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 vector<string> SeqErrorCommand::setParameters(){
  18         try {
  19                 CommandParameter pquery("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pquery);
  20                 CommandParameter preference("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(preference);
  21                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(pqfile);
  22                 CommandParameter preport("report", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(preport);
  23                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  24                 CommandParameter pignorechimeras("ignorechimeras", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pignorechimeras);
  25                 CommandParameter pthreshold("threshold", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pthreshold);
  26                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  27                 CommandParameter pfilter("filter", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pfilter);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "SeqErrorCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string SeqErrorCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The seq.error command reads a query alignment file and a reference alignment file and creates .....\n";
  45                 helpString += "Example seq.error(...).\n";
  46                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  47                 helpString += "For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n";
  48                 return helpString;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "SeqErrorCommand", "getHelpString");
  52                 exit(1);
  53         }
  54 }
  55 //**********************************************************************************************************************
  56 SeqErrorCommand::SeqErrorCommand(){
  57         try {
  58                 abort = true; calledHelp = true;
  59                 vector<string> tempOutNames;
  60                 outputTypes["error.summary"] = tempOutNames;
  61                 outputTypes["error.seq"] = tempOutNames;
  62                 outputTypes["error.quality"] = tempOutNames;
  63                 outputTypes["error.qual.forward"] = tempOutNames;
  64                 outputTypes["error.qual.reverse"] = tempOutNames;
  65                 outputTypes["error.forward"] = tempOutNames;
  66                 outputTypes["error.reverse"] = tempOutNames;
  67                 outputTypes["error.count"] = tempOutNames;
  68                 outputTypes["error.matrix"] = tempOutNames;
  69         }
  70         catch(exception& e) {
  71                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
  72                 exit(1);
  73         }
  74 }
  75 //***************************************************************************************************************
  76
  77 SeqErrorCommand::SeqErrorCommand(string option)  {
  78         try {
  79
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         string temp;
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters = parser.getParameters();
  92
  93                         ValidParameters validParameter;
  94                         map<string,string>::iterator it;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         //initialize outputTypes
 102                         vector<string> tempOutNames;
 103                         outputTypes["error.summary"] = tempOutNames;
 104                         outputTypes["error.seq"] = tempOutNames;
 105                         outputTypes["error.quality"] = tempOutNames;
 106                         outputTypes["error.qual.forward"] = tempOutNames;
 107                         outputTypes["error.qual.reverse"] = tempOutNames;
 108                         outputTypes["error.forward"] = tempOutNames;
 109                         outputTypes["error.reverse"] = tempOutNames;
 110                         outputTypes["error.count"] = tempOutNames;
 111                         outputTypes["error.matrix"] = tempOutNames;
 112
 113
 114                         //if the user changes the input directory command factory will send this info to us in the output parameter
 115                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 116                         if (inputDir == "not found"){   inputDir = "";          }
 117                         else {
 118                                 string path;
 119                                 it = parameters.find("fasta");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 125                                 }
 126
 127                                 it = parameters.find("reference");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 133                                 }
 134
 135                                 it = parameters.find("name");
 136                                 //user has given a names file
 137                                 if(it != parameters.end()){
 138                                         path = m->hasPath(it->second);
 139                                         //if the user has not given a path then, add inputdir. else leave path alone.
 140                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 141                                 }
 142
 143                                 it = parameters.find("qfile");
 144                                 //user has given a quality score file
 145                                 if(it != parameters.end()){
 146                                         path = m->hasPath(it->second);
 147                                         //if the user has not given a path then, add inputdir. else leave path alone.
 148                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 149                                 }
 150
 151                                 it = parameters.find("report");
 152                                 //user has given a alignment report file
 153                                 if(it != parameters.end()){
 154                                         path = m->hasPath(it->second);
 155                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                         if (path == "") {       parameters["report"] = inputDir + it->second;           }
 157                                 }
 158
 159                         }
 160                         //check for required parameters
 161                         queryFileName = validParameter.validFile(parameters, "fasta", true);
 162                         if (queryFileName == "not found") {
 163                                 queryFileName = m->getFastaFile();
 164                                 if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 165                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 166                         }
 167                         else if (queryFileName == "not open") { abort = true; }
 168
 169                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 170                         if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
 171                         else if (referenceFileName == "not open") { abort = true; }
 172
 173
 174                         //check for optional parameters
 175                         namesFileName = validParameter.validFile(parameters, "name", true);
 176                         if(namesFileName == "not found"){       namesFileName = "";     }
 177                         else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
 178
 179                         qualFileName = validParameter.validFile(parameters, "qfile", true);
 180                         if(qualFileName == "not found"){        qualFileName = "";      }
 181                         else if (qualFileName == "not open") { qualFileName = ""; abort = true; }
 182
 183                         reportFileName = validParameter.validFile(parameters, "report", true);
 184                         if(reportFileName == "not found"){      reportFileName = "";    }
 185                         else if (reportFileName == "not open") { reportFileName = ""; abort = true; }
 186
 187                         if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
 188                                 m->mothurOut("if you use either a qual file or a report file, you have to have both.");
 189                                 m->mothurOutEndLine();
 190                                 abort = true;
 191                         }
 192
 193                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 194                         if (outputDir == "not found"){
 195                                 outputDir = "";
 196                                 outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
 197                         }
 198
 199                         //check for optional parameter and set defaults
 200                         // ...at some point should added some additional type checking...
 201                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
 202                         convert(temp, threshold);
 203
 204                         temp = validParameter.validFile(parameters, "ignorechimeras", false);   if (temp == "not found") { temp = "T"; }
 205                         ignoreChimeras = m->isTrue(temp);
 206
 207                         temp = validParameter.validFile(parameters, "filter", false);   if (temp == "not found") { temp = "T"; }
 208                         filter = m->isTrue(temp);
 209
 210                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 211                         m->setProcessors(temp);
 212                         convert(temp, processors);
 213
 214                         substitutionMatrix.resize(6);
 215                         for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
 216                 }
 217         }
 218         catch(exception& e) {
 219                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 220                 exit(1);
 221         }
 222 }
 223 //***************************************************************************************************************
 224
 225 int SeqErrorCommand::execute(){
 226         try{
 227                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 228
 229                 int start = time(NULL);
 230                 maxLength = 2000;
 231                 totalBases = 0;
 232                 totalMatches = 0;
 233
 234                 //run vertical filter on query and reference files.
