl;
 404
 405                                 //output qualForwardMap
 406                                 for(int i = 0; i < qualForwardMap.size(); i++) {
 407                                         for (int j = 0; j < qualForwardMap[i].size(); j++) {
 408                                                 out << qualForwardMap[i][j] << '\t';
 409                                         }
 410                                         out << endl;
 411                                 }
 412                                 out << endl;
 413
 414                                 //output qualReverseMap
 415                                 for(int i = 0; i < qualReverseMap.size(); i++) {
 416                                         for (int j = 0; j < qualReverseMap[i].size(); j++) {
 417                                                 out << qualReverseMap[i][j] << '\t';
 418                                         }
 419                                         out << endl;
 420                                 }
 421                                 out << endl;
 422
 423
 424                                 //output errorForward
 425                                 for (it = errorForward.begin(); it != errorForward.end(); it++) {
 426                                         vector<int> thisErrorForward = it->second;
 427                                         out << it->first << '\t';
 428                                         for (int i = 0; i < thisErrorForward.size(); i++) {
 429                                                 out << thisErrorForward[i] << '\t';
 430                                         }
 431                                         out << endl;
 432                                 }
 433                                 out << endl;
 434
 435                                 //output errorReverse
 436                                 for (it = errorReverse.begin(); it != errorReverse.end(); it++) {
 437                                         vector<int> thisErrorReverse = it->second;
 438                                         out << it->first << '\t';
 439                                         for (int i = 0; i < thisErrorReverse.size(); i++) {
 440                                                 out << thisErrorReverse[i] << '\t';
 441                                         }
 442                                         out << endl;
 443                                 }
 444                                 out << endl;
 445
 446                                 //output misMatchCounts
 447                                 out << misMatchCounts.size() << endl;
 448                                 for (int j = 0; j < misMatchCounts.size(); j++) {
 449                                         out << misMatchCounts[j] << '\t';
 450                                 }
 451                                 out << endl;
 452
 453
 454                                 //output megaAlignVector
 455                                 for (int j = 0; j < megaAlignVector.size(); j++) {
 456                                         out << megaAlignVector[j] << endl;
 457                                 }
 458                                 out << endl;
 459
 460                                 out.close();
 461
 462                                 exit(0);
 463                         }else {
 464                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 465                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 466                                 exit(0);
 467                         }
 468                 }
 469
 470                 //do my part
 471                 num = driver(filename, qFileName, rFileName, summaryFileName, errorOutputFileName, chimeraOutputFileName, lines[0], qLines[0], rLines[0]);
 472
 473                 //force parent to wait until all the processes are done
 474                 for (int i=0;i<processIDS.size();i++) {
 475                         int temp = processIDS[i];
 476                         wait(&temp);
 477                 }
 478
 479                 //append files
 480                 for(int i=0;i<processIDS.size();i++){
 481
 482                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 483
 484                         m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
 485                         m->mothurRemove((summaryFileName + toString(processIDS[i]) + ".temp"));
 486                         m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
 487                         m->mothurRemove((errorOutputFileName + toString(processIDS[i]) + ".temp"));
 488                         m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
 489                         m->mothurRemove((chimeraOutputFileName + toString(processIDS[i]) + ".temp"));
 490
 491                         ifstream in;
 492                         string tempFile =  filename + toString(processIDS[i]) + ".info.temp";
 493                         m->openInputFile(tempFile, in);
 494
 495                         //input totalBases and totalMatches
 496                         int tempBases, tempMatches, tempNumSeqs;
 497                         in >> tempNumSeqs >> tempBases >> tempMatches; m->gobble(in);
 498                         totalBases += tempBases; totalMatches += tempMatches; num += tempNumSeqs;
 499
 500                         //input substitutionMatrix
 501                         int tempNum;
 502                         for(int i = 0; i < substitutionMatrix.size(); i++) {
 503                                 for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 504                                         in >> tempNum; substitutionMatrix[i][j] += tempNum;
 505                                 }
 506                                 m->gobble(in);
 507                         }
 508                         m->gobble(in);
 509
 510                         //input qScoreErrorMap
 511                         char first;
 512                         for (int i = 0; i < qScoreErrorMap.size(); i++) {
 513                                 in >> first;
 514                                 vector<int> thisScoreErrorMap = qScoreErrorMap[first];
 515
 516                                 for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 517                                         in >> tempNum; thisScoreErrorMap[i] += tempNum;
