seqerrorcommand.cpp

   1 /*
   2  *  seqerrorcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 7/15/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "seqerrorcommand.h"
  11 #include "reportfile.h"
  12 #include "qualityscores.h"
  13 #include "refchimeratest.h"
  14 #include "filterseqscommand.h"
  15
  16
  17 //**********************************************************************************************************************
  18 vector<string> SeqErrorCommand::setParameters(){
  19         try {
  20                 CommandParameter pquery("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pquery);
  21                 CommandParameter preference("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(preference);
  22                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(pqfile);
  23                 CommandParameter preport("report", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(preport);
  24                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  25                 CommandParameter pignorechimeras("ignorechimeras", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pignorechimeras);
  26                 CommandParameter pthreshold("threshold", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pthreshold);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  28                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  29                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  30                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "SeqErrorCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string SeqErrorCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The seq.error command reads a query alignment file and a reference alignment file and creates .....\n";
  46                 helpString += "The fasta parameter...\n";
  47                 helpString += "The reference parameter...\n";
  48                 helpString += "The qfile parameter...\n";
  49                 helpString += "The report parameter...\n";
  50                 helpString += "The name parameter...\n";
  51                 helpString += "The ignorechimeras parameter...\n";
  52                 helpString += "The threshold parameter...\n";
  53                 helpString += "The processors parameter...\n";
  54                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  55                 helpString += "Example seq.error(...).\n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  57                 helpString += "For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n";
  58                 return helpString;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "SeqErrorCommand", "getHelpString");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 SeqErrorCommand::SeqErrorCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["error.summary"] = tempOutNames;
  72                 outputTypes["error.seq"] = tempOutNames;
  73                 outputTypes["error.quality"] = tempOutNames;
  74                 outputTypes["error.qual.forward"] = tempOutNames;
  75                 outputTypes["error.qual.reverse"] = tempOutNames;
  76                 outputTypes["error.forward"] = tempOutNames;
  77                 outputTypes["error.reverse"] = tempOutNames;
  78                 outputTypes["error.count"] = tempOutNames;
  79                 outputTypes["error.matrix"] = tempOutNames;
  80         }
  81         catch(exception& e) {
  82                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
  83                 exit(1);
  84         }
  85 }
  86 //***************************************************************************************************************
  87
  88 SeqErrorCommand::SeqErrorCommand(string option)  {
  89         try {
  90
  91                 abort = false; calledHelp = false;
  92                 rdb = ReferenceDB::getInstance();
  93
  94                 //allow user to run help
  95                 if(option == "help") { help(); abort = true; calledHelp = true; }
  96                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  97
  98                 else {
  99                         string temp;
 100                         vector<string> myArray = setParameters();
 101
 102                         OptionParser parser(option);
 103                         map<string,string> parameters = parser.getParameters();
 104
 105                         ValidParameters validParameter;
 106                         map<string,string>::iterator it;
 107
 108                         //check to make sure all parameters are valid for command
 109                         for (it = parameters.begin(); it != parameters.end(); it++) {
 110                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 111                         }
 112
 113                         //initialize outputTypes
 114                         vector<string> tempOutNames;
 115                         outputTypes["error.summary"] = tempOutNames;
 116                         outputTypes["error.seq"] = tempOutNames;
 117                         outputTypes["error.quality"] = tempOutNames;
 118                         outputTypes["error.qual.forward"] = tempOutNames;
 119                         outputTypes["error.qual.reverse"] = tempOutNames;
 120                         outputTypes["error.forward"] = tempOutNames;
 121                         outputTypes["error.reverse"] = tempOutNames;
 122                         outputTypes["error.count"] = tempOutNames;
 123                         outputTypes["error.matrix"] = tempOutNames;
 124
 125
 126                         //if the user changes the input directory command factory will send this info to us in the output parameter
 127                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 128                         if (inputDir == "not found"){   inputDir = "";          }
 129                         else {
 130                                 string path;
 131                                 it = parameters.find("fasta");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 137                                 }
 138
 139                                 it = parameters.find("reference");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 145                                 }
 146
 147                                 it = parameters.find("name");
 148                                 //user has given a names file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 153                                 }
 154
 155                                 it = parameters.find("qfile");
 156                                 //user has given a quality score file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("report");
 164                                 //user has given a alignment report file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["report"] = inputDir + it->second;           }
 169                                 }
 170
 171                         }
 172                         //check for required parameters
 173                         queryFileName = validParameter.validFile(parameters, "fasta", true);
 174                         if (queryFileName == "not found") {
 175                                 queryFileName = m->getFastaFile();
 176                                 if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 177                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 178                         }
 179                         else if (queryFileName == "not open") { abort = true; }
 180                         else { m->setFastaFile(queryFileName); }
 181
 182                         //check for optional parameters
 183                         namesFileName = validParameter.validFile(parameters, "name", true);
 184                         if(namesFileName == "not found"){       namesFileName = "";     }
 185                         else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
 186                         else { m->setNameFile(namesFileName); }
 187
 188                         qualFileName = validParameter.validFile(parameters, "qfile", true);
 189                         if(qualFileName == "not found"){        qualFileName = "";      }
 190                         else if (qualFileName == "not open") { qualFileName = ""; abort = true; }
 191                         else { m->setQualFile(qualFileName); }
 192
 193                         reportFileName = validParameter.validFile(parameters, "report", true);
 194                         if(reportFileName == "not found"){      reportFileName = "";    }
 195                         else if (reportFileName == "not open") { reportFileName = ""; abort = true; }
 196
 197                         if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
 198                                 m->mothurOut("if you use either a qual file or a report file, you have to have both.");
 199                                 m->mothurOutEndLine();
 200                                 abort = true;
 201                         }
 202
 203                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 204                         if (outputDir == "not found"){
 205                                 outputDir = "";
 206                                 outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
 207                         }
 208
 209                         //check for optional parameter and set defaults
 210                         // ...at some point should added some additional type checking...
