trimflowscommand.cpp

   1 /*
   2  *  trimflowscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 12/22/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "trimflowscommand.h"
  11 #include "needlemanoverlap.hpp"
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> TrimFlowsCommand::setParameters(){
  16         try {
  17                 CommandParameter pflow("flow", "InputTypes", "", "", "none", "none", "none","flow",false,true,true); parameters.push_back(pflow);
  18                 CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none","",false,false,true); parameters.push_back(poligos);
  19                 CommandParameter pmaxhomop("maxhomop", "Number", "", "9", "", "", "","",false,false); parameters.push_back(pmaxhomop);
  20                 CommandParameter pmaxflows("maxflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pmaxflows);
  21                 CommandParameter pminflows("minflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pminflows);
  22                 CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(ppdiffs);
  23                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(pbdiffs);
  24         CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pldiffs);
  25                 CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psdiffs);
  26         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  28                 CommandParameter psignal("signal", "Number", "", "0.50", "", "", "","",false,false); parameters.push_back(psignal);
  29                 CommandParameter pnoise("noise", "Number", "", "0.70", "", "", "","",false,false); parameters.push_back(pnoise);
  30                 CommandParameter pallfiles("allfiles", "Boolean", "", "t", "", "", "","",false,false); parameters.push_back(pallfiles);
  31                 CommandParameter porder("order", "String", "", "TACG", "", "", "","",false,false); parameters.push_back(porder);
  32                 CommandParameter pfasta("fasta", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pfasta);
  33                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  34                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  35
  36                 vector<string> myArray;
  37                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  38                 return myArray;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "TrimFlowsCommand", "setParameters");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 string TrimFlowsCommand::getHelpString(){
  47         try {
  48                 string helpString = "";
  49                 helpString += "The trim.flows command reads a flowgram file and creates .....\n";
  50                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  51                 helpString += "For more details please check out the wiki http://www.mothur.org/wiki/Trim.flows.\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "TrimFlowsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 string TrimFlowsCommand::getOutputPattern(string type) {
  61     try {
  62         string pattern = "";
  63
  64         if (type == "flow") {  pattern = "[filename],[tag],flow"; }
  65         else if (type == "fasta") {  pattern = "[filename],flow.fasta"; }
  66         else if (type == "file") {  pattern = "[filename],flow.files"; }
  67         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  68
  69         return pattern;
  70     }
  71     catch(exception& e) {
  72         m->errorOut(e, "TrimFlowsCommand", "getOutputPattern");
  73         exit(1);
  74     }
  75 }
  76 //**********************************************************************************************************************
  77
  78 TrimFlowsCommand::TrimFlowsCommand(){
  79         try {
  80                 abort = true; calledHelp = true;
  81                 setParameters();
  82                 vector<string> tempOutNames;
  83                 outputTypes["flow"] = tempOutNames;
  84                 outputTypes["fasta"] = tempOutNames;
  85         outputTypes["file"] = tempOutNames;
  86         }
  87         catch(exception& e) {
  88                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
  89                 exit(1);
  90         }
  91 }
  92 //**********************************************************************************************************************
  93
  94 TrimFlowsCommand::TrimFlowsCommand(string option)  {
  95         try {
  96
  97                 abort = false; calledHelp = false;
  98                 comboStarts = 0;
  99
 100                 //allow user to run help
 101                 if(option == "help") { help(); abort = true; calledHelp = true; }
 102                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 103
 104                 else {
 105
 106                         vector<string> myArray = setParameters();
 107
 108                         OptionParser parser(option);
 109                         map<string,string> parameters = parser.getParameters();
 110
 111                         ValidParameters validParameter;
 112                         map<string,string>::iterator it;
 113
 114                         //check to make sure all parameters are valid for command
 115                         for (it = parameters.begin(); it != parameters.end(); it++) {
 116                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 117                         }
 118
 119                         //initialize outputTypes
 120                         vector<string> tempOutNames;
 121                         outputTypes["flow"] = tempOutNames;
 122                         outputTypes["fasta"] = tempOutNames;
 123             outputTypes["file"] = tempOutNames;
 124
 125                         //if the user changes the input directory command factory will send this info to us in the output parameter
 126                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 127                         if (inputDir == "not found"){   inputDir = "";          }
 128                         else {
 129                                 string path;
 130                                 it = parameters.find("flow");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["flow"] = inputDir + it->second;             }
 136                                 }
 137
 138                                 it = parameters.find("oligos");
 139                                 //user has given a template file
 140                                 if(it != parameters.end()){
 141                                         path = m->hasPath(it->second);
 142                                         //if the user has not given a path then, add inputdir. else leave path alone.
