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 string SeqErrorCommand::getOutputFileNameTag(string type, string inputName=""){
  67         try {
  68         string outputFileName = "";
  69                 map<string, vector<string> >::iterator it;
  70
  71         //is this a type this command creates
  72         it = outputTypes.find(type);
  73         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  74         else {
  75             if (type == "errorsummary")            {   outputFileName =  "error.summary";   }
  76             else if (type == "errorseq")            {   outputFileName =  "error.seq";   }
  77             else if (type == "errorquality")            {   outputFileName =  "error.quality";   }
  78             else if (type == "errorqualforward")            {   outputFileName =  "error.qual.forward";   }
  79             else if (type == "errorqualreverse")            {   outputFileName =  "error.qual.reverse";   }
  80             else if (type == "errorforward")            {   outputFileName =  "error.seq.forward";   }
  81             else if (type == "errorreverse")            {   outputFileName =  "error.seq.reverse";   }
  82             else if (type == "errorcount")            {   outputFileName =  "error.count";   }
  83             else if (type == "errormatrix")            {   outputFileName =  "error.matrix";   }
  84             else if (type == "errorchimera")            {   outputFileName =  "error.chimera";   }
  85             else if (type == "errorref-query")            {   outputFileName =  "error.ref-query";   }
  86             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  87         }
  88         return outputFileName;
  89         }
  90         catch(exception& e) {
  91                 m->errorOut(e, "SeqErrorCommand", "getOutputFileNameTag");
  92                 exit(1);
  93         }
  94 }
  95 //**********************************************************************************************************************
  96 SeqErrorCommand::SeqErrorCommand(){
  97         try {
  98                 abort = true; calledHelp = true;
  99                 setParameters();
 100                 vector<string> tempOutNames;
 101                 outputTypes["errorsummary"] = tempOutNames;
 102                 outputTypes["errorseq"] = tempOutNames;
 103                 outputTypes["errorquality"] = tempOutNames;
 104                 outputTypes["errorqualforward"] = tempOutNames;
 105                 outputTypes["errorqualreverse"] = tempOutNames;
 106                 outputTypes["errorforward"] = tempOutNames;
 107                 outputTypes["errorreverse"] = tempOutNames;
 108                 outputTypes["errorcount"] = tempOutNames;
 109                 outputTypes["errormatrix"] = tempOutNames;
 110         outputTypes["errorchimera"] = tempOutNames;
 111         outputTypes["errorref-query"] = tempOutNames;
 112         }
 113         catch(exception& e) {
 114                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 115                 exit(1);
 116         }
 117 }
 118 //***************************************************************************************************************
 119
 120 SeqErrorCommand::SeqErrorCommand(string option)  {
 121         try {
 122
 123                 abort = false; calledHelp = false;
 124                 rdb = ReferenceDB::getInstance();
 125
 126                 //allow user to run help
 127                 if(option == "help") { help(); abort = true; calledHelp = true; }
 128                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 129
 130                 else {
 131                         string temp;
 132                         vector<string> myArray = setParameters();
 133
 134                         OptionParser parser(option);
 135                         map<string,string> parameters = parser.getParameters();
 136
 137                         ValidParameters validParameter;
 138                         map<string,string>::iterator it;
 139
 140                         //check to make sure all parameters are valid for command
 141                         for (it = parameters.begin(); it != parameters.end(); it++) {
 142                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 143                         }
 144
 145                         //initialize outputTypes
 146                         vector<string> tempOutNames;
 147                         outputTypes["errorsummary"] = tempOutNames;
 148                         outputTypes["errorseq"] = tempOutNames;
 149                         outputTypes["errorquality"] = tempOutNames;
 150                         outputTypes["errorqualforward"] = tempOutNames;
 151                         outputTypes["errorqualreverse"] = tempOutNames;
 152                         outputTypes["errorforward"] = tempOutNames;
 153                         outputTypes["errorreverse"] = tempOutNames;
 154                         outputTypes["errorcount"] = tempOutNames;
 155                         outputTypes["errormatrix"] = tempOutNames;
 156             outputTypes["errorchimera"] = tempOutNames;
 157             outputTypes["errorref-query"] = tempOutNames;
 158
 159
 160                         //if the user changes the input directory command factory will send this info to us in the output parameter
 161                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 162                         if (inputDir == "not found"){   inputDir = "";          }
 163                         else {
 164                                 string path;
 165                                 it = parameters.find("fasta");
 166                                 //user has given a template file
 167                                 if(it != parameters.end()){
 168                                         path = m->hasPath(it->second);
 169                                         //if the user has not given a path then, add inputdir. else leave path alone.
 170                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 171                                 }
 172
 173                                 it = parameters.find("reference");
 174                                 //user has given a template file
 175                                 if(it != parameters.end()){
 176                                         path = m->hasPath(it->second);
 177                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 179                                 }
 180
 181                                 it = parameters.find("name");
 182                                 //user has given a names file
 183                                 if(it != parameters.end()){
 184                                         path = m->hasPath(it->second);
 185                                         //if the user has not given a path then, add inputdir. else leave path alone.
 186                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 187                                 }
 188
 189                                 it = parameters.find("qfile");
 190                                 //user has given a quality score file
 191                                 if(it != parameters.end()){
 192                                         path = m->hasPath(it->second);
 193                                         //if the user has not given a path then, add inputdir. else leave path alone.
 194                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 195                                 }
 196
 197                                 it = parameters.find("report");
 198                                 //user has given a alignment report file
 199                                 if(it != parameters.end()){
 200                                         path = m->hasPath(it->second);
 201                                         //if the user has not given a path then, add inputdir. else leave path alone.
 202                                         if (path == "") {       parameters["report"] = inputDir + it->second;           }
 203                                 }
 204
 205                         }
 206                         //check for required parameters
 207                         queryFileName = validParameter.validFile(parameters, "fasta", true);
 208                         if (queryFileName == "not found") {
 209                                 queryFileName = m->getFastaFile();
 210                                 if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 211                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 212                         }
 213                         else if (queryFileName == "not open") { queryFileName = ""; abort = true; }
 214                         else { m->setFastaFile(queryFileName); }
 215
 216                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 217                         if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
 218                         else if (referenceFileName == "not open") { abort = true; }
 219
 220                         //check for optional parameters
 221                         namesFileName = validParameter.validFile(parameters, "name", true);
 222                         if(namesFileName == "not found"){       namesFileName = "";     }
 223                         else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
 224                         else { m->setNameFile(namesFileName); }
 225
 226                         qualFileName = validParameter.validFile(parameters, "qfile", true);
 227                         if(qualFileName == "not found"){        qualFileName = "";      }
 228                         else if (qualFileName == "not open") { qualFileName = ""; abort = true; }
 229                         else { m->setQualFile(qualFileName); }
 230
 231                         reportFileName = validParameter.validFile(parameters, "report", true);
 232                         if(reportFileName == "not found"){      reportFileName = "";    }
 233                         else if (reportFileName == "not open") { reportFileName = ""; abort = true; }
 234
 235                         if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
 236                                 m->mothurOut("if you use either a qual file or a report file, you have to have both.");
 237                                 m->mothurOutEndLine();
 238                                 abort = true;
 239                         }
 240
 241                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 242                         if (outputDir == "not found"){
 243                                 outputDir = "";
 244                                 outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
 245                         }
 246
 247                         //check for optional parameter and set defaults
 248                         // ...at some point should added some additional type checking...