 211                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
 212                         convert(temp, threshold);
 213
 214                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 215                         save = m->isTrue(temp);
 216                         rdb->save = save;
 217                         if (save) { //clear out old references
 218                                 rdb->clearMemory();
 219                         }
 220
 221                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 222                         referenceFileName = validParameter.validFile(parameters, "reference", true);
 223                         if (referenceFileName == "not found") {
 224                                 //check for saved reference sequences
 225                                 if (rdb->referenceSeqs.size() != 0) {
 226                                         referenceFileName = "saved";
 227                                 }else {
 228                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 229                                         m->mothurOutEndLine();
 230                                         abort = true;
 231                                 }
 232                         }else if (referenceFileName == "not open") { abort = true; }
 233                         else {  if (save) {     rdb->setSavedReference(referenceFileName);      }       }
 234
 235
 236                         temp = validParameter.validFile(parameters, "ignorechimeras", false);   if (temp == "not found") { temp = "T"; }
 237                         ignoreChimeras = m->isTrue(temp);
 238
 239                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 240                         m->setProcessors(temp);
 241                         convert(temp, processors);
 242
 243                         substitutionMatrix.resize(6);
 244                         for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
 245                 }
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
 249                 exit(1);
 250         }
 251 }
 252 //***************************************************************************************************************
 253
 254 int SeqErrorCommand::execute(){
 255         try{
 256                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 257
 258                 int start = time(NULL);
 259                 maxLength = 2000;
 260                 totalBases = 0;
 261                 totalMatches = 0;
 262
 263                 string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
 264                 outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
 265
 266                 string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
 267                 outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
 268
 269                 string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
 270                 outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
 271
 272                 getReferences();        //read in reference sequences - make sure there's no ambiguous bases
 273
 274                 if(namesFileName != ""){        weights = getWeights(); }
 275
 276                 vector<unsigned long int> fastaFilePos;
 277                 vector<unsigned long int> qFilePos;
 278                 vector<unsigned long int> reportFilePos;
 279
 280                 setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
 281
 282                 if (m->control_pressed) { return 0; }
 283
 284                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
 285                         lines.push_back(linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
 286                         if (qualFileName != "") {  qLines.push_back(linePair(qFilePos[i], qFilePos[(i+1)]));  }
 287                         if (reportFileName != "") {  rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)]));  }
 288                 }
 289                 if(qualFileName == "")  {       qLines = lines; rLines = lines; } //fills with duds
 290
 291                 int numSeqs = 0;
 292 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 293                 if(processors == 1){
 294                         numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 295                 }else{
 296                         numSeqs = createProcesses(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName);
 297                 }
 298 #else
 299                 numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
 300 #endif
 301
 302                 if(qualFileName != "" && reportFileName != ""){
 303                         printErrorQuality(qScoreErrorMap);
 304                         printQualityFR(qualForwardMap, qualReverseMap);
 305                 }
 306
 307                 printErrorFRFile(errorForward, errorReverse);
 308
 309                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 310
 311                 string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
 312                 ofstream errorCountFile;
 313                 m->openOutputFile(errorCountFileName, errorCountFile);
 314                 outputNames.push_back(errorCountFileName);  outputTypes["error.count"].push_back(errorCountFileName);
 315                 m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n");
 316                 m->mothurOut("Errors\tSequences\n");
 317                 errorCountFile << "Errors\tSequences\n";
 318                 for(int i=0;i<misMatchCounts.size();i++){
 319                         m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
 320                         errorCountFile << i << '\t' << misMatchCounts[i] << endl;
 321                 }
 322                 errorCountFile.close();
 323
 324                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 325
 326                 printSubMatrix();
 327
 328                 string megAlignmentFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.ref-query";
 329                 ofstream megAlignmentFile;
 330                 m->openOutputFile(megAlignmentFileName, megAlignmentFile);
 331                 outputNames.push_back(megAlignmentFileName);  outputTypes["error.ref-query"].push_back(megAlignmentFileName);
 332
 333                 for(int i=0;i<numRefs;i++){
 334                         megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
 335                         megAlignmentFile << megaAlignVector[i] << endl;
 336                 }
 337
 338                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
 339                 m->mothurOutEndLine();
 340
 341                 m->mothurOutEndLine();
 342                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 343                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 344                 m->mothurOutEndLine();
 345
 346                 return 0;
 347         }
 348         catch(exception& e) {
 349                 m->errorOut(e, "SeqErrorCommand", "execute");
 350                 exit(1);
 351         }
 352 }
 353 //**********************************************************************************************************************
 354 int SeqErrorCommand::createProcesses(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName) {
 355         try {
 356                 int process = 1;
 357                 processIDS.clear();
 358                 map<char, vector<int> >::iterator it;
 359                 int num = 0;
 360 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 361
 362                 //loop through and create all the processes you want
 363                 while (process != processors) {
 364                         int pid = fork();
 365
 366                         if (pid > 0) {
 367                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 368                                 process++;
 369                         }else if (pid == 0){
 370
 371                                 num = driver(filename, qFileName, rFileName, summaryFileName + toString(getpid()) + ".temp", errorOutputFileName+ toString(getpid()) + ".temp", chimeraOutputFileName + toString(getpid()) + ".temp", lines[process], qLines[process], rLines[process]);
 372
 373                                 //pass groupCounts to parent
 374                                 ofstream out;
 375                                 string tempFile = filename + toString(getpid()) + ".info.temp";
 376                                 m->openOutputFile(tempFile, out);
 377
 378                                 //output totalBases and totalMatches
 379                                 out << num << '\t' << totalBases << '\t' << totalMatches << endl << endl;
 380