 518                                 }
 519                                 qScoreErrorMap[first] = thisScoreErrorMap;
 520                                 m->gobble(in);
 521                         }
 522                         m->gobble(in);
 523
 524                         //input qualForwardMap
 525                         for(int i = 0; i < qualForwardMap.size(); i++) {
 526                                 for (int j = 0; j < qualForwardMap[i].size(); j++) {
 527                                         in >> tempNum; qualForwardMap[i][j] += tempNum;
 528                                 }
 529                                 m->gobble(in);
 530                         }
 531                         m->gobble(in);
 532
 533                         //input qualReverseMap
 534                         for(int i = 0; i < qualReverseMap.size(); i++) {
 535                                 for (int j = 0; j < qualReverseMap[i].size(); j++) {
 536                                         in >> tempNum; qualReverseMap[i][j] += tempNum;
 537                                 }
 538                                 m->gobble(in);
 539                         }
 540                         m->gobble(in);
 541
 542                         //input errorForward
 543                         for (int i = 0; i < errorForward.size(); i++) {
 544                                 in >> first;
 545                                 vector<int> thisErrorForward = errorForward[first];
 546
 547                                 for (int i = 0; i < thisErrorForward.size(); i++) {
 548                                         in >> tempNum; thisErrorForward[i] += tempNum;
 549                                 }
 550                                 errorForward[first] = thisErrorForward;
 551                                 m->gobble(in);
 552                         }
 553                         m->gobble(in);
 554
 555                         //input errorReverse
 556                         for (int i = 0; i < errorReverse.size(); i++) {
 557                                 in >> first;
 558                                 vector<int> thisErrorReverse = errorReverse[first];
 559
 560                                 for (int i = 0; i < thisErrorReverse.size(); i++) {
 561                                         in >> tempNum; thisErrorReverse[i] += tempNum;
 562                                 }
 563                                 errorReverse[first] = thisErrorReverse;
 564                                 m->gobble(in);
 565                         }
 566                         m->gobble(in);
 567
 568                         //input misMatchCounts
 569                         int misMatchSize;
 570                         in >> misMatchSize; m->gobble(in);
 571                         if (misMatchSize > misMatchCounts.size()) {     misMatchCounts.resize(misMatchSize, 0); }
 572                         for (int j = 0; j < misMatchCounts.size(); j++) {
 573                                 in >> tempNum; misMatchCounts[j] += tempNum;
 574                         }
 575                         m->gobble(in);
 576
 577                         //input megaAlignVector
 578                         string thisLine;
 579                         for (int j = 0; j < megaAlignVector.size(); j++) {
 580                                 thisLine = m->getline(in); m->gobble(in); megaAlignVector[j] += thisLine + '\n';
 581                         }
 582                         m->gobble(in);
 583
 584                         in.close(); m->mothurRemove(tempFile);
 585
 586                 }
 587 #endif
 588                 return num;
 589         }
 590         catch(exception& e) {
 591                 m->errorOut(e, "SeqErrorCommand", "createProcesses");
 592                 exit(1);
 593         }
 594 }
 595 //**********************************************************************************************************************
 596 int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
 597
 598         try {
 599                 ReportFile report;
 600                 QualityScores quality;
 601
 602                 misMatchCounts.resize(11, 0);
 603                 int maxMismatch = 0;
 604                 int numSeqs = 0;
 605
 606                 map<string, int>::iterator it;
 607                 qScoreErrorMap['m'].assign(41, 0);
 608                 qScoreErrorMap['s'].assign(41, 0);
 609                 qScoreErrorMap['i'].assign(41, 0);
 610                 qScoreErrorMap['a'].assign(41, 0);
 611
 612                 errorForward['m'].assign(maxLength,0);
 613                 errorForward['s'].assign(maxLength,0);
 614                 errorForward['i'].assign(maxLength,0);
 615                 errorForward['d'].assign(maxLength,0);
 616                 errorForward['a'].assign(maxLength,0);
 617
 618                 errorReverse['m'].assign(maxLength,0);
 619                 errorReverse['s'].assign(maxLength,0);
 620                 errorReverse['i'].assign(maxLength,0);
 621                 errorReverse['d'].assign(maxLength,0);
 622                 errorReverse['a'].assign(maxLength,0);
 623
 624                 //open inputfiles and go to beginning place for this processor
 625                 ifstream queryFile;
 626                 m->openInputFile(filename, queryFile);
 627                 queryFile.seekg(line.start);
 628
 629                 ifstream reportFile;
 630                 ifstream qualFile;
 631                 if(qFileName != "" && rFileName != ""){
 632                         m->openInputFile(qFileName, qualFile);
 633                         qualFile.seekg(qline.start);
 634
 635                         //gobble headers
 636                         if (rline.start == 0) {  report = ReportFile(reportFile, rFileName); }
 637                         else{
 638                                 m->openInputFile(rFileName, reportFile);
 639                                 reportFile.seekg(rline.start);
 640                         }
 641
 642                         qualForwardMap.resize(maxLength);
 643                         qualReverseMap.resize(maxLength);
 644                         for(int i=0;i<maxLength;i++){
 645                                 qualForwardMap[i].assign(41,0);
 646                                 qualReverseMap[i].assign(41,0);
 647                         }