 249                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
 250                         m->mothurConvert(temp, threshold);
 251
 252                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 253                         save = m->isTrue(temp);
 254                         rdb->save = save;
 255                         if (save) { //clear out old references
 256                                 rdb->clearMemory();
 257                         }
 258
 259                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 260                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 261                         if (referenceFileName == "not found") {
 262                                 //check for saved reference sequences
 263                                 if (rdb->referenceSeqs.size() != 0) {
 264                                         referenceFileName = "saved";
 265                                 }else {
 266                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 267                                         m->mothurOutEndLine();
 268                                         abort = true;
 269                                 }
 270                         }else if (referenceFileName == "not open") { abort = true; }
 271                         else {  if (save) {     rdb->setSavedReference(referenceFileName);      }       }
 272
 273
 274                         temp = validParameter.validFile(parameters, "ignorechimeras", false);   if (temp == "not found") { temp = "T"; }
 275                         ignoreChimeras = m->isTrue(temp);
 276
 277                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 278                         m->setProcessors(temp);
 279                         m->mothurConvert(temp, processors);
 280
 281                         substitutionMatrix.resize(6);
 282                         for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
 283
 284                         if ((namesFileName == "") && (queryFileName != "")){
 285                                 vector<string> files; files.push_back(queryFileName);
 286                                 parser.getNameFile(files);
 287                         }
 288                 }
 289         }
 290         catch(exception& e) {
 291                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 292                 exit(1);
 293         }
 294 }
 295 //***************************************************************************************************************
 296
 297 int SeqErrorCommand::execute(){
 298         try{
 299                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 300
 301                 int start = time(NULL);
 302                 maxLength = 2000;
 303                 totalBases = 0;
 304                 totalMatches = 0;
 305
 306         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
 307                 string errorSummaryFileName = fileNameRoot + getOutputFileNameTag("errorsummary");
 308                 outputNames.push_back(errorSummaryFileName); outputTypes["errorsummary"].push_back(errorSummaryFileName);
 309
 310                 string errorSeqFileName = fileNameRoot + getOutputFileNameTag("errorseq");
 311                 outputNames.push_back(errorSeqFileName); outputTypes["errorseq"].push_back(errorSeqFileName);
 312
 313                 string errorChimeraFileName = fileNameRoot + getOutputFileNameTag("errorchimera");
 314                 outputNames.push_back(errorChimeraFileName); outputTypes["errorchimera"].push_back(errorChimeraFileName);
 315
 316                 getReferences();        //read in reference sequences - make sure there's no ambiguous bases
 317
 318                 if(namesFileName != ""){        weights = getWeights(); }
 319
 320                 vector<unsigned long long> fastaFilePos;
 321                 vector<unsigned long long> qFilePos;
 322                 vector<unsigned long long> reportFilePos;
 323
 324                 setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
 325
 326                 if (m->control_pressed) { return 0; }
 327
 328                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 329                         lines.push_back(linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 330                         if (qualFileName != "") {  qLines.push_back(linePair(qFilePos[i], qFilePos[(i+1)]));  }
 331                         if (reportFileName != "") {  rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)]));  }
 332                 }
 333                 if(qualFileName == "")  {       qLines = lines; rLines = lines; } //fills with duds
 334
 335                 int numSeqs = 0;
 336 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 337                 if(processors == 1){
 338                         numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 339                 }else{
 340                         numSeqs = createProcesses(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName);
 341                 }
 342 #else
 343                 numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 344 #endif
 345
 346                 if(qualFileName != "" && reportFileName != ""){
 347                         printErrorQuality(qScoreErrorMap);
 348                         printQualityFR(qualForwardMap, qualReverseMap);
 349                 }
 350
 351                 printErrorFRFile(errorForward, errorReverse);
 352
 353                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 354
 355                 string errorCountFileName = fileNameRoot + getOutputFileNameTag("errorcount");
 356                 ofstream errorCountFile;
 357                 m->openOutputFile(errorCountFileName, errorCountFile);
 358                 outputNames.push_back(errorCountFileName);  outputTypes["errorcount"].push_back(errorCountFileName);
 359                 m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n");
 360                 m->mothurOut("Errors\tSequences\n");
 361                 errorCountFile << "Errors\tSequences\n";
 362                 for(int i=0;i<misMatchCounts.size();i++){
 363                         m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
 364                         errorCountFile << i << '\t' << misMatchCounts[i] << endl;
 365                 }
 366                 errorCountFile.close();
 367
 368                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 369
 370                 printSubMatrix();
 371
 372                 string megAlignmentFileName = fileNameRoot + getOutputFileNameTag("errorref-query");
 373                 ofstream megAlignmentFile;
 374                 m->openOutputFile(megAlignmentFileName, megAlignmentFile);
 375                 outputNames.push_back(megAlignmentFileName);  outputTypes["errorref-query"].push_back(megAlignmentFileName);
 376
 377                 for(int i=0;i<numRefs;i++){
 378                         megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
 379                         megAlignmentFile << megaAlignVector[i] << endl;
 380                 }
 381
 382                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
 383                 m->mothurOutEndLine();
 384
 385                 m->mothurOutEndLine();
 386                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 387                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 388                 m->mothurOutEndLine();
 389
 390                 return 0;
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "SeqErrorCommand", "execute");
 394                 exit(1);
 395         }
 396 }
 397 //**********************************************************************************************************************
 398 int SeqErrorCommand::createProcesses(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName) {
 399         try {
 400                 int process = 1;
 401                 processIDS.clear();
 402                 map<char, vector<int> >::iterator it;
 403                 int num = 0;
 404 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 405
 406                 //loop through and create all the processes you want
 407                 while (process != processors) {
 408                         int pid = fork();
 409
 410                         if (pid > 0) {
 411                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 412                                 process++;
 413                         }else if (pid == 0){
 414
 415                                 num = driver(filename, qFileName, rFileName, summaryFileName + toString(getpid()) + ".temp", errorOutputFileName+ toString(getpid()) + ".temp", chimeraOutputFileName + toString(getpid()) + ".temp", lines[process], qLines[process], rLines[process]);
 416
 417                                 //pass groupCounts to parent
 418                                 ofstream out;