 235                 if (filter) {
 236                         string inputString = "fasta=" + queryFileName + "-" + referenceFileName;
 237                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 238                         m->mothurOut("Running command: filter.seqs(" + inputString + ") to improve processing time."); m->mothurOutEndLine();
 239
 240                         Command* filterCommand = new FilterSeqsCommand(inputString);
 241                         filterCommand->execute();
 242
 243                         map<string, vector<string> > filenames = filterCommand->getOutputFiles();
 244
 245                         delete filterCommand;
 246
 247                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 248
 249                         queryFileName = filenames["fasta"][0];
 250                         referenceFileName = filenames["fasta"][1];
 251                 }
 252
 253                 string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
 254                 outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
 255
 256                 string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
 257                 outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
 258
 259                 string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
 260                 outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
 261
 262                 getReferences();        //read in reference sequences - make sure there's no ambiguous bases
 263
 264                 if(namesFileName != ""){        weights = getWeights(); }
 265
 266                 vector<unsigned long int> fastaFilePos;
 267                 vector<unsigned long int> qFilePos;
 268                 vector<unsigned long int> reportFilePos;
 269
 270                 setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
 271
 272                 if (m->control_pressed) { return 0; }
 273
 274                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 275                         lines.push_back(linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 276                         if (qualFileName != "") {  qLines.push_back(linePair(qFilePos[i], qFilePos[(i+1)]));  }
 277                         if (reportFileName != "") {  rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)]));  }
 278                 }
 279                 if(qualFileName == "")  {       qLines = lines; rLines = lines; } //fills with duds
 280
 281                 int numSeqs = 0;
 282 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 283                 if(processors == 1){
 284                         numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 285                 }else{
 286                         numSeqs = createProcesses(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName);
 287                 }
 288 #else
 289                 numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 290 #endif
 291
 292                 if(qualFileName != "" && reportFileName != ""){
 293                         printErrorQuality(qScoreErrorMap);
 294                         printQualityFR(qualForwardMap, qualReverseMap);
 295                 }
 296
 297                 printErrorFRFile(errorForward, errorReverse);
 298
 299                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 300
 301                 string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
 302                 ofstream errorCountFile;
 303                 m->openOutputFile(errorCountFileName, errorCountFile);
 304                 outputNames.push_back(errorCountFileName);  outputTypes["error.count"].push_back(errorCountFileName);
 305                 m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n");
 306                 m->mothurOut("Errors\tSequences\n");
 307                 errorCountFile << "Errors\tSequences\n";
 308                 for(int i=0;i<misMatchCounts.size();i++){
 309                         m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
 310                         errorCountFile << i << '\t' << misMatchCounts[i] << endl;
 311                 }
 312                 errorCountFile.close();
 313
 314                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 315
 316                 printSubMatrix();
 317
 318                 string megAlignmentFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.ref-query";
 319                 ofstream megAlignmentFile;
 320                 m->openOutputFile(megAlignmentFileName, megAlignmentFile);
 321                 outputNames.push_back(megAlignmentFileName);  outputTypes["error.ref-query"].push_back(megAlignmentFileName);
 322
 323                 for(int i=0;i<numRefs;i++){
 324                         megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
 325                         megAlignmentFile << megaAlignVector[i] << endl;
 326                 }
 327
 328                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
 329                 m->mothurOutEndLine();
 330
 331                 m->mothurOutEndLine();
 332                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 333                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 334                 m->mothurOutEndLine();
 335
 336                 return 0;
 337         }
 338         catch(exception& e) {
 339                 m->errorOut(e, "SeqErrorCommand", "execute");
 340                 exit(1);
 341         }
 342 }
 343 //**********************************************************************************************************************
 344 int SeqErrorCommand::createProcesses(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName) {
 345         try {
 346                 int process = 1;
 347                 processIDS.clear();
 348                 map<char, vector<int> >::iterator it;
 349                 int num = 0;
 350
 351                 //loop through and create all the processes you want
 352                 while (process != processors) {
 353                         int pid = fork();
 354
 355                         if (pid > 0) {
 356                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 357                                 process++;
 358                         }else if (pid == 0){
 359
 360                                 num = driver(filename, qFileName, rFileName, summaryFileName + toString(getpid()) + ".temp", errorOutputFileName+ toString(getpid()) + ".temp", chimeraOutputFileName + toString(getpid()) + ".temp", lines[process], qLines[process], rLines[process]);
 361
 362                                 //pass groupCounts to parent
 363                                 ofstream out;
 364                                 string tempFile = filename + toString(getpid()) + ".info.temp";
 365                                 m->openOutputFile(tempFile, out);
 366
 367                                 //output totalBases and totalMatches
 368                                 out << num << '\t' << totalBases << '\t' << totalMatches << endl << endl;
 369
 370                                 //output substitutionMatrix
 371                                 for(int i = 0; i < substitutionMatrix.size(); i++) {
 372                                         for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 373                                                 out << substitutionMatrix[i][j] << '\t';
 374                                         }
 375                                         out << endl;
 376                                 }
 377                                 out << endl;
 378
 379                                 //output qScoreErrorMap
 380                                 for (it = qScoreErrorMap.begin(); it != qScoreErrorMap.end(); it++) {
 381                                         vector<int> thisScoreErrorMap = it->second;
 382                                         out << it->first << '\t';