 648                 }
 649
 650                 ofstream outChimeraReport;
 651                 m->openOutputFile(chimeraOutputFileName, outChimeraReport);
 652                 RefChimeraTest chimeraTest(referenceSeqs);
 653                 if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
 654
 655                 ofstream errorSummaryFile;
 656                 m->openOutputFile(summaryFileName, errorSummaryFile);
 657                 if (line.start == 0) { printErrorHeader(errorSummaryFile); }
 658
 659                 ofstream errorSeqFile;
 660                 m->openOutputFile(errorOutputFileName, errorSeqFile);
 661
 662                 megaAlignVector.resize(numRefs, "");
 663
 664                 int index = 0;
 665                 bool ignoreSeq = 0;
 666
 667                 bool moreSeqs = 1;
 668                 while (moreSeqs) {
 669
 670                         if (m->control_pressed) { queryFile.close(); if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); } outChimeraReport.close(); errorSummaryFile.close();errorSeqFile.close(); return 0; }
 671
 672                         Sequence query(queryFile);
 673
 674                         int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
 675                         int closestRefIndex = chimeraTest.getClosestRefIndex();
 676
 677                         if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
 678                         else                                                                                    {       ignoreSeq = 0;  }
 679
 680                         Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
 681
 682                         if(namesFileName != ""){
 683                                 it = weights.find(query.getName());
 684                                 minCompare.weight = it->second;
 685                         }
 686                         else{   minCompare.weight = 1;  }
 687
 688                         printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
 689
 690                         if(!ignoreSeq){
 691
 692                                 for(int i=0;i<minCompare.sequence.length();i++){
 693                                         char letter = minCompare.sequence[i];
 694
 695                                         if(letter != 'r'){
 696                                                 errorForward[letter][i] += minCompare.weight;
 697                                                 errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
 698                                         }
 699                                 }
 700                         }
 701
 702                         if(qualFileName != "" && reportFileName != ""){
 703                                 report = ReportFile(reportFile);
 704
 705                                 //                              int origLength = report.getQueryLength();
 706                                 int startBase = report.getQueryStart();
 707                                 int endBase = report.getQueryEnd();
 708
 709                                 quality = QualityScores(qualFile);
 710
 711                                 if(!ignoreSeq){
 712 //                                      cout << query.getName() << '\t';
 713
 714                                         quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
 715                                         quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
 716                                         quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
 717
 718 //                                      cout << endl;
 719                                 }
 720                         }
 721
 722                         if(minCompare.errorRate < threshold && !ignoreSeq){
 723                                 totalBases += (minCompare.total * minCompare.weight);
 724                                 totalMatches += minCompare.matches * minCompare.weight;
 725                                 if(minCompare.mismatches > maxMismatch){
 726                                         maxMismatch = minCompare.mismatches;
 727                                         misMatchCounts.resize(maxMismatch + 1, 0);
 728                                 }
 729                                 misMatchCounts[minCompare.mismatches] += minCompare.weight;
 730                                 numSeqs++;
 731
 732                                 megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
 733                         }
 734
 735                         index++;
 736
 737                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 738                                 unsigned long long pos = queryFile.tellg();
 739                                 if ((pos == -1) || (pos >= line.end)) { break; }
 740                         #else
 741                                 if (queryFile.eof()) { break; }
 742                         #endif
 743
 744                         if(index % 100 == 0){   m->mothurOut(toString(index) + '\n');   }
 745                 }
 746                 queryFile.close();
 747                 if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); }
 748                 errorSummaryFile.close();
 749                 errorSeqFile.close();
 750
 751                 //report progress
 752                 if(index % 100 != 0){   m->mothurOut(toString(index) + '\n');   }
 753
 754                 return index;
 755         }
 756         catch(exception& e) {
 757                 m->errorOut(e, "SeqErrorCommand", "driver");
 758                 exit(1);
 759         }
 760 }
 761 //***************************************************************************************************************
 762
 763 void SeqErrorCommand::getReferences(){
 764         try {
 765                 int numAmbigSeqs = 0;
 766
 767                 int maxStartPos = 0;
 768                 int minEndPos = 100000;
 769
 770                 if (referenceFileName == "saved") {
 771                         int start = time(NULL);
 772                         m->mothurOutEndLine();  m->mothurOut("Using sequences from " + rdb->getSavedReference() + " that are saved in memory.");        m->mothurOutEndLine();
 773
 774                         for (int i = 0; i < rdb->referenceSeqs.size(); i++) {