 419                                 string tempFile = filename + toString(getpid()) + ".info.temp";
 420                                 m->openOutputFile(tempFile, out);
 421
 422                                 //output totalBases and totalMatches
 423                                 out << num << '\t' << totalBases << '\t' << totalMatches << endl << endl;
 424
 425                                 //output substitutionMatrix
 426                                 for(int i = 0; i < substitutionMatrix.size(); i++) {
 427                                         for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 428                                                 out << substitutionMatrix[i][j] << '\t';
 429                                         }
 430                                         out << endl;
 431                                 }
 432                                 out << endl;
 433
 434                                 //output qScoreErrorMap
 435                                 for (it = qScoreErrorMap.begin(); it != qScoreErrorMap.end(); it++) {
 436                                         vector<int> thisScoreErrorMap = it->second;
 437                                         out << it->first << '\t';
 438                                         for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 439                                                 out << thisScoreErrorMap[i] << '\t';
 440                                         }
 441                                         out << endl;
 442                                 }
 443                                 out << endl;
 444
 445                                 //output qualForwardMap
 446                                 for(int i = 0; i < qualForwardMap.size(); i++) {
 447                                         for (int j = 0; j < qualForwardMap[i].size(); j++) {
 448                                                 out << qualForwardMap[i][j] << '\t';
 449                                         }
 450                                         out << endl;
 451                                 }
 452                                 out << endl;
 453
 454                                 //output qualReverseMap
 455                                 for(int i = 0; i < qualReverseMap.size(); i++) {
 456                                         for (int j = 0; j < qualReverseMap[i].size(); j++) {
 457                                                 out << qualReverseMap[i][j] << '\t';
 458                                         }
 459                                         out << endl;
 460                                 }
 461                                 out << endl;
 462
 463
 464                                 //output errorForward
 465                                 for (it = errorForward.begin(); it != errorForward.end(); it++) {
 466                                         vector<int> thisErrorForward = it->second;
 467                                         out << it->first << '\t';
 468                                         for (int i = 0; i < thisErrorForward.size(); i++) {
 469                                                 out << thisErrorForward[i] << '\t';
 470                                         }
 471                                         out << endl;
 472                                 }
 473                                 out << endl;
 474
 475                                 //output errorReverse
 476                                 for (it = errorReverse.begin(); it != errorReverse.end(); it++) {
 477                                         vector<int> thisErrorReverse = it->second;
 478                                         out << it->first << '\t';
 479                                         for (int i = 0; i < thisErrorReverse.size(); i++) {
 480                                                 out << thisErrorReverse[i] << '\t';
 481                                         }
 482                                         out << endl;
 483                                 }
 484                                 out << endl;
 485
 486                                 //output misMatchCounts
 487                                 out << misMatchCounts.size() << endl;
 488                                 for (int j = 0; j < misMatchCounts.size(); j++) {
 489                                         out << misMatchCounts[j] << '\t';
 490                                 }
 491                                 out << endl;
 492
 493
 494                                 //output megaAlignVector
 495                                 for (int j = 0; j < megaAlignVector.size(); j++) {
 496                                         out << megaAlignVector[j] << endl;
 497                                 }
 498                                 out << endl;
 499
 500                                 out.close();
 501
 502                                 exit(0);
 503                         }else {
 504                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 505                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 506                                 exit(0);
 507                         }
 508                 }
 509
 510                 //do my part
 511                 num = driver(filename, qFileName, rFileName, summaryFileName, errorOutputFileName, chimeraOutputFileName, lines[0], qLines[0], rLines[0]);
 512
 513                 //force parent to wait until all the processes are done
 514                 for (int i=0;i<processIDS.size();i++) {
 515                         int temp = processIDS[i];
 516                         wait(&temp);
 517                 }
 518
 519                 //append files
 520                 for(int i=0;i<processIDS.size();i++){
 521
 522                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 523
 524                         m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
 525                         m->mothurRemove((summaryFileName + toString(processIDS[i]) + ".temp"));
 526                         m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
 527                         m->mothurRemove((errorOutputFileName + toString(processIDS[i]) + ".temp"));
 528                         m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
 529                         m->mothurRemove((chimeraOutputFileName + toString(processIDS[i]) + ".temp"));
 530
 531                         ifstream in;
 532                         string tempFile =  filename + toString(processIDS[i]) + ".info.temp";
 533                         m->openInputFile(tempFile, in);
 534
 535                         //input totalBases and totalMatches
 536                         int tempBases, tempMatches, tempNumSeqs;
 537                         in >> tempNumSeqs >> tempBases >> tempMatches; m->gobble(in);
 538                         totalBases += tempBases; totalMatches += tempMatches; num += tempNumSeqs;
 539
 540                         //input substitutionMatrix
 541                         int tempNum;
 542                         for(int i = 0; i < substitutionMatrix.size(); i++) {
 543                                 for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 544                                         in >> tempNum; substitutionMatrix[i][j] += tempNum;
 545                                 }
 546                                 m->gobble(in);
 547                         }
 548                         m->gobble(in);
 549
 550                         //input qScoreErrorMap
 551                         char first;
 552                         for (int i = 0; i < qScoreErrorMap.size(); i++) {
 553                                 in >> first;
 554                                 vector<int> thisScoreErrorMap = qScoreErrorMap[first];
 555
 556                                 for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 557                                         in >> tempNum; thisScoreErrorMap[i] += tempNum;
 558                                 }
 559                                 qScoreErrorMap[first] = thisScoreErrorMap;
 560                                 m->gobble(in);
 561                         }
 562                         m->gobble(in);
 563
 564                         //input qualForwardMap
 565                         for(int i = 0; i < qualForwardMap.size(); i++) {
 566                                 for (int j = 0; j < qualForwardMap[i].size(); j++) {
 567                                         in >> tempNum; qualForwardMap[i][j] += tempNum;
 568                                 }
 569                                 m->gobble(in);
 570                         }
 571                         m->gobble(in);
 572
 573                         //input qualReverseMap
 574                         for(int i = 0; i < qualReverseMap.size(); i++) {