 381                                 //output substitutionMatrix
 382                                 for(int i = 0; i < substitutionMatrix.size(); i++) {
 383                                         for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 384                                                 out << substitutionMatrix[i][j] << '\t';
 385                                         }
 386                                         out << endl;
 387                                 }
 388                                 out << endl;
 389
 390                                 //output qScoreErrorMap
 391                                 for (it = qScoreErrorMap.begin(); it != qScoreErrorMap.end(); it++) {
 392                                         vector<int> thisScoreErrorMap = it->second;
 393                                         out << it->first << '\t';
 394                                         for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 395                                                 out << thisScoreErrorMap[i] << '\t';
 396                                         }
 397                                         out << endl;
 398                                 }
 399                                 out << endl;
 400
 401                                 //output qualForwardMap
 402                                 for(int i = 0; i < qualForwardMap.size(); i++) {
 403                                         for (int j = 0; j < qualForwardMap[i].size(); j++) {
 404                                                 out << qualForwardMap[i][j] << '\t';
 405                                         }
 406                                         out << endl;
 407                                 }
 408                                 out << endl;
 409
 410                                 //output qualReverseMap
 411                                 for(int i = 0; i < qualReverseMap.size(); i++) {
 412                                         for (int j = 0; j < qualReverseMap[i].size(); j++) {
 413                                                 out << qualReverseMap[i][j] << '\t';
 414                                         }
 415                                         out << endl;
 416                                 }
 417                                 out << endl;
 418
 419
 420                                 //output errorForward
 421                                 for (it = errorForward.begin(); it != errorForward.end(); it++) {
 422                                         vector<int> thisErrorForward = it->second;
 423                                         out << it->first << '\t';
 424                                         for (int i = 0; i < thisErrorForward.size(); i++) {
 425                                                 out << thisErrorForward[i] << '\t';
 426                                         }
 427                                         out << endl;
 428                                 }
 429                                 out << endl;
 430
 431                                 //output errorReverse
 432                                 for (it = errorReverse.begin(); it != errorReverse.end(); it++) {
 433                                         vector<int> thisErrorReverse = it->second;
 434                                         out << it->first << '\t';
 435                                         for (int i = 0; i < thisErrorReverse.size(); i++) {
 436                                                 out << thisErrorReverse[i] << '\t';
 437                                         }
 438                                         out << endl;
 439                                 }
 440                                 out << endl;
 441
 442                                 //output misMatchCounts
 443                                 out << misMatchCounts.size() << endl;
 444                                 for (int j = 0; j < misMatchCounts.size(); j++) {
 445                                         out << misMatchCounts[j] << '\t';
 446                                 }
 447                                 out << endl;
 448
 449
 450                                 //output megaAlignVector
 451                                 for (int j = 0; j < megaAlignVector.size(); j++) {
 452                                         out << megaAlignVector[j] << endl;
 453                                 }
 454                                 out << endl;
 455
 456                                 out.close();
 457
 458                                 exit(0);
 459                         }else {
 460                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 461                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 462                                 exit(0);
 463                         }
 464                 }
 465
 466                 //do my part
 467                 num = driver(filename, qFileName, rFileName, summaryFileName, errorOutputFileName, chimeraOutputFileName, lines[0], qLines[0], rLines[0]);
 468
 469                 //force parent to wait until all the processes are done
 470                 for (int i=0;i<processIDS.size();i++) {
 471                         int temp = processIDS[i];
 472                         wait(&temp);
 473                 }
 474
 475                 //append files
 476                 for(int i=0;i<processIDS.size();i++){
 477
 478                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 479
 480                         m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
 481                         remove((summaryFileName + toString(processIDS[i]) + ".temp").c_str());
 482                         m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
 483                         remove((errorOutputFileName + toString(processIDS[i]) + ".temp").c_str());
 484                         m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
 485                         remove((chimeraOutputFileName + toString(processIDS[i]) + ".temp").c_str());
 486
 487                         ifstream in;
 488                         string tempFile =  filename + toString(processIDS[i]) + ".info.temp";
 489                         m->openInputFile(tempFile, in);
 490
 491                         //input totalBases and totalMatches
 492                         int tempBases, tempMatches, tempNumSeqs;
 493                         in >> tempNumSeqs >> tempBases >> tempMatches; m->gobble(in);
 494                         totalBases += tempBases; totalMatches += tempMatches; num += tempNumSeqs;
 495
 496                         //input substitutionMatrix
 497                         int tempNum;
 498                         for(int i = 0; i < substitutionMatrix.size(); i++) {
 499                                 for (int j = 0; j < substitutionMatrix[i].size(); j++) {
 500                                         in >> tempNum; substitutionMatrix[i][j] += tempNum;
 501                                 }
 502                                 m->gobble(in);
 503                         }
 504                         m->gobble(in);
 505
 506                         //input qScoreErrorMap
 507                         char first;
 508                         for (int i = 0; i < qScoreErrorMap.size(); i++) {
 509                                 in >> first;
 510                                 vector<int> thisScoreErrorMap = qScoreErrorMap[first];
 511
 512                                 for (int i = 0; i < thisScoreErrorMap.size(); i++) {
 513                                         in >> tempNum; thisScoreErrorMap[i] += tempNum;
 514                                 }
 515                                 qScoreErrorMap[first] = thisScoreErrorMap;
 516                                 m->gobble(in);
 517                         }
 518                         m->gobble(in);
 519
 520                         //input qualForwardMap
 521                         for(int i = 0; i < qualForwardMap.size(); i++) {
 522                                 for (int j = 0; j < qualForwardMap[i].size(); j++) {
 523                                         in >> tempNum; qualForwardMap[i][j] += tempNum;
 524                                 }
 525                                 m->gobble(in);
 526                         }
 527                         m->gobble(in);
 528
 529                         //input qualReverseMap
 530                         for(int i = 0; i < qualReverseMap.size(); i++) {
 531                                 for (int j = 0; j < qualReverseMap[i].size(); j++) {
 532                                         in >> tempNum; qualReverseMap[i][j] += tempNum;
 533                                 }
 534                                 m->gobble(in);