 775                                 int numAmbigs = rdb->referenceSeqs[i].getAmbigBases();
 776                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 777
 778                                 //                      int startPos = rdb->referenceSeqs[i].getStartPos();
 779                                 //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 780                                 //
 781                                 //                      int endPos = rdb->referenceSeqs[i].getEndPos();
 782                                 //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 783
 784                                 referenceSeqs.push_back(rdb->referenceSeqs[i]);
 785                         }
 786                         referenceFileName = rdb->getSavedReference();
 787
 788                         m->mothurOut("It took " + toString(time(NULL) - start) + " to load " + toString(rdb->referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 789
 790                 }else {
 791                         int start = time(NULL);
 792
 793                         ifstream referenceFile;
 794                         m->openInputFile(referenceFileName, referenceFile);
 795
 796                         while(referenceFile){
 797                                 Sequence currentSeq(referenceFile);
 798                                 int numAmbigs = currentSeq.getAmbigBases();
 799                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 800
 801         //                      int startPos = currentSeq.getStartPos();
 802         //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 803         //
 804         //                      int endPos = currentSeq.getEndPos();
 805         //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 806                                 referenceSeqs.push_back(currentSeq);
 807
 808                                 if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
 809
 810                                 m->gobble(referenceFile);
 811                         }
 812                         referenceFile.close();
 813
 814                         m->mothurOut("It took " + toString(time(NULL) - start) + " to read " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 815                 }
 816
 817                 numRefs = referenceSeqs.size();
 818
 819                 for(int i=0;i<numRefs;i++){
 820                         referenceSeqs[i].padToPos(maxStartPos);
 821                         referenceSeqs[i].padFromPos(minEndPos);
 822                 }
 823
 824                 if(numAmbigSeqs != 0){
 825                         m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
 826                 }
 827
 828         }
 829         catch(exception& e) {
 830                 m->errorOut(e, "SeqErrorCommand", "getReferences");
 831                 exit(1);
 832         }
 833 }
 834
 835 //***************************************************************************************************************
 836
 837 Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
 838         try {
 839                 if(query.getAlignLength() != reference.getAlignLength()){
 840                         m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
 841                 }
 842                 int alignLength = query.getAlignLength();
 843
 844                 string q = query.getAligned();
 845                 string r = reference.getAligned();
 846
 847                 int started = 0;
 848                 Compare errors;
 849
 850                 for(int i=0;i<alignLength;i++){
 851 //                      cout << r[i] << '\t' << q[i] << '\t';
 852                         if(q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){                 //      no missing data and no double gaps
 853                                 if(r[i] != 'N'){
 854                                         started = 1;
 855
 856                                         if(q[i] == 'A'){
 857                                                 if(r[i] == 'A'){        errors.AA++;    errors.matches++;       errors.sequence += 'm'; }
 858                                                 if(r[i] == 'T'){        errors.AT++;    errors.sequence += 's'; }
 859                                                 if(r[i] == 'G'){        errors.AG++;    errors.sequence += 's'; }
 860                                                 if(r[i] == 'C'){        errors.AC++;    errors.sequence += 's'; }
 861                                                 if(r[i] == '-'){        errors.Ai++;    errors.sequence += 'i'; }
 862                                         }
 863                                         else if(q[i] == 'T'){
 864                                                 if(r[i] == 'A'){        errors.TA++;    errors.sequence += 's'; }
 865                                                 if(r[i] == 'T'){        errors.TT++;    errors.matches++;       errors.sequence += 'm'; }
 866                                                 if(r[i] == 'G'){        errors.TG++;    errors.sequence += 's'; }
 867                                                 if(r[i] == 'C'){        errors.TC++;    errors.sequence += 's'; }
 868                                                 if(r[i] == '-'){        errors.Ti++;    errors.sequence += 'i'; }
 869                                         }
 870                                         else if(q[i] == 'G'){
 871                                                 if(r[i] == 'A'){        errors.GA++;    errors.sequence += 's'; }
 872                                                 if(r[i] == 'T'){        errors.GT++;    errors.sequence += 's'; }
 873                                                 if(r[i] == 'G'){        errors.GG++;    errors.matches++;       errors.sequence += 'm'; }
 874                                                 if(r[i] == 'C'){        errors.GC++;    errors.sequence += 's'; }
 875                                                 if(r[i] == '-'){        errors.Gi++;    errors.sequence += 'i'; }
 876                                         }
 877                                         else if(q[i] == 'C'){
 878                                                 if(r[i] == 'A'){        errors.CA++;    errors.sequence += 's'; }
 879                                                 if(r[i] == 'T'){        errors.CT++;    errors.sequence += 's'; }