 575                                 for (int j = 0; j < qualReverseMap[i].size(); j++) {
 576                                         in >> tempNum; qualReverseMap[i][j] += tempNum;
 577                                 }
 578                                 m->gobble(in);
 579                         }
 580                         m->gobble(in);
 581
 582                         //input errorForward
 583                         for (int i = 0; i < errorForward.size(); i++) {
 584                                 in >> first;
 585                                 vector<int> thisErrorForward = errorForward[first];
 586
 587                                 for (int i = 0; i < thisErrorForward.size(); i++) {
 588                                         in >> tempNum; thisErrorForward[i] += tempNum;
 589                                 }
 590                                 errorForward[first] = thisErrorForward;
 591                                 m->gobble(in);
 592                         }
 593                         m->gobble(in);
 594
 595                         //input errorReverse
 596                         for (int i = 0; i < errorReverse.size(); i++) {
 597                                 in >> first;
 598                                 vector<int> thisErrorReverse = errorReverse[first];
 599
 600                                 for (int i = 0; i < thisErrorReverse.size(); i++) {
 601                                         in >> tempNum; thisErrorReverse[i] += tempNum;
 602                                 }
 603                                 errorReverse[first] = thisErrorReverse;
 604                                 m->gobble(in);
 605                         }
 606                         m->gobble(in);
 607
 608                         //input misMatchCounts
 609                         int misMatchSize;
 610                         in >> misMatchSize; m->gobble(in);
 611                         if (misMatchSize > misMatchCounts.size()) {     misMatchCounts.resize(misMatchSize, 0); }
 612                         for (int j = 0; j < misMatchCounts.size(); j++) {
 613                                 in >> tempNum; misMatchCounts[j] += tempNum;
 614                         }
 615                         m->gobble(in);
 616
 617                         //input megaAlignVector
 618                         string thisLine;
 619                         for (int j = 0; j < megaAlignVector.size(); j++) {
 620                                 thisLine = m->getline(in); m->gobble(in); megaAlignVector[j] += thisLine + '\n';
 621                         }
 622                         m->gobble(in);
 623
 624                         in.close(); m->mothurRemove(tempFile);
 625
 626                 }
 627 #endif
 628                 return num;
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "SeqErrorCommand", "createProcesses");
 632                 exit(1);
 633         }
 634 }
 635 //**********************************************************************************************************************
 636 int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
 637
 638         try {
 639                 ReportFile report;
 640                 QualityScores quality;
 641
 642                 misMatchCounts.resize(11, 0);
 643                 int maxMismatch = 0;
 644                 int numSeqs = 0;
 645
 646                 map<string, int>::iterator it;
 647                 qScoreErrorMap['m'].assign(41, 0);
 648                 qScoreErrorMap['s'].assign(41, 0);
 649                 qScoreErrorMap['i'].assign(41, 0);
 650                 qScoreErrorMap['a'].assign(41, 0);
 651
 652                 errorForward['m'].assign(maxLength,0);
 653                 errorForward['s'].assign(maxLength,0);
 654                 errorForward['i'].assign(maxLength,0);
 655                 errorForward['d'].assign(maxLength,0);
 656                 errorForward['a'].assign(maxLength,0);
 657
 658                 errorReverse['m'].assign(maxLength,0);
 659                 errorReverse['s'].assign(maxLength,0);
 660                 errorReverse['i'].assign(maxLength,0);
 661                 errorReverse['d'].assign(maxLength,0);
 662                 errorReverse['a'].assign(maxLength,0);
 663
 664                 //open inputfiles and go to beginning place for this processor
 665                 ifstream queryFile;
 666                 m->openInputFile(filename, queryFile);
 667                 queryFile.seekg(line.start);
 668
 669                 ifstream reportFile;
 670                 ifstream qualFile;
 671                 if(qFileName != "" && rFileName != ""){
 672                         m->openInputFile(qFileName, qualFile);
 673                         qualFile.seekg(qline.start);
 674
 675                         //gobble headers
 676                         if (rline.start == 0) {  report = ReportFile(reportFile, rFileName); }
 677                         else{
 678                                 m->openInputFile(rFileName, reportFile);
 679                                 reportFile.seekg(rline.start);
 680                         }
 681
 682                         qualForwardMap.resize(maxLength);
 683                         qualReverseMap.resize(maxLength);
 684                         for(int i=0;i<maxLength;i++){
 685                                 qualForwardMap[i].assign(41,0);
 686                                 qualReverseMap[i].assign(41,0);
 687                         }
 688                 }
 689
 690                 ofstream outChimeraReport;
 691                 m->openOutputFile(chimeraOutputFileName, outChimeraReport);
 692                 RefChimeraTest chimeraTest(referenceSeqs);
 693                 if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
 694
 695                 ofstream errorSummaryFile;
 696                 m->openOutputFile(summaryFileName, errorSummaryFile);
 697                 if (line.start == 0) { printErrorHeader(errorSummaryFile); }
 698
 699                 ofstream errorSeqFile;
 700                 m->openOutputFile(errorOutputFileName, errorSeqFile);
 701
 702                 megaAlignVector.resize(numRefs, "");
 703
 704                 int index = 0;
 705                 bool ignoreSeq = 0;
 706
 707                 bool moreSeqs = 1;
 708                 while (moreSeqs) {
 709
 710                         if (m->control_pressed) { queryFile.close(); if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); } outChimeraReport.close(); errorSummaryFile.close();errorSeqFile.close(); return 0; }
 711
 712                         Sequence query(queryFile);
 713
 714                         int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
 715                         int closestRefIndex = chimeraTest.getClosestRefIndex();
 716
 717                         if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
 718                         else                                                                                    {       ignoreSeq = 0;  }
 719
 720                         Compare minCompare;
 721             getErrors(query, referenceSeqs[closestRefIndex], minCompare);
 722
 723                         if(namesFileName != ""){
 724                                 it = weights.find(query.getName());
 725                                 minCompare.weight = it->second;
 726                         }
 727                         else{   minCompare.weight = 1;  }
 728
 729                         printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
 730
 731                         if(!ignoreSeq){
 732
 733                                 for(int i=0;i<minCompare.sequence.length();i++){
 734                                         char letter = minCompare.sequence[i];
 735
 736                                         if(letter != 'r'){
 737                                                 errorForward[letter][i] += minCompare.weight;
 738                                                 errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
 739                                         }
 740                                 }
 741                         }
 742
 743                         if(qualFileName != "" && reportFileName != ""){
 744                                 report = ReportFile(reportFile);
 745
 746                                 //                              int origLength = report.getQueryLength();
 747                                 int startBase = report.getQueryStart();
 748                                 int endBase = report.getQueryEnd();
 749
 750                                 quality = QualityScores(qualFile);
 751
 752                                 if(!ignoreSeq){
 753 //                                      cout << query.getName() << '\t';
 754
 755                                         quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
 756                                         quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