 535                         }
 536                         m->gobble(in);
 537
 538                         //input errorForward
 539                         for (int i = 0; i < errorForward.size(); i++) {
 540                                 in >> first;
 541                                 vector<int> thisErrorForward = errorForward[first];
 542
 543                                 for (int i = 0; i < thisErrorForward.size(); i++) {
 544                                         in >> tempNum; thisErrorForward[i] += tempNum;
 545                                 }
 546                                 errorForward[first] = thisErrorForward;
 547                                 m->gobble(in);
 548                         }
 549                         m->gobble(in);
 550
 551                         //input errorReverse
 552                         for (int i = 0; i < errorReverse.size(); i++) {
 553                                 in >> first;
 554                                 vector<int> thisErrorReverse = errorReverse[first];
 555
 556                                 for (int i = 0; i < thisErrorReverse.size(); i++) {
 557                                         in >> tempNum; thisErrorReverse[i] += tempNum;
 558                                 }
 559                                 errorReverse[first] = thisErrorReverse;
 560                                 m->gobble(in);
 561                         }
 562                         m->gobble(in);
 563
 564                         //input misMatchCounts
 565                         int misMatchSize;
 566                         in >> misMatchSize; m->gobble(in);
 567                         if (misMatchSize > misMatchCounts.size()) {     misMatchCounts.resize(misMatchSize, 0); }
 568                         for (int j = 0; j < misMatchCounts.size(); j++) {
 569                                 in >> tempNum; misMatchCounts[j] += tempNum;
 570                         }
 571                         m->gobble(in);
 572
 573                         //input megaAlignVector
 574                         string thisLine;
 575                         for (int j = 0; j < megaAlignVector.size(); j++) {
 576                                 thisLine = m->getline(in); m->gobble(in); megaAlignVector[j] += thisLine + '\n';
 577                         }
 578                         m->gobble(in);
 579
 580                         in.close(); remove(tempFile.c_str());
 581
 582                 }
 583 #endif
 584                 return num;
 585         }
 586         catch(exception& e) {
 587                 m->errorOut(e, "SeqErrorCommand", "createProcesses");
 588                 exit(1);
 589         }
 590 }
 591 //**********************************************************************************************************************
 592 int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
 593
 594         try {
 595                 ReportFile report;
 596                 QualityScores quality;
 597
 598                 misMatchCounts.resize(11, 0);
 599                 int maxMismatch = 0;
 600                 int numSeqs = 0;
 601
 602                 map<string, int>::iterator it;
 603                 qScoreErrorMap['m'].assign(41, 0);
 604                 qScoreErrorMap['s'].assign(41, 0);
 605                 qScoreErrorMap['i'].assign(41, 0);
 606                 qScoreErrorMap['a'].assign(41, 0);
 607
 608                 errorForward['m'].assign(maxLength,0);
 609                 errorForward['s'].assign(maxLength,0);
 610                 errorForward['i'].assign(maxLength,0);
 611                 errorForward['d'].assign(maxLength,0);
 612                 errorForward['a'].assign(maxLength,0);
 613
 614                 errorReverse['m'].assign(maxLength,0);
 615                 errorReverse['s'].assign(maxLength,0);
 616                 errorReverse['i'].assign(maxLength,0);
 617                 errorReverse['d'].assign(maxLength,0);
 618                 errorReverse['a'].assign(maxLength,0);
 619
 620                 //open inputfiles and go to beginning place for this processor
 621                 ifstream queryFile;
 622                 m->openInputFile(filename, queryFile);
 623                 queryFile.seekg(line.start);
 624
 625                 ifstream reportFile;
 626                 ifstream qualFile;
 627                 if(qFileName != "" && rFileName != ""){
 628                         m->openInputFile(qFileName, qualFile);
 629                         qualFile.seekg(qline.start);
 630
 631                         //gobble headers
 632                         if (rline.start == 0) {  report = ReportFile(reportFile, rFileName); }
 633                         else{
 634                                 m->openInputFile(rFileName, reportFile);
 635                                 reportFile.seekg(rline.start);
 636                         }
 637
 638                         qualForwardMap.resize(maxLength);
 639                         qualReverseMap.resize(maxLength);
 640                         for(int i=0;i<maxLength;i++){
 641                                 qualForwardMap[i].assign(41,0);
 642                                 qualReverseMap[i].assign(41,0);
 643                         }
 644                 }
 645
 646                 ofstream outChimeraReport;
 647                 m->openOutputFile(chimeraOutputFileName, outChimeraReport);
 648                 RefChimeraTest chimeraTest(referenceSeqs);
 649                 if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
 650
 651                 ofstream errorSummaryFile;
 652                 m->openOutputFile(summaryFileName, errorSummaryFile);
 653                 if (line.start == 0) { printErrorHeader(errorSummaryFile); }
 654
 655                 ofstream errorSeqFile;
 656                 m->openOutputFile(errorOutputFileName, errorSeqFile);
 657
 658                 megaAlignVector.resize(numRefs, "");
 659
 660                 int index = 0;
 661                 bool ignoreSeq = 0;
 662
 663                 bool moreSeqs = 1;
 664                 while (moreSeqs) {
 665
 666                         if (m->control_pressed) { queryFile.close(); if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); } outChimeraReport.close(); errorSummaryFile.close();errorSeqFile.close(); return 0; }
 667
 668                         Sequence query(queryFile);
 669
 670                         int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
 671                         int closestRefIndex = chimeraTest.getClosestRefIndex();
 672
 673                         if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
 674                         else                                                                                    {       ignoreSeq = 0;  }
 675
 676                         Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
 677
 678                         if(namesFileName != ""){
 679                                 it = weights.find(query.getName());
 680                                 minCompare.weight = it->second;
 681                         }
 682                         else{   minCompare.weight = 1;  }
 683
 684                         printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
 685
 686                         if(!ignoreSeq){
 687
 688                                 for(int i=0;i<minCompare.sequence.length();i++){
 689                                         char letter = minCompare.sequence[i];
 690
 691                                         if(letter != 'r'){
 692                                                 errorForward[letter][i] += minCompare.weight;
 693                                                 errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
 694                                         }
 695                                 }
 696                         }
 697
 698                         if(qualFileName != "" && reportFileName != ""){
 699                                 report = ReportFile(reportFile);
 700
 701                                 //                              int origLength = report.getQueryLength();
 702                                 int startBase = report.getQueryStart();
 703                                 int endBase = report.getQueryEnd();
 704
 705                                 quality = QualityScores(qualFile);
 706
 707                                 if(!ignoreSeq){
 708 //                                      cout << query.getName() << '\t';
 709
 710                                         quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
 711                                         quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