 880                                                 if(r[i] == 'G'){        errors.CG++;    errors.sequence += 's'; }
 881                                                 if(r[i] == 'C'){        errors.CC++;    errors.matches++;       errors.sequence += 'm'; }
 882                                                 if(r[i] == '-'){        errors.Ci++;    errors.sequence += 'i'; }
 883                                         }
 884                                         else if(q[i] == 'N'){
 885                                                 if(r[i] == 'A'){        errors.NA++;    errors.sequence += 'a'; }
 886                                                 if(r[i] == 'T'){        errors.NT++;    errors.sequence += 'a'; }
 887                                                 if(r[i] == 'G'){        errors.NG++;    errors.sequence += 'a'; }
 888                                                 if(r[i] == 'C'){        errors.NC++;    errors.sequence += 'a'; }
 889                                                 if(r[i] == '-'){        errors.Ni++;    errors.sequence += 'a'; }
 890                                         }
 891                                         else if(q[i] == '-' && r[i] != '-'){
 892                                                 if(r[i] == 'A'){        errors.dA++;    errors.sequence += 'd'; }
 893                                                 if(r[i] == 'T'){        errors.dT++;    errors.sequence += 'd'; }
 894                                                 if(r[i] == 'G'){        errors.dG++;    errors.sequence += 'd'; }
 895                                                 if(r[i] == 'C'){        errors.dC++;    errors.sequence += 'd'; }
 896                                         }
 897                                         errors.total++;
 898                                 }
 899                                 else{
 900
 901                                         if(q[i] == '-'){
 902                                                 errors.sequence += 'd'; errors.total++;
 903                                         }
 904                                         else{
 905                                                 errors.sequence += 'r';
 906                                         }
 907                                 }
 908                         }
 909
 910                         else if(q[i] == '.' && r[i] != '.'){            //      reference extends beyond query
 911                                 if(started == 1){       break;  }
 912                         }
 913                         else if(q[i] != '.' && r[i] == '.'){            //      query extends beyond reference
 914                                 if(started == 1){       break;  }
 915                         }
 916                         else if(q[i] == '.' && r[i] == '.'){            //      both are missing data
 917                                 if(started == 1){       break;  }
 918                         }
 919 //                      cout << errors.sequence[errors.sequence.length()-1] << endl;
 920                 }
 921 //              cout << errors.sequence << endl;
 922                 errors.mismatches = errors.total-errors.matches;
 923                 errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
 924                 errors.queryName = query.getName();
 925                 errors.refName = reference.getName();
 926
 927                 return errors;
 928         }
 929         catch(exception& e) {
 930                 m->errorOut(e, "SeqErrorCommand", "getErrors");
 931                 exit(1);
 932         }
 933 }
 934
 935 //***************************************************************************************************************
 936
 937 map<string, int> SeqErrorCommand::getWeights(){
 938         ifstream nameFile;
 939         m->openInputFile(namesFileName, nameFile);
 940
 941         string seqName;
 942         string redundantSeqs;
 943         map<string, int> nameCountMap;
 944
 945         while(nameFile){
 946                 nameFile >> seqName >> redundantSeqs;
 947                 nameCountMap[seqName] = m->getNumNames(redundantSeqs);
 948                 m->gobble(nameFile);
 949         }
 950         return nameCountMap;
 951 }
 952
 953
 954 //***************************************************************************************************************
 955
 956 void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
 957         try {
 958                 errorSummaryFile << "query\treference\tweight\t";
 959                 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";
 960                 errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
 961
 962                 errorSummaryFile << setprecision(6);
 963                 errorSummaryFile.setf(ios::fixed);
 964         }
 965         catch(exception& e) {
 966                 m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
 967                 exit(1);
 968         }
 969 }
 970
 971 //***************************************************************************************************************
 972
 973 void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs, ofstream& errorSummaryFile, ofstream& errorSeqFile){
 974         try {
 975
 976                 errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
 977                 errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
 978                 errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
 979                 errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
 980                 errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
 981                 errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
 982                 errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
 983                 errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
 984
 985                 errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t';                  //insertions
 986                 errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t';                  //deletions
 987                 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
 988                 errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t';       //ambiguities