 757                                         quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
 758
 759 //                                      cout << endl;
 760                                 }
 761                         }
 762
 763                         if(minCompare.errorRate < threshold && !ignoreSeq){
 764                                 totalBases += (minCompare.total * minCompare.weight);
 765                                 totalMatches += minCompare.matches * minCompare.weight;
 766                                 if(minCompare.mismatches > maxMismatch){
 767                                         maxMismatch = minCompare.mismatches;
 768                                         misMatchCounts.resize(maxMismatch + 1, 0);
 769                                 }
 770                                 misMatchCounts[minCompare.mismatches] += minCompare.weight;
 771                                 numSeqs++;
 772
 773                                 megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
 774                         }
 775
 776                         index++;
 777
 778                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 779                                 unsigned long long pos = queryFile.tellg();
 780                                 if ((pos == -1) || (pos >= line.end)) { break; }
 781                         #else
 782                                 if (queryFile.eof()) { break; }
 783                         #endif
 784
 785                         if(index % 100 == 0){   m->mothurOut(toString(index));  m->mothurOutEndLine(); }
 786                 }
 787                 queryFile.close();
 788                 if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); }
 789                 errorSummaryFile.close();
 790                 errorSeqFile.close();
 791
 792                 //report progress
 793                 if(index % 100 != 0){   m->mothurOut(toString(index));  m->mothurOutEndLine(); }
 794
 795                 return index;
 796         }
 797         catch(exception& e) {
 798                 m->errorOut(e, "SeqErrorCommand", "driver");
 799                 exit(1);
 800         }
 801 }
 802 //***************************************************************************************************************
 803
 804 void SeqErrorCommand::getReferences(){
 805         try {
 806                 int numAmbigSeqs = 0;
 807
 808                 int maxStartPos = 0;
 809                 int minEndPos = 100000;
 810
 811                 if (referenceFileName == "saved") {
 812                         int start = time(NULL);
 813                         m->mothurOutEndLine();  m->mothurOut("Using sequences from " + rdb->getSavedReference() + " that are saved in memory.");        m->mothurOutEndLine();
 814
 815                         for (int i = 0; i < rdb->referenceSeqs.size(); i++) {
 816                                 int numAmbigs = rdb->referenceSeqs[i].getAmbigBases();
 817                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 818
 819                                 //                      int startPos = rdb->referenceSeqs[i].getStartPos();
 820                                 //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 821                                 //
 822                                 //                      int endPos = rdb->referenceSeqs[i].getEndPos();
 823                                 //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 824                                 if (rdb->referenceSeqs[i].getNumBases() == 0) {
 825                     m->mothurOut("[WARNING]: " + rdb->referenceSeqs[i].getName() + " is blank, ignoring.");m->mothurOutEndLine();
 826                 }else {
 827                     referenceSeqs.push_back(rdb->referenceSeqs[i]);
 828                 }
 829
 830                         }
 831                         referenceFileName = rdb->getSavedReference();
 832
 833                         m->mothurOut("It took " + toString(time(NULL) - start) + " to load " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 834
 835                 }else {
 836                         int start = time(NULL);
 837
 838                         ifstream referenceFile;
 839                         m->openInputFile(referenceFileName, referenceFile);
 840
 841                         while(referenceFile){
 842                                 Sequence currentSeq(referenceFile);
 843                                 int numAmbigs = currentSeq.getAmbigBases();
 844                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 845
 846         //                      int startPos = currentSeq.getStartPos();
 847         //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 848         //
 849         //                      int endPos = currentSeq.getEndPos();
 850         //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 851                 if (currentSeq.getNumBases() == 0) {
 852                     m->mothurOut("[WARNING]: " + currentSeq.getName() + " is blank, ignoring.");m->mothurOutEndLine();
 853                 }else {
 854                     referenceSeqs.push_back(currentSeq);
 855                     if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
 856                 }
 857
 858                                 m->gobble(referenceFile);
 859                         }
 860                         referenceFile.close();
 861
 862                         m->mothurOut("It took " + toString(time(NULL) - start) + " to read " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 863                 }
 864
 865                 numRefs = referenceSeqs.size();
 866
 867                 for(int i=0;i<numRefs;i++){
 868                         referenceSeqs[i].padToPos(maxStartPos);
 869                         referenceSeqs[i].padFromPos(minEndPos);
 870         }
 871
 872                 if(numAmbigSeqs != 0){
 873                         m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
 874                 }
 875
 876         }
 877         catch(exception& e) {
 878                 m->errorOut(e, "SeqErrorCommand", "getReferences");
 879                 exit(1);
 880         }
 881 }
 882
 883 //***************************************************************************************************************
 884
 885 int SeqErrorCommand::getErrors(Sequence query, Sequence reference, Compare& errors){
 886         try {
 887                 if(query.getAlignLength() != reference.getAlignLength()){
 888                         m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
 889                 }
 890                 int alignLength = query.getAlignLength();
 891
 892                 string q = query.getAligned();
 893                 string r = reference.getAligned();
 894
 895                 int started = 0;
 896                 //Compare errors;
 897
 898                 for(int i=0;i<alignLength;i++){
 899 //                      cout << r[i] << '\t' << q[i] << '\t';
 900                         if(q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){                 //      no missing data and no double gaps
 901                                 if(r[i] != 'N'){
 902                                         started = 1;
 903
 904                                         if(q[i] == 'A'){
 905                                                 if(r[i] == 'A'){        errors.AA++;    errors.matches++;       errors.sequence += 'm'; }
 906                                                 if(r[i] == 'T'){        errors.AT++;    errors.sequence += 's'; }
 907                                                 if(r[i] == 'G'){        errors.AG++;    errors.sequence += 's'; }
 908                                                 if(r[i] == 'C'){        errors.AC++;    errors.sequence += 's'; }
 909                                                 if(r[i] == '-'){        errors.Ai++;    errors.sequence += 'i'; }
 910                                         }
 911                                         else if(q[i] == 'T'){
 912                                                 if(r[i] == 'A'){        errors.TA++;    errors.sequence += 's'; }
 913                                                 if(r[i] == 'T'){        errors.TT++;    errors.matches++;       errors.sequence += 'm'; }
 914                                                 if(r[i] == 'G'){        errors.TG++;    errors.sequence += 's'; }
 915                                                 if(r[i] == 'C'){        errors.TC++;    errors.sequence += 's'; }
 916                                                 if(r[i] == '-'){        errors.Ti++;    errors.sequence += 'i'; }
 917                                         }
 918                                         else if(q[i] == 'G'){
 919                                                 if(r[i] == 'A'){        errors.GA++;    errors.sequence += 's'; }