 383                                         for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 384                                                 out << thisScoreErrorMap[i] << '\t';
 385                                         }
 386                                         out << endl;
 387                                 }
 388                                 out << endl;
 389
 390                                 //output qualForwardMap
 391                                 for(int i = 0; i < qualForwardMap.size(); i++) {
 392                                         for (int j = 0; j < qualForwardMap[i].size(); j++) {
 393                                                 out << qualForwardMap[i][j] << '\t';
 394                                         }
 395                                         out << endl;
 396                                 }
 397                                 out << endl;
 398
 399                                 //output qualReverseMap
 400                                 for(int i = 0; i < qualReverseMap.size(); i++) {
 401                                         for (int j = 0; j < qualReverseMap[i].size(); j++) {
 402                                                 out << qualReverseMap[i][j] << '\t';
 403                                         }
 404                                         out << endl;
 405                                 }
 406                                 out << endl;
 407
 408
 409                                 //output errorForward
 410                                 for (it = errorForward.begin(); it != errorForward.end(); it++) {
 411                                         vector<int> thisErrorForward = it->second;
 412                                         out << it->first << '\t';
 413                                         for (int i = 0; i < thisErrorForward.size(); i++) {
 414                                                 out << thisErrorForward[i] << '\t';
 415                                         }
 416                                         out << endl;
 417                                 }
 418                                 out << endl;
 419
 420                                 //output errorReverse
 421                                 for (it = errorReverse.begin(); it != errorReverse.end(); it++) {
 422                                         vector<int> thisErrorReverse = it->second;
 423                                         out << it->first << '\t';
 424                                         for (int i = 0; i < thisErrorReverse.size(); i++) {
 425                                                 out << thisErrorReverse[i] << '\t';
 426                                         }
 427                                         out << endl;
 428                                 }
 429                                 out << endl;
 430
 431                                 //output misMatchCounts
 432                                 out << misMatchCounts.size() << endl;
 433                                 for (int j = 0; j < misMatchCounts.size(); j++) {
 434                                         out << misMatchCounts[j] << '\t';
 435                                 }
 436                                 out << endl;
 437
 438
 439                                 //output megaAlignVector
 440                                 for (int j = 0; j < megaAlignVector.size(); j++) {
 441                                         out << megaAlignVector[j] << endl;
 442                                 }
 443                                 out << endl;
 444
 445                                 out.close();
 446
 447                                 exit(0);
 448                         }else {
 449                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 450                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 451                                 exit(0);
 452                         }
 453                 }
 454
 455                 //do my part
 456                 num = driver(filename, qFileName, rFileName, summaryFileName, errorOutputFileName, chimeraOutputFileName, lines[0], qLines[0], rLines[0]);
 457
 458                 //force parent to wait until all the processes are done
 459                 for (int i=0;i<processIDS.size();i++) {
 460                         int temp = processIDS[i];
 461                         wait(&temp);
 462                 }
 463
 464                 //append files
 465                 for(int i=0;i<processIDS.size();i++){
 466
 467                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 468
 469                         m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
 470                         remove((summaryFileName + toString(processIDS[i]) + ".temp").c_str());
 471                         m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
 472                         remove((errorOutputFileName + toString(processIDS[i]) + ".temp").c_str());
 473                         m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
 474                         remove((chimeraOutputFileName + toString(processIDS[i]) + ".temp").c_str());
 475
 476                         ifstream in;
 477                         string tempFile =  filename + toString(processIDS[i]) + ".info.temp";
 478                         m->openInputFile(tempFile, in);
 479
 480                         //input totalBases and totalMatches
 481                         int tempBases, tempMatches, tempNumSeqs;
 482                         in >> tempNumSeqs >> tempBases >> tempMatches; m->gobble(in);
 483                         totalBases += tempBases; totalMatches += tempMatches; num += tempNumSeqs;
 484
 485                         //input substitutionMatrix
 486                         int tempNum;
 487                         for(int i = 0; i < substitutionMatrix.size(); i++) {
 488                                 for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 489                                         in >> tempNum; substitutionMatrix[i][j] += tempNum;
 490                                 }
 491                                 m->gobble(in);
 492                         }
 493                         m->gobble(in);
 494
 495                         //input qScoreErrorMap
 496                         char first;
 497                         for (int i = 0; i < qScoreErrorMap.size(); i++) {
 498                                 in >> first;
 499                                 vector<int> thisScoreErrorMap = qScoreErrorMap[first];
 500
 501                                 for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 502                                         in >> tempNum; thisScoreErrorMap[i] += tempNum;
 503                                 }
 504                                 qScoreErrorMap[first] = thisScoreErrorMap;
 505                                 m->gobble(in);
 506                         }
 507                         m->gobble(in);
 508
 509                         //input qualForwardMap
 510                         for(int i = 0; i < qualForwardMap.size(); i++) {
 511                                 for (int j = 0; j < qualForwardMap[i].size(); j++) {
 512                                         in >> tempNum; qualForwardMap[i][j] += tempNum;
 513                                 }
 514                                 m->gobble(in);
 515                         }
 516                         m->gobble(in);
 517
 518                         //input qualReverseMap
 519                         for(int i = 0; i < qualReverseMap.size(); i++) {
 520                                 for (int j = 0; j < qualReverseMap[i].size(); j++) {
 521                                         in >> tempNum; qualReverseMap[i][j] += tempNum;
 522                                 }
 523                                 m->gobble(in);
 524                         }
 525                         m->gobble(in);
 526
 527                         //input errorForward