 712                                         quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
 713
 714 //                                      cout << endl;
 715                                 }
 716                         }
 717
 718                         if(minCompare.errorRate < threshold && !ignoreSeq){
 719                                 totalBases += (minCompare.total * minCompare.weight);
 720                                 totalMatches += minCompare.matches * minCompare.weight;
 721                                 if(minCompare.mismatches > maxMismatch){
 722                                         maxMismatch = minCompare.mismatches;
 723                                         misMatchCounts.resize(maxMismatch + 1, 0);
 724                                 }
 725                                 misMatchCounts[minCompare.mismatches] += minCompare.weight;
 726                                 numSeqs++;
 727
 728                                 megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
 729                         }
 730
 731                         index++;
 732
 733                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 734                                 unsigned long int pos = queryFile.tellg();
 735                                 if ((pos == -1) || (pos >= line.end)) { break; }
 736                         #else
 737                                 if (queryFile.eof()) { break; }
 738                         #endif
 739
 740                         if(index % 100 == 0){   m->mothurOut(toString(index) + '\n');   }
 741                 }
 742                 queryFile.close();
 743                 if(qFileName != "" && rFileName != ""){  reportFile.close(); qualFile.close(); }
 744                 errorSummaryFile.close();
 745                 errorSeqFile.close();
 746
 747                 //report progress
 748                 if(index % 100 != 0){   m->mothurOut(toString(index) + '\n');   }
 749
 750                 return index;
 751         }
 752         catch(exception& e) {
 753                 m->errorOut(e, "SeqErrorCommand", "driver");
 754                 exit(1);
 755         }
 756 }
 757 //***************************************************************************************************************
 758
 759 void SeqErrorCommand::getReferences(){
 760         try {
 761                 int numAmbigSeqs = 0;
 762
 763                 int maxStartPos = 0;
 764                 int minEndPos = 100000;
 765
 766                 if (referenceFileName == "saved") {
 767                         int start = time(NULL);
 768                         m->mothurOutEndLine();  m->mothurOut("Using sequences from " + rdb->getSavedReference() + " that are saved in memory.");        m->mothurOutEndLine();
 769
 770                         for (int i = 0; i < rdb->referenceSeqs.size(); i++) {
 771                                 int numAmbigs = rdb->referenceSeqs[i].getAmbigBases();
 772                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 773
 774                                 //                      int startPos = rdb->referenceSeqs[i].getStartPos();
 775                                 //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 776                                 //
 777                                 //                      int endPos = rdb->referenceSeqs[i].getEndPos();
 778                                 //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 779
 780                                 referenceSeqs.push_back(rdb->referenceSeqs[i]);
 781                         }
 782                         referenceFileName = rdb->getSavedReference();
 783
 784                         m->mothurOut("It took " + toString(time(NULL) - start) + " to load " + toString(rdb->referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 785
 786                 }else {
 787                         int start = time(NULL);
 788
 789                         ifstream referenceFile;
 790                         m->openInputFile(referenceFileName, referenceFile);
 791
 792                         while(referenceFile){
 793                                 Sequence currentSeq(referenceFile);
 794                                 int numAmbigs = currentSeq.getAmbigBases();
 795                                 if(numAmbigs > 0){      numAmbigSeqs++; }
 796
 797         //                      int startPos = currentSeq.getStartPos();
 798         //                      if(startPos > maxStartPos)      {       maxStartPos = startPos; }
 799         //
 800         //                      int endPos = currentSeq.getEndPos();
 801         //                      if(endPos < minEndPos)          {       minEndPos = endPos;             }
 802                                 referenceSeqs.push_back(currentSeq);
 803
 804                                 if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
 805
 806                                 m->gobble(referenceFile);
 807                         }
 808                         referenceFile.close();
 809
 810                         m->mothurOut("It took " + toString(time(NULL) - start) + " to read " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
 811                 }
 812
 813                 numRefs = referenceSeqs.size();
 814
 815                 for(int i=0;i<numRefs;i++){
 816                         referenceSeqs[i].padToPos(maxStartPos);
 817                         referenceSeqs[i].padFromPos(minEndPos);
 818                 }
 819
 820                 if(numAmbigSeqs != 0){
 821                         m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
 822                 }
 823
 824         }
 825         catch(exception& e) {
 826                 m->errorOut(e, "SeqErrorCommand", "getReferences");
 827                 exit(1);
 828         }
 829 }
 830
 831 //***************************************************************************************************************
 832
 833 Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
 834         try {
 835                 if(query.getAlignLength() != reference.getAlignLength()){
 836                         m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
 837                 }
 838                 int alignLength = query.getAlignLength();
 839
 840                 string q = query.getAligned();
 841                 string r = reference.getAligned();
 842
 843                 int started = 0;
 844                 Compare errors;
 845
 846                 for(int i=0;i<alignLength;i++){
 847 //                      cout << r[i] << '\t' << q[i] << '\t';
 848                         if(q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){                 //      no missing data and no double gaps
 849                                 if(r[i] != 'N'){
 850                                         started = 1;
 851
 852                                         if(q[i] == 'A'){
 853                                                 if(r[i] == 'A'){        errors.AA++;    errors.matches++;       errors.sequence += 'm'; }
 854                                                 if(r[i] == 'T'){        errors.AT++;    errors.sequence += 's'; }
 855                                                 if(r[i] == 'G'){        errors.AG++;    errors.sequence += 's'; }
 856                                                 if(r[i] == 'C'){        errors.AC++;    errors.sequence += 's'; }
 857                                                 if(r[i] == '-'){        errors.Ai++;    errors.sequence += 'i'; }
 858                                         }
 859                                         else if(q[i] == 'T'){
 860                                                 if(r[i] == 'A'){        errors.TA++;    errors.sequence += 's'; }
 861                                                 if(r[i] == 'T'){        errors.TT++;    errors.matches++;       errors.sequence += 'm'; }
 862                                                 if(r[i] == 'G'){        errors.TG++;    errors.sequence += 's'; }
 863                                                 if(r[i] == 'C'){        errors.TC++;    errors.sequence += 's'; }
 864                                                 if(r[i] == '-'){        errors.Ti++;    errors.sequence += 'i'; }
 865                                         }
 866                                         else if(q[i] == 'G'){
 867                                                 if(r[i] == 'A'){        errors.GA++;    errors.sequence += 's'; }
 868                                                 if(r[i] == 'T'){        errors.GT++;    errors.sequence += 's'; }
 869                                                 if(r[i] == 'G'){        errors.GG++;    errors.matches++;       errors.sequence += 'm'; }