 989                 errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
 990
 991                 errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
 992
 993                 int a=0;                int t=1;                int g=2;                int c=3;
 994                 int gap=4;              int n=5;
 995
 996                 if(numParentSeqs == 1 || ignoreChimeras == 0){
 997                         substitutionMatrix[a][a] += error.weight * error.AA;
 998                         substitutionMatrix[a][t] += error.weight * error.TA;
 999                         substitutionMatrix[a][g] += error.weight * error.GA;
1000                         substitutionMatrix[a][c] += error.weight * error.CA;
1001                         substitutionMatrix[a][gap] += error.weight * error.dA;
1002                         substitutionMatrix[a][n] += error.weight * error.NA;
1003
1004                         substitutionMatrix[t][a] += error.weight * error.AT;
1005                         substitutionMatrix[t][t] += error.weight * error.TT;
1006                         substitutionMatrix[t][g] += error.weight * error.GT;
1007                         substitutionMatrix[t][c] += error.weight * error.CT;
1008                         substitutionMatrix[t][gap] += error.weight * error.dT;
1009                         substitutionMatrix[t][n] += error.weight * error.NT;
1010
1011                         substitutionMatrix[g][a] += error.weight * error.AG;
1012                         substitutionMatrix[g][t] += error.weight * error.TG;
1013                         substitutionMatrix[g][g] += error.weight * error.GG;
1014                         substitutionMatrix[g][c] += error.weight * error.CG;
1015                         substitutionMatrix[g][gap] += error.weight * error.dG;
1016                         substitutionMatrix[g][n] += error.weight * error.NG;
1017
1018                         substitutionMatrix[c][a] += error.weight * error.AC;
1019                         substitutionMatrix[c][t] += error.weight * error.TC;
1020                         substitutionMatrix[c][g] += error.weight * error.GC;
1021                         substitutionMatrix[c][c] += error.weight * error.CC;
1022                         substitutionMatrix[c][gap] += error.weight * error.dC;
1023                         substitutionMatrix[c][n] += error.weight * error.NC;
1024
1025                         substitutionMatrix[gap][a] += error.weight * error.Ai;
1026                         substitutionMatrix[gap][t] += error.weight * error.Ti;
1027                         substitutionMatrix[gap][g] += error.weight * error.Gi;
1028                         substitutionMatrix[gap][c] += error.weight * error.Ci;
1029                         substitutionMatrix[gap][n] += error.weight * error.Ni;
1030                 }
1031         }
1032         catch(exception& e) {
1033                 m->errorOut(e, "SeqErrorCommand", "printErrorData");
1034                 exit(1);
1035         }
1036 }
1037
1038 //***************************************************************************************************************
1039
1040 void SeqErrorCommand::printSubMatrix(){
1041         try {
1042                 string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
1043                 ofstream subMatrixFile;
1044                 m->openOutputFile(subMatrixFileName, subMatrixFile);
1045                 outputNames.push_back(subMatrixFileName);  outputTypes["error.matrix"].push_back(subMatrixFileName);
1046                 vector<string> bases(6);
1047                 bases[0] = "A";
1048                 bases[1] = "T";
1049                 bases[2] = "G";
1050                 bases[3] = "C";
1051                 bases[4] = "Gap";
1052                 bases[5] = "N";
1053                 vector<int> refSums(5,1);
1054
1055                 for(int i=0;i<5;i++){
1056                         subMatrixFile << "\tr" << bases[i];
1057
1058                         for(int j=0;j<6;j++){
1059                                 refSums[i] += substitutionMatrix[i][j];
1060                         }
1061                 }
1062                 subMatrixFile << endl;
1063
1064                 for(int i=0;i<6;i++){
1065                         subMatrixFile << 'q' << bases[i];
1066                         for(int j=0;j<5;j++){
1067                                 subMatrixFile << '\t' << substitutionMatrix[j][i];
1068                         }
1069                         subMatrixFile << endl;
1070                 }
1071
1072                 subMatrixFile << "total";
1073                 for(int i=0;i<5;i++){
1074                         subMatrixFile << '\t' << refSums[i];
1075                 }
1076                 subMatrixFile << endl;
1077                 subMatrixFile.close();
1078         }
1079         catch(exception& e) {
1080                 m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
1081                 exit(1);
1082         }
1083 }
1084 //***************************************************************************************************************
1085
1086 void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
1087         try{
1088                 string errorForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.forward";
1089                 ofstream errorForwardFile;
1090                 m->openOutputFile(errorForwardFileName, errorForwardFile);
1091                 outputNames.push_back(errorForwardFileName);  outputTypes["error.forward"].push_back(errorForwardFileName);
1092
1093                 errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1094                 for(int i=0;i<maxLength;i++){
1095                         float match = (float)errorForward['m'][i];
1096                         float subst = (float)errorForward['s'][i];
1097                         float insert = (float)errorForward['i'][i];
1098                         float del = (float)errorForward['d'][i];
1099                         float amb = (float)errorForward['a'][i];
1100                         float total = match + subst + insert + del + amb;
1101                         if(total == 0){ break;  }
1102                         errorForwardFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1103                 }
1104                 errorForwardFile.close();
1105
1106                 string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