 528                         for (int i = 0; i < errorForward.size(); i++) {
 529                                 in >> first;
 530                                 vector<int> thisErrorForward = errorForward[first];
 531
 532                                 for (int i = 0; i < thisErrorForward.size(); i++) {
 533                                         in >> tempNum; thisErrorForward[i] += tempNum;
 534                                 }
 535                                 errorForward[first] = thisErrorForward;
 536                                 m->gobble(in);
 537                         }
 538                         m->gobble(in);
 539
 540                         //input errorReverse
 541                         for (int i = 0; i < errorReverse.size(); i++) {
 542                                 in >> first;
 543                                 vector<int> thisErrorReverse = errorReverse[first];
 544
 545                                 for (int i = 0; i < thisErrorReverse.size(); i++) {
 546                                         in >> tempNum; thisErrorReverse[i] += tempNum;
 547                                 }
 548                                 errorReverse[first] = thisErrorReverse;
 549                                 m->gobble(in);
 550                         }
 551                         m->gobble(in);
 552
 553                         //input misMatchCounts
 554                         int misMatchSize;
 555                         in >> misMatchSize; m->gobble(in);
 556                         if (misMatchSize > misMatchCounts.size()) {     misMatchCounts.resize(misMatchSize, 0); }
 557                         for (int j = 0; j < misMatchCounts.size(); j++) {
 558                                 in >> tempNum; misMatchCounts[j] += tempNum;
 559                         }
 560                         m->gobble(in);
 561
 562                         //input megaAlignVector
 563                         string thisLine;
 564                         for (int j = 0; j < megaAlignVector.size(); j++) {
 565                                 thisLine = m->getline(in); m->gobble(in); megaAlignVector[j] += thisLine + '\n';
 566                         }
 567                         m->gobble(in);
 568
 569                         in.close(); remove(tempFile.c_str());
 570
 571                 }
 572
 573                 return num;
 574         }
 575         catch(exception& e) {
 576                 m->errorOut(e, "SeqErrorCommand", "createProcesses");
 577                 exit(1);
 578         }
 579 }
 580 //**********************************************************************************************************************
 581 int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
 582
 583         try {
 584                 ReportFile report;
 585                 QualityScores quality;
 586
 587                 misMatchCounts.resize(11, 0);
 588                 int maxMismatch = 0;
 589                 int numSeqs = 0;
 590
 591                 map<string, int>::iterator it;
 592                 qScoreErrorMap['m'].assign(41, 0);
 593                 qScoreErrorMap['s'].assign(41, 0);
 594                 qScoreErrorMap['i'].assign(41, 0);
 595                 qScoreErrorMap['a'].assign(41, 0);
 596
 597                 errorForward['m'].assign(maxLength,0);
 598                 errorForward['s'].assign(maxLength,0);
 599                 errorForward['i'].assign(maxLength,0);
 600                 errorForward['d'].assign(maxLength,0);
 601                 errorForward['a'].assign(maxLength,0);
 602
 603                 errorReverse['m'].assign(maxLength,0);
 604                 errorReverse['s'].assign(maxLength,0);
 605                 errorReverse['i'].assign(maxLength,0);
 606                 errorReverse['d'].assign(maxLength,0);
 607                 errorReverse['a'].assign(maxLength,0);
 608
 609                 //open inputfiles and go to beginning place for this processor
 610                 ifstream queryFile;
 611                 m->openInputFile(filename, queryFile);
 612                 queryFile.seekg(line.start);
 613
 614                 ifstream reportFile;
 615                 ifstream qualFile;
 616                 if(qFileName != "" && rFileName != ""){
 617                         m->openInputFile(qFileName, qualFile);
 618                         qualFile.seekg(qline.start);
 619
 620                         //gobble headers
 621                         if (rline.start == 0) {  report = ReportFile(reportFile, rFileName); }
 622                         else{
 623                                 m->openInputFile(rFileName, reportFile);
 624                                 reportFile.seekg(rline.start);
 625                         }
 626
 627                         qualForwardMap.resize(maxLength);
 628                         qualReverseMap.resize(maxLength);
 629                         for(int i=0;i<maxLength;i++){
 630                                 qualForwardMap[i].assign(41,0);
 631                                 qualReverseMap[i].assign(41,0);
 632                         }
 633                 }
 634
 635                 ofstream outChimeraReport;
 636                 m->openOutputFile(chimeraOutputFileName, outChimeraReport);
 637                 RefChimeraTest chimeraTest(referenceSeqs);
 638                 if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
 639
 640                 ofstream errorSummaryFile;
 641                 m->openOutputFile(summaryFileName, errorSummaryFile);
 642                 if (line.start == 0) { printErrorHeader(errorSummaryFile); }
 643
 644                 ofstream errorSeqFile;
 645                 m->openOutputFile(errorOutputFileName, errorSeqFile);
 646
 647                 megaAlignVector.resize(numRefs, "");
 648
 649                 int index = 0;
 650                 bool ignoreSeq = 0;
 651
 652                 bool moreSeqs = 1;
 653                 while (moreSeqs) {
 654
 655                         if (m->control_pressed) { queryFile.close(); if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); } outChimeraReport.close(); errorSummaryFile.close();errorSeqFile.close(); return 0; }
 656
 657                         Sequence query(queryFile);
 658
 659                         int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
 660                         int closestRefIndex = chimeraTest.getClosestRefIndex();
 661
 662                         if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
 663                         else                                                                                    {       ignoreSeq = 0;  }
 664
 665                         Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
 666
 667                         if(namesFileName != ""){
 668                                 it = weights.find(query.getName());
 669                                 minCompare.weight = it->second;
 670                         }
 671                         else{   minCompare.weight = 1;  }
 672
 673                         printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
 674
 675                         if(!ignoreSeq){
 676
 677                                 for(int i=0;i<minCompare.sequence.length();i++){
 678                                         char letter = minCompare.sequence[i];
 679
 680                                         errorForward[letter][i] += minCompare.weight;
 681                                         errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
 682                                 }
 683                         }
 684
 685                         if(qualFileName != "" && reportFileName != ""){
 686                                 report = ReportFile(reportFile);
 687
 688                                 //                              int origLength = report.getQueryLength();
 689                                 int startBase = report.getQueryStart();
 690                                 int endBase = report.getQueryEnd();