 870                                                 if(r[i] == 'C'){        errors.GC++;    errors.sequence += 's'; }
 871                                                 if(r[i] == '-'){        errors.Gi++;    errors.sequence += 'i'; }
 872                                         }
 873                                         else if(q[i] == 'C'){
 874                                                 if(r[i] == 'A'){        errors.CA++;    errors.sequence += 's'; }
 875                                                 if(r[i] == 'T'){        errors.CT++;    errors.sequence += 's'; }
 876                                                 if(r[i] == 'G'){        errors.CG++;    errors.sequence += 's'; }
 877                                                 if(r[i] == 'C'){        errors.CC++;    errors.matches++;       errors.sequence += 'm'; }
 878                                                 if(r[i] == '-'){        errors.Ci++;    errors.sequence += 'i'; }
 879                                         }
 880                                         else if(q[i] == 'N'){
 881                                                 if(r[i] == 'A'){        errors.NA++;    errors.sequence += 'a'; }
 882                                                 if(r[i] == 'T'){        errors.NT++;    errors.sequence += 'a'; }
 883                                                 if(r[i] == 'G'){        errors.NG++;    errors.sequence += 'a'; }
 884                                                 if(r[i] == 'C'){        errors.NC++;    errors.sequence += 'a'; }
 885                                                 if(r[i] == '-'){        errors.Ni++;    errors.sequence += 'a'; }
 886                                         }
 887                                         else if(q[i] == '-' && r[i] != '-'){
 888                                                 if(r[i] == 'A'){        errors.dA++;    errors.sequence += 'd'; }
 889                                                 if(r[i] == 'T'){        errors.dT++;    errors.sequence += 'd'; }
 890                                                 if(r[i] == 'G'){        errors.dG++;    errors.sequence += 'd'; }
 891                                                 if(r[i] == 'C'){        errors.dC++;    errors.sequence += 'd'; }
 892                                         }
 893                                         errors.total++;
 894                                 }
 895                                 else{
 896
 897                                         if(q[i] == '-'){
 898                                                 errors.sequence += 'd'; errors.total++;
 899                                         }
 900                                         else{
 901                                                 errors.sequence += 'r';
 902                                         }
 903                                 }
 904                         }
 905
 906                         else if(q[i] == '.' && r[i] != '.'){            //      reference extends beyond query
 907                                 if(started == 1){       break;  }
 908                         }
 909                         else if(q[i] != '.' && r[i] == '.'){            //      query extends beyond reference
 910                                 if(started == 1){       break;  }
 911                         }
 912                         else if(q[i] == '.' && r[i] == '.'){            //      both are missing data
 913                                 if(started == 1){       break;  }
 914                         }
 915 //                      cout << errors.sequence[errors.sequence.length()-1] << endl;
 916                 }
 917 //              cout << errors.sequence << endl;
 918                 errors.mismatches = errors.total-errors.matches;
 919                 errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
 920                 errors.queryName = query.getName();
 921                 errors.refName = reference.getName();
 922
 923                 return errors;
 924         }
 925         catch(exception& e) {
 926                 m->errorOut(e, "SeqErrorCommand", "getErrors");
 927                 exit(1);
 928         }
 929 }
 930
 931 //***************************************************************************************************************
 932
 933 map<string, int> SeqErrorCommand::getWeights(){
 934         ifstream nameFile;
 935         m->openInputFile(namesFileName, nameFile);
 936
 937         string seqName;
 938         string redundantSeqs;
 939         map<string, int> nameCountMap;
 940
 941         while(nameFile){
 942                 nameFile >> seqName >> redundantSeqs;
 943                 nameCountMap[seqName] = m->getNumNames(redundantSeqs);
 944                 m->gobble(nameFile);
 945         }
 946         return nameCountMap;
 947 }
 948
 949
 950 //***************************************************************************************************************
 951
 952 void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
 953         try {
 954                 errorSummaryFile << "query\treference\tweight\t";
 955                 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";
 956                 errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
 957
 958                 errorSummaryFile << setprecision(6);
 959                 errorSummaryFile.setf(ios::fixed);
 960         }
 961         catch(exception& e) {
 962                 m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
 963                 exit(1);
 964         }
 965 }
 966
 967 //***************************************************************************************************************
 968
 969 void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs, ofstream& errorSummaryFile, ofstream& errorSeqFile){
 970         try {
 971
 972                 errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
 973                 errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
 974                 errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
 975                 errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
 976                 errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
 977                 errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
 978                 errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
 979                 errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
 980
 981                 errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t';                  //insertions
 982                 errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t';                  //deletions
 983                 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
 984                 errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t';       //ambiguities
 985                 errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
 986
 987                 errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
 988
 989                 int a=0;                int t=1;                int g=2;                int c=3;
 990                 int gap=4;              int n=5;
 991
 992                 if(numParentSeqs == 1 || ignoreChimeras == 0){
 993                         substitutionMatrix[a][a] += error.weight * error.AA;
 994                         substitutionMatrix[a][t] += error.weight * error.TA;
 995                         substitutionMatrix[a][g] += error.weight * error.GA;
 996                         substitutionMatrix[a][c] += error.weight * error.CA;
 997                         substitutionMatrix[a][gap] += error.weight * error.dA;
 998                         substitutionMatrix[a][n] += error.weight * error.NA;
 999
1000                         substitutionMatrix[t][a] += error.weight * error.AT;
1001                         substitutionMatrix[t][t] += error.weight * error.TT;
1002                         substitutionMatrix[t][g] += error.weight * error.GT;
1003                         substitutionMatrix[t][c] += error.weight * error.CT;
1004                         substitutionMatrix[t][gap] += error.weight * error.dT;
1005                         substitutionMatrix[t][n] += error.weight * error.NT;
1006
1007                         substitutionMatrix[g][a] += error.weight * error.AG;
1008                         substitutionMatrix[g][t] += error.weight * error.TG;
1009                         substitutionMatrix[g][g] += error.weight * error.GG;
1010                         substitutionMatrix[g][c] += error.weight * error.CG;
1011                         substitutionMatrix[g][gap] += error.weight * error.dG;