1107                 ofstream errorReverseFile;
1108                 m->openOutputFile(errorReverseFileName, errorReverseFile);
1109                 outputNames.push_back(errorReverseFileName);  outputTypes["error.reverse"].push_back(errorReverseFileName);
1110
1111                 errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1112                 for(int i=0;i<maxLength;i++){
1113                         float match = (float)errorReverse['m'][i];
1114                         float subst = (float)errorReverse['s'][i];
1115                         float insert = (float)errorReverse['i'][i];
1116                         float del = (float)errorReverse['d'][i];
1117                         float amb = (float)errorReverse['a'][i];
1118                         float total = match + subst + insert + del + amb;
1119                         if(total == 0){ break;  }
1120                         errorReverseFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1121                 }
1122                 errorReverseFile.close();
1123         }
1124         catch(exception& e) {
1125                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1126                 exit(1);
1127         }
1128 }
1129
1130 //***************************************************************************************************************
1131
1132 void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
1133         try{
1134
1135                 string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
1136                 ofstream errorQualityFile;
1137                 m->openOutputFile(errorQualityFileName, errorQualityFile);
1138                 outputNames.push_back(errorQualityFileName);  outputTypes["error.quality"].push_back(errorQualityFileName);
1139
1140                 errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
1141                 for(int i=0;i<41;i++){
1142                         errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
1143                 }
1144                 errorQualityFile.close();
1145         }
1146         catch(exception& e) {
1147                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1148                 exit(1);
1149         }
1150 }
1151
1152
1153 //***************************************************************************************************************
1154
1155 void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
1156
1157         try{
1158                 int numRows = 0;
1159                 int numColumns = qualForwardMap[0].size();
1160
1161                 for(int i=0;i<qualForwardMap.size();i++){
1162                         for(int j=0;j<numColumns;j++){
1163                                 if(qualForwardMap[i][j] != 0){
1164                                         if(numRows < i)         {       numRows = i+20;         }
1165                                 }
1166                         }
1167                 }
1168
1169                 string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
1170                 ofstream qualityForwardFile;
1171                 m->openOutputFile(qualityForwardFileName, qualityForwardFile);
1172                 outputNames.push_back(qualityForwardFileName);  outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
1173
1174                 for(int i=0;i<numColumns;i++){  qualityForwardFile << '\t' << i;        }       qualityForwardFile << endl;
1175
1176                 for(int i=0;i<numRows;i++){
1177                         qualityForwardFile << i+1;
1178                         for(int j=0;j<numColumns;j++){
1179                                 qualityForwardFile << '\t' << qualForwardMap[i][j];
1180                         }
1181
1182                         qualityForwardFile << endl;
1183                 }
1184                 qualityForwardFile.close();
1185
1186
1187                 string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
1188                 ofstream qualityReverseFile;
1189                 m->openOutputFile(qualityReverseFileName, qualityReverseFile);
1190                 outputNames.push_back(qualityReverseFileName);  outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
1191
1192                 for(int i=0;i<numColumns;i++){  qualityReverseFile << '\t' << i;        }       qualityReverseFile << endl;
1193                 for(int i=0;i<numRows;i++){
1194
1195                         qualityReverseFile << i+1;
1196                         for(int j=0;j<numColumns;j++){
1197                                 qualityReverseFile << '\t' << qualReverseMap[i][j];
1198                         }
1199                         qualityReverseFile << endl;
1200                 }
1201                 qualityReverseFile.close();
1202         }
1203         catch(exception& e) {
1204                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1205                 exit(1);
1206         }
1207
1208 }
1209 /**************************************************************************************************/
1210
1211 int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long long>& fastaFilePos, vector<unsigned long long>& qfileFilePos, vector<unsigned long long>& rfileFilePos) {
1212         try {
1213 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1214                 //set file positions for fasta file
1215                 fastaFilePos = m->divideFile(filename, processors);
1216
1217                 if (qfilename == "") { return processors; }
1218
1219                 //get name of first sequence in each chunk
1220                 map<string, int> firstSeqNames;
1221                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
1222                         ifstream in;
1223                         m->openInputFile(filename, in);
1224                         in.seekg(fastaFilePos[i]);
1225
1226                         Sequence temp(in);
1227                         firstSeqNames[temp.getName()] = i;
1228
1229                         in.close();
1230                 }
1231
1232                 //make copy to use below
1233                 map<string, int> firstSeqNamesReport = firstSeqNames;
1234
1235                 //seach for filePos of each first name in the qfile and save in qfileFilePos
1236                 ifstream inQual;
1237                 m->openInputFile(qfilename, inQual);
1238
1239                 string input;