 691
 692                                 quality = QualityScores(qualFile);
 693
 694                                 if(!ignoreSeq){
 695                                         quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
 696                                         quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
 697                                         quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
 698                                 }
 699                         }
 700
 701                         if(minCompare.errorRate < threshold && !ignoreSeq){
 702                                 totalBases += (minCompare.total * minCompare.weight);
 703                                 totalMatches += minCompare.matches * minCompare.weight;
 704                                 if(minCompare.mismatches > maxMismatch){
 705                                         maxMismatch = minCompare.mismatches;
 706                                         misMatchCounts.resize(maxMismatch + 1, 0);
 707                                 }
 708                                 misMatchCounts[minCompare.mismatches] += minCompare.weight;
 709                                 numSeqs++;
 710
 711                                 megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
 712                         }
 713
 714                         index++;
 715
 716                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 717                                 unsigned long int pos = queryFile.tellg();
 718                                 if ((pos == -1) || (pos >= line.end)) { break; }
 719                         #else
 720                                 if (queryFile.eof()) { break; }
 721                         #endif
 722
 723                         if(index % 100 == 0){   m->mothurOut(toString(index) + '\n');   }
 724                 }
 725                 queryFile.close();
 726                 if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); }
 727                 errorSummaryFile.close();
 728                 errorSeqFile.close();
 729
 730                 //report progress
 731                 if(index % 100 != 0){   m->mothurOut(toString(index) + '\n');   }
 732
 733                 return index;
 734         }
 735         catch(exception& e) {
 736                 m->errorOut(e, "SeqErrorCommand", "driver");
 737                 exit(1);
 738         }
 739 }
 740 //***************************************************************************************************************
 741
 742 void SeqErrorCommand::getReferences(){
 743         try {
 744
 745                 ifstream referenceFile;
 746                 m->openInputFile(referenceFileName, referenceFile);
 747
 748                 int numAmbigSeqs = 0;
 749
 750                 int maxStartPos = 0;
 751                 int minEndPos = 100000;
 752
 753                 while(referenceFile){
 754                         Sequence currentSeq(referenceFile);
 755                         int numAmbigs = currentSeq.getAmbigBases();
 756                         if(numAmbigs > 0){      numAmbigSeqs++; }
 757
 758 //                      int startPos = currentSeq.getStartPos();
 759 //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 760 //
 761 //                      int endPos = currentSeq.getEndPos();
 762 //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 763                         referenceSeqs.push_back(currentSeq);
 764
 765                         m->gobble(referenceFile);
 766                 }
 767                 referenceFile.close();
 768                 numRefs = referenceSeqs.size();
 769
 770
 771                 for(int i=0;i<numRefs;i++){
 772                         referenceSeqs[i].padToPos(maxStartPos);
 773                         referenceSeqs[i].padFromPos(minEndPos);
 774                 }
 775
 776                 if(numAmbigSeqs != 0){
 777                         m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
 778                 }
 779
 780         }
 781         catch(exception& e) {
 782                 m->errorOut(e, "SeqErrorCommand", "getReferences");
 783                 exit(1);
 784         }
 785 }
 786
 787 //***************************************************************************************************************
 788
 789 Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
 790         try {
 791                 if(query.getAlignLength() != reference.getAlignLength()){
 792                         m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
 793                 }
 794                 int alignLength = query.getAlignLength();
 795
 796                 string q = query.getAligned();
 797                 string r = reference.getAligned();
 798
 799                 int started = 0;
 800                 Compare errors;
 801
 802                 for(int i=0;i<alignLength;i++){
 803                         if(r[i] != 'N' && q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){                  //      no missing data and no double gaps
 804                                 started = 1;
 805
 806                                 if(q[i] == 'A'){
 807                                         if(r[i] == 'A'){        errors.AA++;    errors.matches++;       errors.sequence += 'm'; }
 808                                         if(r[i] == 'T'){        errors.AT++;    errors.sequence += 's'; }
 809                                         if(r[i] == 'G'){        errors.AG++;    errors.sequence += 's'; }
 810                                         if(r[i] == 'C'){        errors.AC++;    errors.sequence += 's'; }
 811                                         if(r[i] == '-'){        errors.Ai++;    errors.sequence += 'i'; }
 812                                 }
 813                                 else if(q[i] == 'T'){
 814                                         if(r[i] == 'A'){        errors.TA++;    errors.sequence += 's'; }
 815                                         if(r[i] == 'T'){        errors.TT++;    errors.matches++;       errors.sequence += 'm'; }
 816                                         if(r[i] == 'G'){        errors.TG++;    errors.sequence += 's'; }
 817                                         if(r[i] == 'C'){        errors.TC++;    errors.sequence += 's'; }
 818                                         if(r[i] == '-'){        errors.Ti++;    errors.sequence += 'i'; }
 819                                 }
 820                                 else if(q[i] == 'G'){
 821                                         if(r[i] == 'A'){        errors.GA++;    errors.sequence += 's'; }
 822                                         if(r[i] == 'T'){        errors.GT++;    errors.sequence += 's'; }
 823                                         if(r[i] == 'G'){        errors.GG++;    errors.matches++;       errors.sequence += 'm'; }
 824                                         if(r[i] == 'C'){        errors.GC++;    errors.sequence += 's'; }
 825                                         if(r[i] == '-'){        errors.Gi++;    errors.sequence += 'i'; }
 826                                 }
 827                                 else if(q[i] == 'C'){
 828                                         if(r[i] == 'A'){        errors.CA++;    errors.sequence += 's'; }
 829                                         if(r[i] == 'T'){        errors.CT++;    errors.sequence += 's'; }
 830                                         if(r[i] == 'G'){        errors.CG++;    errors.sequence += 's'; }
 831                                         if(r[i] == 'C'){        errors.CC++;    errors.matches++;       errors.sequence += 'm'; }
 832                                         if(r[i] == '-'){        errors.Ci++;    errors.sequence += 'i'; }
 833                                 }
 834                                 else if(q[i] == 'N'){
 835                                         if(r[i] == 'A'){        errors.NA++;    errors.sequence += 'a'; }