1012                         substitutionMatrix[g][n] += error.weight * error.NG;
1013
1014                         substitutionMatrix[c][a] += error.weight * error.AC;
1015                         substitutionMatrix[c][t] += error.weight * error.TC;
1016                         substitutionMatrix[c][g] += error.weight * error.GC;
1017                         substitutionMatrix[c][c] += error.weight * error.CC;
1018                         substitutionMatrix[c][gap] += error.weight * error.dC;
1019                         substitutionMatrix[c][n] += error.weight * error.NC;
1020
1021                         substitutionMatrix[gap][a] += error.weight * error.Ai;
1022                         substitutionMatrix[gap][t] += error.weight * error.Ti;
1023                         substitutionMatrix[gap][g] += error.weight * error.Gi;
1024                         substitutionMatrix[gap][c] += error.weight * error.Ci;
1025                         substitutionMatrix[gap][n] += error.weight * error.Ni;
1026                 }
1027         }
1028         catch(exception& e) {
1029                 m->errorOut(e, "SeqErrorCommand", "printErrorData");
1030                 exit(1);
1031         }
1032 }
1033
1034 //***************************************************************************************************************
1035
1036 void SeqErrorCommand::printSubMatrix(){
1037         try {
1038                 string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
1039                 ofstream subMatrixFile;
1040                 m->openOutputFile(subMatrixFileName, subMatrixFile);
1041                 outputNames.push_back(subMatrixFileName);  outputTypes["error.matrix"].push_back(subMatrixFileName);
1042                 vector<string> bases(6);
1043                 bases[0] = "A";
1044                 bases[1] = "T";
1045                 bases[2] = "G";
1046                 bases[3] = "C";
1047                 bases[4] = "Gap";
1048                 bases[5] = "N";
1049                 vector<int> refSums(5,1);
1050
1051                 for(int i=0;i<5;i++){
1052                         subMatrixFile << "\tr" << bases[i];
1053
1054                         for(int j=0;j<6;j++){
1055                                 refSums[i] += substitutionMatrix[i][j];
1056                         }
1057                 }
1058                 subMatrixFile << endl;
1059
1060                 for(int i=0;i<6;i++){
1061                         subMatrixFile << 'q' << bases[i];
1062                         for(int j=0;j<5;j++){
1063                                 subMatrixFile << '\t' << substitutionMatrix[j][i];
1064                         }
1065                         subMatrixFile << endl;
1066                 }
1067
1068                 subMatrixFile << "total";
1069                 for(int i=0;i<5;i++){
1070                         subMatrixFile << '\t' << refSums[i];
1071                 }
1072                 subMatrixFile << endl;
1073                 subMatrixFile.close();
1074         }
1075         catch(exception& e) {
1076                 m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
1077                 exit(1);
1078         }
1079 }
1080 //***************************************************************************************************************
1081
1082 void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
1083         try{
1084                 string errorForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.forward";
1085                 ofstream errorForwardFile;
1086                 m->openOutputFile(errorForwardFileName, errorForwardFile);
1087                 outputNames.push_back(errorForwardFileName);  outputTypes["error.forward"].push_back(errorForwardFileName);
1088
1089                 errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1090                 for(int i=0;i<maxLength;i++){
1091                         float match = (float)errorForward['m'][i];
1092                         float subst = (float)errorForward['s'][i];
1093                         float insert = (float)errorForward['i'][i];
1094                         float del = (float)errorForward['d'][i];
1095                         float amb = (float)errorForward['a'][i];
1096                         float total = match + subst + insert + del + amb;
1097                         if(total == 0){ break;  }
1098                         errorForwardFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1099                 }
1100                 errorForwardFile.close();
1101
1102                 string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
1103                 ofstream errorReverseFile;
1104                 m->openOutputFile(errorReverseFileName, errorReverseFile);
1105                 outputNames.push_back(errorReverseFileName);  outputTypes["error.reverse"].push_back(errorReverseFileName);
1106
1107                 errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
1108                 for(int i=0;i<maxLength;i++){
1109                         float match = (float)errorReverse['m'][i];
1110                         float subst = (float)errorReverse['s'][i];
1111                         float insert = (float)errorReverse['i'][i];
1112                         float del = (float)errorReverse['d'][i];
1113                         float amb = (float)errorReverse['a'][i];
1114                         float total = match + subst + insert + del + amb;
1115                         if(total == 0){ break;  }
1116                         errorReverseFile << i+1 << '\t' << total << '\t' << match/total  << '\t' << subst/total  << '\t' << insert/total  << '\t' << del/total  << '\t' << amb/total << endl;
1117                 }
1118                 errorReverseFile.close();
1119         }
1120         catch(exception& e) {
1121                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1122                 exit(1);
1123         }
1124 }
1125
1126 //***************************************************************************************************************
1127
1128 void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
1129         try{
1130
1131                 string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
1132                 ofstream errorQualityFile;
1133                 m->openOutputFile(errorQualityFileName, errorQualityFile);
1134                 outputNames.push_back(errorQualityFileName);  outputTypes["error.quality"].push_back(errorQualityFileName);
1135
1136                 errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
1137                 for(int i=0;i<41;i++){
1138                         errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
1139                 }
1140                 errorQualityFile.close();
1141         }
1142         catch(exception& e) {
1143                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1144                 exit(1);
1145         }
1146 }
1147
1148
1149 //***************************************************************************************************************
1150
1151 void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
1152
1153         try{
1154                 int numRows = 0;
1155                 int numColumns = qualForwardMap[0].size();
1156
1157                 for(int i=0;i<qualForwardMap.size();i++){
1158                         for(int j=0;j<numColumns;j++){
1159                                 if(qualForwardMap[i][j] != 0){
1160                                         if(numRows < i)         {       numRows = i+20;         }
1161                                 }
1162                         }
1163                 }
1164
1165                 string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
1166                 ofstream qualityForwardFile;
1167                 m->openOutputFile(qualityForwardFileName, qualityForwardFile);
1168                 outputNames.push_back(qualityForwardFileName);  outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
1169
1170                 for(int i=0;i<numColumns;i++){  qualityForwardFile << '\t' << i;        }       qualityForwardFile << endl;
1171
1172                 for(int i=0;i<numRows;i++){
1173                         qualityForwardFile << i+1;
1174                         for(int j=0;j<numColumns;j++){
1175                                 qualityForwardFile << '\t' << qualForwardMap[i][j];
1176                         }
1177
1178                         qualityForwardFile << endl;
1179                 }
1180                 qualityForwardFile.close();
1181
1182
1183                 string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
1184                 ofstream qualityReverseFile;
1185                 m->openOutputFile(qualityReverseFileName, qualityReverseFile);