 920                                                 if(r[i] == 'T'){        errors.GT++;    errors.sequence += 's'; }
 921                                                 if(r[i] == 'G'){        errors.GG++;    errors.matches++;       errors.sequence += 'm'; }
 922                                                 if(r[i] == 'C'){        errors.GC++;    errors.sequence += 's'; }
 923                                                 if(r[i] == '-'){        errors.Gi++;    errors.sequence += 'i'; }
 924                                         }
 925                                         else if(q[i] == 'C'){
 926                                                 if(r[i] == 'A'){        errors.CA++;    errors.sequence += 's'; }
 927                                                 if(r[i] == 'T'){        errors.CT++;    errors.sequence += 's'; }
 928                                                 if(r[i] == 'G'){        errors.CG++;    errors.sequence += 's'; }
 929                                                 if(r[i] == 'C'){        errors.CC++;    errors.matches++;       errors.sequence += 'm'; }
 930                                                 if(r[i] == '-'){        errors.Ci++;    errors.sequence += 'i'; }
 931                                         }
 932                                         else if(q[i] == 'N'){
 933                                                 if(r[i] == 'A'){        errors.NA++;    errors.sequence += 'a'; }
 934                                                 if(r[i] == 'T'){        errors.NT++;    errors.sequence += 'a'; }
 935                                                 if(r[i] == 'G'){        errors.NG++;    errors.sequence += 'a'; }
 936                                                 if(r[i] == 'C'){        errors.NC++;    errors.sequence += 'a'; }
 937                                                 if(r[i] == '-'){        errors.Ni++;    errors.sequence += 'a'; }
 938                                         }
 939                                         else if(q[i] == '-' && r[i] != '-'){
 940                                                 if(r[i] == 'A'){        errors.dA++;    errors.sequence += 'd'; }
 941                                                 if(r[i] == 'T'){        errors.dT++;    errors.sequence += 'd'; }
 942                                                 if(r[i] == 'G'){        errors.dG++;    errors.sequence += 'd'; }
 943                                                 if(r[i] == 'C'){        errors.dC++;    errors.sequence += 'd'; }
 944                                         }
 945                                         errors.total++;
 946                                 }
 947                                 else{
 948
 949                                         if(q[i] == '-'){
 950                                                 errors.sequence += 'd'; errors.total++;
 951                                         }
 952                                         else{
 953                                                 errors.sequence += 'r';
 954                                         }
 955                                 }
 956                         }
 957
 958                         else if(q[i] == '.' && r[i] != '.'){            //      reference extends beyond query
 959                                 if(started == 1){       break;  }
 960                         }
 961                         else if(q[i] != '.' && r[i] == '.'){            //      query extends beyond reference
 962                                 if(started == 1){       break;  }
 963                         }
 964                         else if(q[i] == '.' && r[i] == '.'){            //      both are missing data
 965                                 if(started == 1){       break;  }
 966                         }
 967 //                      cout << errors.sequence[errors.sequence.length()-1] << endl;
 968                 }
 969 //              cout << errors.sequence << endl;
 970                 errors.mismatches = errors.total-errors.matches;
 971                 errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
 972                 errors.queryName = query.getName();
 973                 errors.refName = reference.getName();
 974                 //return errors;
 975         return 0;
 976         }
 977         catch(exception& e) {
 978                 m->errorOut(e, "SeqErrorCommand", "getErrors");
 979                 exit(1);
 980         }
 981 }
 982
 983 //***************************************************************************************************************
 984
 985 map<string, int> SeqErrorCommand::getWeights(){
 986         ifstream nameFile;
 987         m->openInputFile(namesFileName, nameFile);
 988
 989         string seqName;
 990         string redundantSeqs;
 991         map<string, int> nameCountMap;
 992
 993         while(nameFile){
 994                 nameFile >> seqName >> redundantSeqs;
 995                 nameCountMap[seqName] = m->getNumNames(redundantSeqs);
 996                 m->gobble(nameFile);
 997         }
 998         return nameCountMap;
 999 }
1000
1001
1002 //***************************************************************************************************************
1003
1004 void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
1005         try {
1006                 errorSummaryFile << "query\treference\tweight\t";
1007                 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";
1008                 errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
1009
1010                 errorSummaryFile << setprecision(6);
1011                 errorSummaryFile.setf(ios::fixed);
1012         }
1013         catch(exception& e) {
1014                 m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
1015                 exit(1);
1016         }
1017 }
1018
1019 //***************************************************************************************************************
1020
1021 void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs, ofstream& errorSummaryFile, ofstream& errorSeqFile){
1022         try {
1023
1024                 errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
1025                 errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
1026                 errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
1027                 errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
1028                 errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
1029                 errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
1030                 errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
1031                 errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
1032
1033                 errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t';                  //insertions
1034                 errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t';                  //deletions
1035                 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
1036                 errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t';       //ambiguities
1037                 errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
1038
1039                 errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
1040
1041                 int a=0;                int t=1;                int g=2;                int c=3;
1042                 int gap=4;              int n=5;
1043
1044                 if(numParentSeqs == 1 || ignoreChimeras == 0){
1045                         substitutionMatrix[a][a] += error.weight * error.AA;
1046                         substitutionMatrix[a][t] += error.weight * error.TA;
1047                         substitutionMatrix[a][g] += error.weight * error.GA;
1048                         substitutionMatrix[a][c] += error.weight * error.CA;
1049                         substitutionMatrix[a][gap] += error.weight * error.dA;
1050                         substitutionMatrix[a][n] += error.weight * error.NA;
1051
1052                         substitutionMatrix[t][a] += error.weight * error.AT;
1053                         substitutionMatrix[t][t] += error.weight * error.TT;
1054                         substitutionMatrix[t][g] += error.weight * error.GT;
1055                         substitutionMatrix[t][c] += error.weight * error.CT;
1056                         substitutionMatrix[t][gap] += error.weight * error.dT;
1057                         substitutionMatrix[t][n] += error.weight * error.NT;
1058
1059                         substitutionMatrix[g][a] += error.weight * error.AG;
1060                         substitutionMatrix[g][t] += error.weight * error.TG;
1061                         substitutionMatrix[g][g] += error.weight * error.GG;
1062                         substitutionMatrix[g][c] += error.weight * error.CG;