 836                                         if(r[i] == 'T'){        errors.NT++;    errors.sequence += 'a'; }
 837                                         if(r[i] == 'G'){        errors.NG++;    errors.sequence += 'a'; }
 838                                         if(r[i] == 'C'){        errors.NC++;    errors.sequence += 'a'; }
 839                                         if(r[i] == '-'){        errors.Ni++;    errors.sequence += 'a'; }
 840                                 }
 841                                 else if(q[i] == '-' && r[i] != '-'){
 842                                         if(r[i] == 'A'){        errors.dA++;    errors.sequence += 'd'; }
 843                                         if(r[i] == 'T'){        errors.dT++;    errors.sequence += 'd'; }
 844                                         if(r[i] == 'G'){        errors.dG++;    errors.sequence += 'd'; }
 845                                         if(r[i] == 'C'){        errors.dC++;    errors.sequence += 'd'; }
 846                                 }
 847                                 errors.total++;
 848
 849                         }
 850                         else if(q[i] == '.' && r[i] != '.'){            //      reference extends beyond query
 851                                 if(started == 1){       break;  }
 852                         }
 853                         else if(q[i] != '.' && r[i] == '.'){            //      query extends beyond reference
 854                                 if(started == 1){       break;  }
 855                         }
 856                         else if(q[i] == '.' && r[i] == '.'){            //      both are missing data
 857                                 if(started == 1){       break;  }
 858                         }
 859
 860                 }
 861                 errors.mismatches = errors.total-errors.matches;
 862                 errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
 863                 errors.queryName = query.getName();
 864                 errors.refName = reference.getName();
 865
 866                 return errors;
 867         }
 868         catch(exception& e) {
 869                 m->errorOut(e, "SeqErrorCommand", "getErrors");
 870                 exit(1);
 871         }
 872 }
 873
 874 //***************************************************************************************************************
 875
 876 map<string, int> SeqErrorCommand::getWeights(){
 877         ifstream nameFile;
 878         m->openInputFile(namesFileName, nameFile);
 879
 880         string seqName;
 881         string redundantSeqs;
 882         map<string, int> nameCountMap;
 883
 884         while(nameFile){
 885                 nameFile >> seqName >> redundantSeqs;
 886                 nameCountMap[seqName] = m->getNumNames(redundantSeqs);
 887                 m->gobble(nameFile);
 888         }
 889         return nameCountMap;
 890 }
 891
 892
 893 //***************************************************************************************************************
 894
 895 void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
 896         try {
 897                 errorSummaryFile << "query\treference\tweight\t";
 898                 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";
 899                 errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
 900
 901                 errorSummaryFile << setprecision(6);
 902                 errorSummaryFile.setf(ios::fixed);
 903         }
 904         catch(exception& e) {
 905                 m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
 906                 exit(1);
 907         }
 908 }
 909
 910 //***************************************************************************************************************
 911
 912 void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs, ofstream& errorSummaryFile, ofstream& errorSeqFile){
 913         try {
 914
 915                 errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
 916                 errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
 917                 errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
 918                 errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
 919                 errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
 920                 errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
 921                 errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
 922                 errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
 923
 924                 errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t';                  //insertions
 925                 errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t';                  //deletions
 926                 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
 927                 errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t';       //ambiguities
 928                 errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
 929
 930                 errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
 931
 932                 int a=0;                int t=1;                int g=2;                int c=3;
 933                 int gap=4;              int n=5;
 934
 935                 if(numParentSeqs == 1 || ignoreChimeras == 0){
 936                         substitutionMatrix[a][a] += error.weight * error.AA;
 937                         substitutionMatrix[a][t] += error.weight * error.TA;
 938                         substitutionMatrix[a][g] += error.weight * error.GA;
 939                         substitutionMatrix[a][c] += error.weight * error.CA;
 940                         substitutionMatrix[a][gap] += error.weight * error.dA;
 941                         substitutionMatrix[a][n] += error.weight * error.NA;
 942
 943                         substitutionMatrix[t][a] += error.weight * error.AT;
 944                         substitutionMatrix[t][t] += error.weight * error.TT;
 945                         substitutionMatrix[t][g] += error.weight * error.GT;
 946                         substitutionMatrix[t][c] += error.weight * error.CT;
 947                         substitutionMatrix[t][gap] += error.weight * error.dT;
 948                         substitutionMatrix[t][n] += error.weight * error.NT;
 949
 950                         substitutionMatrix[g][a] += error.weight * error.AG;
 951                         substitutionMatrix[g][t] += error.weight * error.TG;
 952                         substitutionMatrix[g][g] += error.weight * error.GG;
 953                         substitutionMatrix[g][c] += error.weight * error.CG;
 954                         substitutionMatrix[g][gap] += error.weight * error.dG;
 955                         substitutionMatrix[g][n] += error.weight * error.NG;
 956
 957                         substitutionMatrix[c][a] += error.weight * error.AC;
 958                         substitutionMatrix[c][t] += error.weight * error.TC;
 959                         substitutionMatrix[c][g] += error.weight * error.GC;
 960                         substitutionMatrix[c][c] += error.weight * error.CC;
 961                         substitutionMatrix[c][gap] += error.weight * error.dC;
 962                         substitutionMatrix[c][n] += error.weight * error.NC;
 963
 964                         substitutionMatrix[gap][a] += error.weight * error.Ai;
 965                         substitutionMatrix[gap][t] += error.weight * error.Ti;
 966                         substitutionMatrix[gap][g] += error.weight * error.Gi;
 967                         substitutionMatrix[gap][c] += error.weight * error.Ci;
 968                         substitutionMatrix[gap][n] += error.weight * error.Ni;
 969                 }
 970         }
 971         catch(exception& e) {
 972                 m->errorOut(e, "SeqErrorCommand", "printErrorData");
 973                 exit(1);
 974         }
 975 }
 976
 977 //***************************************************************************************************************