1240                 while(!inQual.eof()){
1241                         input = m->getline(inQual);
1242
1243                         if (input.length() != 0) {
1244                                 if(input[0] == '>'){ //this is a sequence name line
1245                                         istringstream nameStream(input);
1246
1247                                         string sname = "";  nameStream >> sname;
1248                                         sname = sname.substr(1);
1249
1250                                         map<string, int>::iterator it = firstSeqNames.find(sname);
1251
1252                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
1253                                                 unsigned long long pos = inQual.tellg();
1254                                                 qfileFilePos.push_back(pos - input.length() - 1);
1255                                                 firstSeqNames.erase(it);
1256                                         }
1257                                 }
1258                         }
1259
1260                         if (firstSeqNames.size() == 0) { break; }
1261                 }
1262                 inQual.close();
1263
1264                 if (firstSeqNames.size() != 0) {
1265                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
1266                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, aborting."); m->mothurOutEndLine();
1267                         }
1268                         m->control_pressed = true;
1269                         return processors;
1270                 }
1271
1272                 //get last file position of qfile
1273                 FILE * pFile;
1274                 unsigned long long size;
1275
1276                 //get num bytes in file
1277                 pFile = fopen (qfilename.c_str(),"rb");
1278                 if (pFile==NULL) perror ("Error opening file");
1279                 else{
1280                         fseek (pFile, 0, SEEK_END);
1281                         size=ftell (pFile);
1282                         fclose (pFile);
1283                 }
1284
1285                 qfileFilePos.push_back(size);
1286
1287                 //seach for filePos of each first name in the rfile and save in rfileFilePos
1288                 string junk;
1289                 ifstream inR;
1290                 m->openInputFile(rfilename, inR);
1291
1292                 //read column headers
1293                 for (int i = 0; i < 16; i++) {
1294                         if (!inR.eof()) {       inR >> junk;    }
1295                         else                    {       break;                  }
1296                 }
1297
1298                 while(!inR.eof()){
1299
1300                         if (m->control_pressed) { inR.close();  return processors; }
1301
1302                         input = m->getline(inR);
1303
1304                         if (input.length() != 0) {
1305
1306                                 istringstream nameStream(input);
1307                                 string sname = "";  nameStream >> sname;
1308
1309                                 map<string, int>::iterator it = firstSeqNamesReport.find(sname);
1310
1311                                 if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
1312                                         unsigned long long pos = inR.tellg();
1313                                         rfileFilePos.push_back(pos - input.length() - 1);
1314                                         firstSeqNamesReport.erase(it);
1315                                 }
1316                         }
1317
1318                         if (firstSeqNamesReport.size() == 0) { break; }
1319                         m->gobble(inR);
1320                 }
1321                 inR.close();
1322
1323                 if (firstSeqNamesReport.size() != 0) {
1324                         for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
1325                                 m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
1326                         }
1327                         m->control_pressed = true;
1328                         return processors;
1329                 }
1330
1331                 //get last file position of qfile
1332                 FILE * rFile;
1333                 unsigned long long sizeR;
1334
1335                 //get num bytes in file
1336                 rFile = fopen (rfilename.c_str(),"rb");
1337                 if (rFile==NULL) perror ("Error opening file");
1338                 else{
1339                         fseek (rFile, 0, SEEK_END);
1340                         sizeR=ftell (rFile);
1341                         fclose (rFile);
1342                 }
1343
1344                 rfileFilePos.push_back(sizeR);
1345
1346                 return processors;
1347
1348 #else
1349
1350                 fastaFilePos.push_back(0); qfileFilePos.push_back(0);
1351                 //get last file position of fastafile
1352                 FILE * pFile;
1353                 unsigned long long size;
1354
1355                 //get num bytes in file
1356                 pFile = fopen (filename.c_str(),"rb");
1357                 if (pFile==NULL) perror ("Error opening file");
1358                 else{
1359                         fseek (pFile, 0, SEEK_END);
1360                         size=ftell (pFile);
1361                         fclose (pFile);
1362                 }
1363                 fastaFilePos.push_back(size);
1364
1365                 //get last file position of fastafile
1366                 FILE * qFile;
1367
1368                 //get num bytes in file
1369                 qFile = fopen (qfilename.c_str(),"rb");
1370                 if (qFile==NULL) perror ("Error opening file");
1371                 else{
1372                         fseek (qFile, 0, SEEK_END);
1373                         size=ftell (qFile);
1374                         fclose (qFile);
1375                 }
1376                 qfileFilePos.push_back(size);
1377
1378                 return 1;
1379
1380 #endif
1381         }
1382         catch(exception& e) {
1383                 m->errorOut(e, "SeqErrorCommand", "setLines");
1384                 exit(1);
1385         }
1386 }
1387 //***************************************************************************************************************