1063                         substitutionMatrix[g][gap] += error.weight * error.dG;
1064                         substitutionMatrix[g][n] += error.weight * error.NG;
1065
1066                         substitutionMatrix[c][a] += error.weight * error.AC;
1067                         substitutionMatrix[c][t] += error.weight * error.TC;
1068                         substitutionMatrix[c][g] += error.weight * error.GC;
1069                         substitutionMatrix[c][c] += error.weight * error.CC;
1070                         substitutionMatrix[c][gap] += error.weight * error.dC;
1071                         substitutionMatrix[c][n] += error.weight * error.NC;
1072
1073                         substitutionMatrix[gap][a] += error.weight * error.Ai;
1074                         substitutionMatrix[gap][t] += error.weight * error.Ti;
1075                         substitutionMatrix[gap][g] += error.weight * error.Gi;
1076                         substitutionMatrix[gap][c] += error.weight * error.Ci;
1077                         substitutionMatrix[gap][n] += error.weight * error.Ni;
1078                 }
1079         }
1080         catch(exception& e) {
1081                 m->errorOut(e, "SeqErrorCommand", "printErrorData");
1082                 exit(1);
1083         }
1084 }
1085
1086 //***************************************************************************************************************
1087
1088 void SeqErrorCommand::printSubMatrix(){
1089         try {
1090         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
1091                 string subMatrixFileName = fileNameRoot + getOutputFileNameTag("errormatrix");
1092                 ofstream subMatrixFile;
1093                 m->openOutputFile(subMatrixFileName, subMatrixFile);
1094                 outputNames.push_back(subMatrixFileName);  outputTypes["errormatrix"].push_back(subMatrixFileName);
1095                 vector<string> bases(6);
1096                 bases[0] = "A";
1097                 bases[1] = "T";
1098                 bases[2] = "G";
1099                 bases[3] = "C";
1100                 bases[4] = "Gap";
1101                 bases[5] = "N";
1102                 vector<int> refSums(5,1);
1103
1104                 for(int i=0;i<5;i++){
1105                         subMatrixFile << "\tr" << bases[i];
1106
1107                         for(int j=0;j<6;j++){
1108                                 refSums[i] += substitutionMatrix[i][j];
1109                         }
1110                 }
1111                 subMatrixFile << endl;
1112
1113                 for(int i=0;i<6;i++){
1114                         subMatrixFile << 'q' << bases[i];
1115                         for(int j=0;j<5;j++){
1116                                 subMatrixFile << '\t' << substitutionMatrix[j][i];
1117                         }
1118                         subMatrixFile << endl;
1119                 }
1120
1121                 subMatrixFile << "total";
1122                 for(int i=0;i<5;i++){
1123                         subMatrixFile << '\t' << refSums[i];
1124                 }
1125                 subMatrixFile << endl;
1126                 subMatrixFile.close();
1127         }
1128         catch(exception& e) {
1129                 m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
1130                 exit(1);
1131         }
1132 }
1133 //***************************************************************************************************************
1134
1135 void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
1136         try{
1137         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
1138                 string errorForwardFileName = fileNameRoot + getOutputFileNameTag("errorforward");
1139                 ofstream errorForwardFile;
1140                 m->openOutputFile(errorForwardFileName, errorForwardFile);
1141                 outputNames.push_back(errorForwardFileName);  outputTypes["errorforward"].push_back(errorForwardFileName);
1142
1143                 errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1144                 for(int i=0;i<maxLength;i++){
1145                         float match = (float)errorForward['m'][i];
1146                         float subst = (float)errorForward['s'][i];
1147                         float insert = (float)errorForward['i'][i];
1148                         float del = (float)errorForward['d'][i];
1149                         float amb = (float)errorForward['a'][i];
1150                         float total = match + subst + insert + del + amb;
1151                         if(total == 0){ break;  }
1152                         errorForwardFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1153                 }
1154                 errorForwardFile.close();
1155
1156                 string errorReverseFileName = fileNameRoot + getOutputFileNameTag("errorreverse");
1157                 ofstream errorReverseFile;
1158                 m->openOutputFile(errorReverseFileName, errorReverseFile);
1159                 outputNames.push_back(errorReverseFileName);  outputTypes["errorreverse"].push_back(errorReverseFileName);
1160
1161                 errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1162                 for(int i=0;i<maxLength;i++){
1163                         float match = (float)errorReverse['m'][i];
1164                         float subst = (float)errorReverse['s'][i];
1165                         float insert = (float)errorReverse['i'][i];
1166                         float del = (float)errorReverse['d'][i];
1167                         float amb = (float)errorReverse['a'][i];
1168                         float total = match + subst + insert + del + amb;
1169                         if(total == 0){ break;  }
1170                         errorReverseFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1171                 }
1172                 errorReverseFile.close();
1173         }
1174         catch(exception& e) {
1175                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1176                 exit(1);
1177         }
1178 }
1179
1180 //***************************************************************************************************************
1181
1182 void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
1183         try{
1184         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
1185                 string errorQualityFileName = fileNameRoot + getOutputFileNameTag("errorquality");
1186                 ofstream errorQualityFile;
1187                 m->openOutputFile(errorQualityFileName, errorQualityFile);
1188                 outputNames.push_back(errorQualityFileName);  outputTypes["errorquality"].push_back(errorQualityFileName);
1189
1190                 errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
1191                 for(int i=0;i<41;i++){
1192                         errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
1193                 }
1194                 errorQualityFile.close();
1195         }
1196         catch(exception& e) {
1197                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1198                 exit(1);
1199         }
1200 }
1201
1202
1203 //***************************************************************************************************************
1204
1205 void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
1206
1207         try{
1208                 int numRows = 0;
1209                 int numColumns = qualForwardMap[0].size();
1210
1211                 for(int i=0;i<qualForwardMap.size();i++){
1212                         for(int j=0;j<numColumns;j++){
1213                                 if(qualForwardMap[i][j] != 0){
1214                                         if(numRows < i)         {       numRows = i+20;         }
1215                                 }
1216                         }
1217                 }
1218         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
1219                 string qualityForwardFileName = fileNameRoot + getOutputFileNameTag("errorqualforward");
1220                 ofstream qualityForwardFile;
1221                 m->openOutputFile(qualityForwardFileName, qualityForwardFile);
1222                 outputNames.push_back(qualityForwardFileName);  outputTypes["errorqualforward"].push_back(qualityForwardFileName);
1223
1224                 for(int i=0;i<numColumns;i++){  qualityForwardFile << '\t' << i;        }       qualityForwardFile << endl;
1225
1226                 for(int i=0;i<numRows;i++){
1227                         qualityForwardFile << i+1;
1228                         for(int j=0;j<numColumns;j++){
1229                                 qualityForwardFile << '\t' << qualForwardMap[i][j];
1230                         }
1231
1232                         qualityForwardFile << endl;
1233                 }
1234                 qualityForwardFile.close();
1235
1236