1186                 outputNames.push_back(qualityReverseFileName);  outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
1187
1188                 for(int i=0;i<numColumns;i++){  qualityReverseFile << '\t' << i;        }       qualityReverseFile << endl;
1189                 for(int i=0;i<numRows;i++){
1190
1191                         qualityReverseFile << i+1;
1192                         for(int j=0;j<numColumns;j++){
1193                                 qualityReverseFile << '\t' << qualReverseMap[i][j];
1194                         }
1195                         qualityReverseFile << endl;
1196                 }
1197                 qualityReverseFile.close();
1198         }
1199         catch(exception& e) {
1200                 m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
1201                 exit(1);
1202         }
1203
1204 }
1205 /**************************************************************************************************/
1206
1207 int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos, vector<unsigned long int>& rfileFilePos) {
1208         try {
1209 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1210                 //set file positions for fasta file
1211                 fastaFilePos = m->divideFile(filename, processors);
1212
1213                 if (qfilename == "") { return processors; }
1214
1215                 //get name of first sequence in each chunk
1216                 map<string, int> firstSeqNames;
1217                 for (int i = 0; i < (fastaFilePos.size()-1); i++) {
1218                         ifstream in;
1219                         m->openInputFile(filename, in);
1220                         in.seekg(fastaFilePos[i]);
1221
1222                         Sequence temp(in);
1223                         firstSeqNames[temp.getName()] = i;
1224
1225                         in.close();
1226                 }
1227
1228                 //make copy to use below
1229                 map<string, int> firstSeqNamesReport = firstSeqNames;
1230
1231                 //seach for filePos of each first name in the qfile and save in qfileFilePos
1232                 ifstream inQual;
1233                 m->openInputFile(qfilename, inQual);
1234
1235                 string input;
1236                 while(!inQual.eof()){
1237                         input = m->getline(inQual);
1238
1239                         if (input.length() != 0) {
1240                                 if(input[0] == '>'){ //this is a sequence name line
1241                                         istringstream nameStream(input);
1242
1243                                         string sname = "";  nameStream >> sname;
1244                                         sname = sname.substr(1);
1245
1246                                         map<string, int>::iterator it = firstSeqNames.find(sname);
1247
1248                                         if(it != firstSeqNames.end()) { //this is the start of a new chunk
1249                                                 unsigned long int pos = inQual.tellg();
1250                                                 qfileFilePos.push_back(pos - input.length() - 1);
1251                                                 firstSeqNames.erase(it);
1252                                         }
1253                                 }
1254                         }
1255
1256                         if (firstSeqNames.size() == 0) { break; }
1257                 }
1258                 inQual.close();
1259
1260                 if (firstSeqNames.size() != 0) {
1261                         for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
1262                                 m->mothurOut(it->first + " is in your fasta file and not in your quality file, aborting."); m->mothurOutEndLine();
1263                         }
1264                         m->control_pressed = true;
1265                         return processors;
1266                 }
1267
1268                 //get last file position of qfile
1269                 FILE * pFile;
1270                 unsigned long int size;
1271
1272                 //get num bytes in file
1273                 pFile = fopen (qfilename.c_str(),"rb");
1274                 if (pFile==NULL) perror ("Error opening file");
1275                 else{
1276                         fseek (pFile, 0, SEEK_END);
1277                         size=ftell (pFile);
1278                         fclose (pFile);
1279                 }
1280
1281                 qfileFilePos.push_back(size);
1282
1283                 //seach for filePos of each first name in the rfile and save in rfileFilePos
1284                 string junk;
1285                 ifstream inR;
1286                 m->openInputFile(rfilename, inR);
1287
1288                 //read column headers
1289                 for (int i = 0; i < 16; i++) {
1290                         if (!inR.eof()) {       inR >> junk;    }
1291                         else                    {       break;                  }
1292                 }
1293
1294                 while(!inR.eof()){
1295
1296                         if (m->control_pressed) { inR.close();  return processors; }
1297
1298                         input = m->getline(inR);
1299
1300                         if (input.length() != 0) {
1301
1302                                 istringstream nameStream(input);
1303                                 string sname = "";  nameStream >> sname;
1304
1305                                 map<string, int>::iterator it = firstSeqNamesReport.find(sname);
1306
1307                                 if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
1308                                         unsigned long int pos = inR.tellg();
1309                                         rfileFilePos.push_back(pos - input.length() - 1);
1310                                         firstSeqNamesReport.erase(it);
1311                                 }
1312                         }
1313
1314                         if (firstSeqNamesReport.size() == 0) { break; }
1315                         m->gobble(inR);
1316                 }
1317                 inR.close();
1318
1319                 if (firstSeqNamesReport.size() != 0) {
1320                         for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
1321                                 m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
1322                         }
1323                         m->control_pressed = true;
1324                         return processors;
1325                 }
1326
1327                 //get last file position of qfile
1328                 FILE * rFile;
1329                 unsigned long int sizeR;
1330
1331                 //get num bytes in file
1332                 rFile = fopen (rfilename.c_str(),"rb");
1333                 if (rFile==NULL) perror ("Error opening file");
1334                 else{
1335                         fseek (rFile, 0, SEEK_END);
1336                         sizeR=ftell (rFile);
1337                         fclose (rFile);
1338                 }
1339
1340                 rfileFilePos.push_back(sizeR);
1341
1342                 return processors;
1343
1344 #else
1345
1346                 fastaFilePos.push_back(0); qfileFilePos.push_back(0);
1347                 //get last file position of fastafile
1348                 FILE * pFile;
1349                 unsigned long int size;
1350
1351                 //get num bytes in file
1352                 pFile = fopen (filename.c_str(),"rb");
1353                 if (pFile==NULL) perror ("Error opening file");
1354                 else{
1355                         fseek (pFile, 0, SEEK_END);
1356                         size=ftell (pFile);
1357                         fclose (pFile);
1358                 }
1359                 fastaFilePos.push_back(size);
1360
1361                 //get last file position of fastafile
1362                 FILE * qFile;
1363
1364                 //get num bytes in file
1365                 qFile = fopen (qfilename.c_str(),"rb");
1366                 if (qFile==NULL) perror ("Error opening file");
1367                 else{
1368                         fseek (qFile, 0, SEEK_END);
1369                         size=ftell (qFile);
1370                         fclose (qFile);
1371                 }
1372                 qfileFilePos.push_back(size);
1373
1374                 return 1;
1375
1376 #endif
1377         }
1378         catch(exception& e) {
1379                 m->errorOut(e, "SeqErrorCommand", "setLines");
1380                 exit(1);
1381         }
1382 }
1383 //***************************************************************************************************************