 978
 979 void SeqErrorCommand::printSubMatrix(){
 980         try {
 981                 string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
 982                 ofstream subMatrixFile;
 983                 m->openOutputFile(subMatrixFileName, subMatrixFile);
 984                 outputNames.push_back(subMatrixFileName);  outputTypes["error.matrix"].push_back(subMatrixFileName);
 985                 vector<string> bases(6);
 986                 bases[0] = "A";
 987                 bases[1] = "T";
 988                 bases[2] = "G";
 989                 bases[3] = "C";
 990                 bases[4] = "Gap";
 991                 bases[5] = "N";
 992                 vector<int> refSums(5,1);
 993
 994                 for(int i=0;i<5;i++){
 995                         subMatrixFile << "\tr" << bases[i];
 996
 997                         for(int j=0;j<6;j++){
 998                                 refSums[i] += substitutionMatrix[i][j];
 999                         }
1000                 }
1001                 subMatrixFile << endl;
1002
1003                 for(int i=0;i<6;i++){
1004                         subMatrixFile << 'q' << bases[i];
1005                         for(int j=0;j<5;j++){
1006                                 subMatrixFile << '\t' << substitutionMatrix[j][i];
1007                         }
1008                         subMatrixFile << endl;
1009                 }
1010
1011                 subMatrixFile << "total";
1012                 for(int i=0;i<5;i++){
1013                         subMatrixFile << '\t' << refSums[i];
1014                 }
1015                 subMatrixFile << endl;
1016                 subMatrixFile.close();
1017         }
1018         catch(exception& e) {
1019                 m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
1020                 exit(1);
1021         }
1022 }
1023 //***************************************************************************************************************
1024
1025 void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
1026         try{
1027                 string errorForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.forward";
1028                 ofstream errorForwardFile;
1029                 m->openOutputFile(errorForwardFileName, errorForwardFile);
1030                 outputNames.push_back(errorForwardFileName);  outputTypes["error.forward"].push_back(errorForwardFileName);
1031
1032                 errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1033                 for(int i=0;i<maxLength;i++){
1034                         float match = (float)errorForward['m'][i];
1035                         float subst = (float)errorForward['s'][i];
1036                         float insert = (float)errorForward['i'][i];
1037                         float del = (float)errorForward['d'][i];
1038                         float amb = (float)errorForward['a'][i];
1039                         float total = match + subst + insert + del + amb;
1040                         if(total == 0){ break;  }
1041                         errorForwardFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1042                 }
1043                 errorForwardFile.close();
1044
1045                 string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
1046                 ofstream errorReverseFile;
1047                 m->openOutputFile(errorReverseFileName, errorReverseFile);
1048                 outputNames.push_back(errorReverseFileName);  outputTypes["error.reverse"].push_back(errorReverseFileName);
1049
1050                 errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1051                 for(int i=0;i<maxLength;i++){
1052                         float match = (float)errorReverse['m'][i];
1053                         float subst = (float)errorReverse['s'][i];
1054                         float insert = (float)errorReverse['i'][i];
1055                         float del = (float)errorReverse['d'][i];
1056                         float amb = (float)errorReverse['a'][i];
1057                         float total = match + subst + insert + del + amb;
1058                         if(total == 0){ break;  }
1059                         errorReverseFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1060                 }
1061                 errorReverseFile.close();
1062         }
1063         catch(exception& e) {
1064                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1065                 exit(1);
1066         }
1067 }
1068
1069 //***************************************************************************************************************
1070
1071 void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
1072         try{
1073
1074                 string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
1075                 ofstream errorQualityFile;
1076                 m->openOutputFile(errorQualityFileName, errorQualityFile);
1077                 outputNames.push_back(errorQualityFileName);  outputTypes["error.quality"].push_back(errorQualityFileName);
1078
1079                 errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
1080                 for(int i=0;i<41;i++){
1081                         errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
1082                 }
1083                 errorQualityFile.close();
1084         }
1085         catch(exception& e) {
1086                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1087                 exit(1);
1088         }
1089 }
1090
1091
1092 //***************************************************************************************************************
1093
1094 void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
1095
1096         try{
1097                 int numRows = 0;
1098                 int numColumns = qualForwardMap[0].size();
1099
1100                 for(int i=0;i<qualForwardMap.size();i++){
1101                         for(int j=0;j<numColumns;j++){
1102                                 if(qualForwardMap[i][j] != 0){
1103                                         if(numRows < i)         {       numRows = i+20;         }
1104                                 }
1105                         }
1106                 }
1107
1108                 string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
1109                 ofstream qualityForwardFile;
1110                 m->openOutputFile(qualityForwardFileName, qualityForwardFile);
1111                 outputNames.push_back(qualityForwardFileName);  outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
1112
1113                 for(int i=0;i<numColumns;i++){  qualityForwardFile << '\t' << i;        }       qualityForwardFile << endl;
1114
1115                 for(int i=0;i<numRows;i++){
1116                         qualityForwardFile << i+1;
1117                         for(int j=0;j<numColumns;j++){
1118                                 qualityForwardFile << '\t' << qualForwardMap[i][j];
1119                         }
1120
1121                         qualityForwardFile << endl;
1122                 }
1123                 qualityForwardFile.close();
1124
1125
1126                 string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
1127                 ofstream qualityReverseFile;
1128                 m->openOutputFile(qualityReverseFileName, qualityReverseFile);
1129                 outputNames.push_back(qualityReverseFileName);  outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
1130
1131                 for(int i=0;i<numColumns;i++){  qualityReverseFile << '\t' << i;        }       qualityReverseFile << endl;
1132                 for(int i=0;i<numRows;i++){
1133
1134                         qualityReverseFile << i+1;
1135                         for(int j=0;j<numColumns;j++){
1136                                 qualityReverseFile << '\t' << qualReverseMap[i][j];
1137                         }
1138                         qualityReverseFile << endl;