1237                 string qualityReverseFileName = fileNameRoot + getOutputFileNameTag("errorqualreverse");
1238                 ofstream qualityReverseFile;
1239                 m->openOutputFile(qualityReverseFileName, qualityReverseFile);
1240                 outputNames.push_back(qualityReverseFileName);  outputTypes["errorqualreverse"].push_back(qualityReverseFileName);
1241
1242                 for(int i=0;i<numColumns;i++){  qualityReverseFile << '\t' << i;        }       qualityReverseFile << endl;
1243                 for(int i=0;i<numRows;i++){
1244
1245                         qualityReverseFile << i+1;
1246                         for(int j=0;j<numColumns;j++){
1247                                 qualityReverseFile << '\t' << qualReverseMap[i][j];
1248                         }
1249                         qualityReverseFile << endl;
1250                 }
1251                 qualityReverseFile.close();
1252         }
1253         catch(exception& e) {
1254                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1255                 exit(1);
1256         }
1257
1258 }
1259 /**************************************************************************************************/
1260
1261 int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long long>& fastaFilePos, vector<unsigned long long>& qfileFilePos, vector<unsigned long long>& rfileFilePos) {
1262         try {
1263 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1264                 //set file positions for fasta file
1265                 fastaFilePos = m->divideFile(filename, processors);
1266
1267                 if (qfilename == "") { return processors; }
1268
1269                 //get name of first sequence in each chunk
1270                 map<string, int> firstSeqNames;
1271                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
1272                         ifstream in;
1273                         m->openInputFile(filename, in);
1274                         in.seekg(fastaFilePos[i]);
1275
1276                         Sequence temp(in);
1277                         firstSeqNames[temp.getName()] = i;
1278
1279                         in.close();
1280                 }
1281
1282                 //make copy to use below
1283                 map<string, int> firstSeqNamesReport = firstSeqNames;
1284
1285                 //seach for filePos of each first name in the qfile and save in qfileFilePos
1286                 ifstream inQual;
1287                 m->openInputFile(qfilename, inQual);
1288
1289                 string input;
1290                 while(!inQual.eof()){
1291                         input = m->getline(inQual);
1292
1293                         if (input.length() != 0) {
1294                                 if(input[0] == '>'){ //this is a sequence name line
1295                                         istringstream nameStream(input);
1296
1297                                         string sname = "";  nameStream >> sname;
1298                                         sname = sname.substr(1);
1299
1300                                         map<string, int>::iterator it = firstSeqNames.find(sname);
1301
1302                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
1303                                                 unsigned long long pos = inQual.tellg();
1304                                                 qfileFilePos.push_back(pos - input.length() - 1);
1305                                                 firstSeqNames.erase(it);
1306                                         }
1307                                 }
1308                         }
1309
1310                         if (firstSeqNames.size() == 0) { break; }
1311                 }
1312                 inQual.close();
1313
1314                 if (firstSeqNames.size() != 0) {
1315                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
1316                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, aborting."); m->mothurOutEndLine();
1317                         }
1318                         m->control_pressed = true;
1319                         return processors;
1320                 }
1321
1322                 //get last file position of qfile
1323                 FILE * pFile;
1324                 unsigned long long size;
1325
1326                 //get num bytes in file
1327                 pFile = fopen (qfilename.c_str(),"rb");
1328                 if (pFile==NULL) perror ("Error opening file");
1329                 else{
1330                         fseek (pFile, 0, SEEK_END);
1331                         size=ftell (pFile);
1332                         fclose (pFile);
1333                 }
1334
1335                 qfileFilePos.push_back(size);
1336
1337                 //seach for filePos of each first name in the rfile and save in rfileFilePos
1338                 string junk;
1339                 ifstream inR;
1340                 m->openInputFile(rfilename, inR);
1341
1342                 //read column headers
1343                 for (int i = 0; i < 16; i++) {
1344                         if (!inR.eof()) {       inR >> junk;    }
1345                         else                    {       break;                  }
1346                 }
1347
1348                 while(!inR.eof()){
1349
1350                         if (m->control_pressed) { inR.close();  return processors; }
1351
1352                         input = m->getline(inR);
1353
1354                         if (input.length() != 0) {
1355
1356                                 istringstream nameStream(input);
1357                                 string sname = "";  nameStream >> sname;
1358
1359                                 map<string, int>::iterator it = firstSeqNamesReport.find(sname);
1360
1361                                 if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
1362                                         unsigned long long pos = inR.tellg();
1363                                         rfileFilePos.push_back(pos - input.length() - 1);
1364                                         firstSeqNamesReport.erase(it);
1365                                 }
1366                         }
1367
1368                         if (firstSeqNamesReport.size() == 0) { break; }
1369                         m->gobble(inR);
1370                 }
1371                 inR.close();
1372
1373                 if (firstSeqNamesReport.size() != 0) {
1374                         for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
1375                                 m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
1376                         }
1377                         m->control_pressed = true;
1378                         return processors;
1379                 }
1380
1381                 //get last file position of qfile
1382                 FILE * rFile;
1383                 unsigned long long sizeR;
1384
1385                 //get num bytes in file
1386                 rFile = fopen (rfilename.c_str(),"rb");
1387                 if (rFile==NULL) perror ("Error opening file");
1388                 else{
1389                         fseek (rFile, 0, SEEK_END);
1390                         sizeR=ftell (rFile);
1391                         fclose (rFile);
1392                 }
1393
1394                 rfileFilePos.push_back(sizeR);
1395
1396                 return processors;
1397
1398 #else
1399
1400                 fastaFilePos.push_back(0); qfileFilePos.push_back(0);
1401                 //get last file position of fastafile
1402                 FILE * pFile;
1403                 unsigned long long size;
1404
1405                 //get num bytes in file
1406                 pFile = fopen (filename.c_str(),"rb");
1407                 if (pFile==NULL) perror ("Error opening file");
1408                 else{
1409                         fseek (pFile, 0, SEEK_END);
1410                         size=ftell (pFile);
1411                         fclose (pFile);
1412                 }
1413                 fastaFilePos.push_back(size);
1414
1415                 //get last file position of fastafile
1416                 FILE * qFile;
1417
1418                 //get num bytes in file
1419                 qFile = fopen (qfilename.c_str(),"rb");
1420                 if (qFile==NULL) perror ("Error opening file");
1421                 else{
1422                         fseek (qFile, 0, SEEK_END);
1423                         size=ftell (qFile);
1424                         fclose (qFile);
1425                 }
1426                 qfileFilePos.push_back(size);
1427
1428                 return 1;
1429
1430 #endif
1431         }
1432         catch(exception& e) {
1433                 m->errorOut(e, "SeqErrorCommand", "setLines");
1434                 exit(1);
1435         }
1436 }
1437 //***************************************************************************************************************