1139                 }
1140                 qualityReverseFile.close();
1141         }
1142         catch(exception& e) {
1143                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1144                 exit(1);
1145         }
1146
1147 }
1148 /**************************************************************************************************/
1149
1150 int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos, vector<unsigned long int>& rfileFilePos) {
1151         try {
1152 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1153                 //set file positions for fasta file
1154                 fastaFilePos = m->divideFile(filename, processors);
1155
1156                 if (qfilename == "") { return processors; }
1157
1158                 //get name of first sequence in each chunk
1159                 map<string, int> firstSeqNames;
1160                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
1161                         ifstream in;
1162                         m->openInputFile(filename, in);
1163                         in.seekg(fastaFilePos[i]);
1164
1165                         Sequence temp(in);
1166                         firstSeqNames[temp.getName()] = i;
1167
1168                         in.close();
1169                 }
1170
1171                 //make copy to use below
1172                 map<string, int> firstSeqNamesReport = firstSeqNames;
1173
1174                 //seach for filePos of each first name in the qfile and save in qfileFilePos
1175                 ifstream inQual;
1176                 m->openInputFile(qfilename, inQual);
1177
1178                 string input;
1179                 while(!inQual.eof()){
1180                         input = m->getline(inQual);
1181
1182                         if (input.length() != 0) {
1183                                 if(input[0] == '>'){ //this is a sequence name line
1184                                         istringstream nameStream(input);
1185
1186                                         string sname = "";  nameStream >> sname;
1187                                         sname = sname.substr(1);
1188
1189                                         map<string, int>::iterator it = firstSeqNames.find(sname);
1190
1191                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
1192                                                 unsigned long int pos = inQual.tellg();
1193                                                 qfileFilePos.push_back(pos - input.length() - 1);
1194                                                 firstSeqNames.erase(it);
1195                                         }
1196                                 }
1197                         }
1198
1199                         if (firstSeqNames.size() == 0) { break; }
1200                 }
1201                 inQual.close();
1202
1203                 if (firstSeqNames.size() != 0) {
1204                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
1205                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, aborting."); m->mothurOutEndLine();
1206                         }
1207                         m->control_pressed = true;
1208                         return processors;
1209                 }
1210
1211                 //get last file position of qfile
1212                 FILE * pFile;
1213                 unsigned long int size;
1214
1215                 //get num bytes in file
1216                 pFile = fopen (qfilename.c_str(),"rb");
1217                 if (pFile==NULL) perror ("Error opening file");
1218                 else{
1219                         fseek (pFile, 0, SEEK_END);
1220                         size=ftell (pFile);
1221                         fclose (pFile);
1222                 }
1223
1224                 qfileFilePos.push_back(size);
1225
1226                 //seach for filePos of each first name in the rfile and save in rfileFilePos
1227                 string junk;
1228                 ifstream inR;
1229                 m->openInputFile(rfilename, inR);
1230
1231                 //read column headers
1232                 for (int i = 0; i < 16; i++) {
1233                         if (!inR.eof()) {       inR >> junk;    }
1234                         else                    {       break;                  }
1235                 }
1236
1237                 while(!inR.eof()){
1238
1239                         if (m->control_pressed) { inR.close();  return processors; }
1240
1241                         input = m->getline(inR);
1242
1243                         if (input.length() != 0) {
1244
1245                                 istringstream nameStream(input);
1246                                 string sname = "";  nameStream >> sname;
1247
1248                                 map<string, int>::iterator it = firstSeqNamesReport.find(sname);
1249
1250                                 if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
1251                                         unsigned long int pos = inR.tellg();
1252                                         rfileFilePos.push_back(pos - input.length() - 1);
1253                                         firstSeqNamesReport.erase(it);
1254                                 }
1255                         }
1256
1257                         if (firstSeqNamesReport.size() == 0) { break; }
1258                         m->gobble(inR);
1259                 }
1260                 inR.close();
1261
1262                 if (firstSeqNamesReport.size() != 0) {
1263                         for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
1264                                 m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
1265                         }
1266                         m->control_pressed = true;
1267                         return processors;
1268                 }
1269
1270                 //get last file position of qfile
1271                 FILE * rFile;
1272                 unsigned long int sizeR;
1273
1274                 //get num bytes in file
1275                 rFile = fopen (rfilename.c_str(),"rb");
1276                 if (rFile==NULL) perror ("Error opening file");
1277                 else{
1278                         fseek (rFile, 0, SEEK_END);
1279                         sizeR=ftell (rFile);
1280                         fclose (rFile);
1281                 }
1282
1283                 rfileFilePos.push_back(sizeR);
1284
1285                 return processors;
1286
1287 #else
1288
1289                 fastaFilePos.push_back(0); qfileFilePos.push_back(0);
1290                 //get last file position of fastafile
1291                 FILE * pFile;
1292                 unsigned long int size;
1293
1294                 //get num bytes in file
1295                 pFile = fopen (filename.c_str(),"rb");
1296                 if (pFile==NULL) perror ("Error opening file");
1297                 else{
1298                         fseek (pFile, 0, SEEK_END);
1299                         size=ftell (pFile);
1300                         fclose (pFile);
1301                 }
1302                 fastaFilePos.push_back(size);
1303
1304                 //get last file position of fastafile
1305                 FILE * qFile;
1306
1307                 //get num bytes in file
1308                 qFile = fopen (qfilename.c_str(),"rb");
1309                 if (qFile==NULL) perror ("Error opening file");
1310                 else{
1311                         fseek (qFile, 0, SEEK_END);
1312                         size=ftell (qFile);
1313                         fclose (qFile);
1314                 }
1315                 qfileFilePos.push_back(size);
1316
1317                 return 1;
1318
1319 #endif
1320         }
1321         catch(exception& e) {
1322                 m->errorOut(e, "SeqErrorCommand", "setLines");
1323                 exit(1);
1324         }
1325 }
1326 //***************************************************************************************************************