 143                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 144                                 }
 145
 146                         }
 147
 148
 149                         //check for required parameters
 150                         flowFileName = validParameter.validFile(parameters, "flow", true);
 151                         if (flowFileName == "not found") {
 152                                 flowFileName = m->getFlowFile();
 153                                 if (flowFileName != "") {  m->mothurOut("Using " + flowFileName + " as input file for the flow parameter."); m->mothurOutEndLine(); }
 154                                 else {
 155                                         m->mothurOut("No valid current flow file. You must provide a flow file."); m->mothurOutEndLine();
 156                                         abort = true;
 157                                 }
 158                         }else if (flowFileName == "not open") { flowFileName = ""; abort = true; }
 159
 160                         //if the user changes the output directory command factory will send this info to us in the output parameter
 161                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 162                                 outputDir = "";
 163                                 outputDir += m->hasPath(flowFileName); //if user entered a file with a path then preserve it
 164                         }
 165
 166
 167                         //check for optional parameter and set defaults
 168                         // ...at some point should added some additional type checking...
 169
 170                         string temp;
 171                         temp = validParameter.validFile(parameters, "minflows", false); if (temp == "not found") { temp = "450"; }
 172                         m->mothurConvert(temp, minFlows);
 173
 174                         temp = validParameter.validFile(parameters, "maxflows", false); if (temp == "not found") { temp = "450"; }
 175                         m->mothurConvert(temp, maxFlows);
 176
 177
 178                         temp = validParameter.validFile(parameters, "oligos", true);
 179                         if (temp == "not found")        {       oligoFileName = "";             }
 180                         else if(temp == "not open")     {       abort = true;                   }
 181                         else                                            {       oligoFileName = temp;   m->setOligosFile(oligoFileName); }
 182
 183                         temp = validParameter.validFile(parameters, "fasta", false);            if (temp == "not found"){       fasta = 0;              }
 184                         else if(m->isTrue(temp))        {       fasta = 1;      }
 185
 186                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found"){       temp = "9";             }
 187                         m->mothurConvert(temp, maxHomoP);
 188
 189                         temp = validParameter.validFile(parameters, "signal", false);           if (temp == "not found"){       temp = "0.50";  }
 190                         m->mothurConvert(temp, signal);
 191
 192                         temp = validParameter.validFile(parameters, "noise", false);            if (temp == "not found"){       temp = "0.70";  }
 193                         m->mothurConvert(temp, noise);
 194
 195                         temp = validParameter.validFile(parameters, "bdiffs", false);           if (temp == "not found"){       temp = "0";             }
 196                         m->mothurConvert(temp, bdiffs);
 197
 198                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found"){       temp = "0";             }
 199                         m->mothurConvert(temp, pdiffs);
 200
 201             temp = validParameter.validFile(parameters, "ldiffs", false);               if (temp == "not found") { temp = "0"; }
 202                         m->mothurConvert(temp, ldiffs);
 203
 204             temp = validParameter.validFile(parameters, "sdiffs", false);               if (temp == "not found") { temp = "0"; }
 205                         m->mothurConvert(temp, sdiffs);
 206
 207                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs;  temp = toString(tempTotal); }
 208                         m->mothurConvert(temp, tdiffs);
 209
 210                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs + ldiffs + sdiffs;     }
 211
 212
 213                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 214                         m->setProcessors(temp);
 215                         m->mothurConvert(temp, processors);
 216
 217                         flowOrder = validParameter.validFile(parameters, "order", false);
 218                         if (flowOrder == "not found"){ flowOrder = "TACG";              }
 219                         else if(flowOrder.length() != 4){
 220                                 m->mothurOut("The value of the order option must be four bases long\n");
 221                         }
 222
 223                         if(oligoFileName == "") {       allFiles = 0;           }
 224                         else                                    {       allFiles = 1;           }
 225
 226                         numFPrimers = 0;
 227                         numRPrimers = 0;
 228             numLinkers = 0;
 229             numSpacers = 0;
 230                 }
 231
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
 235                 exit(1);
 236         }
 237 }
 238
 239 //***************************************************************************************************************
 240
 241 int TrimFlowsCommand::execute(){
 242         try{
 243
 244                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 245
 246         map<string, string> variables;
 247                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
 248         string fastaFileName = getOutputFileName("fasta",variables);
 249                 if(fasta){ outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName); }
 250
 251         variables["[tag]"] = "trim";
 252                 string trimFlowFileName = getOutputFileName("flow",variables);
 253                 outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
 254
 255         variables["[tag]"] = "scrap";
 256                 string scrapFlowFileName = getOutputFileName("flow",variables);
 257                 outputNames.push_back(scrapFlowFileName); outputTypes["flow"].push_back(scrapFlowFileName);
 258
 259
 260
 261                 vector<unsigned long long> flowFilePos;
 262         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 263                 flowFilePos = getFlowFileBreaks();
 264                 for (int i = 0; i < (flowFilePos.size()-1); i++) {
 265                         lines.push_back(new linePair(flowFilePos[i], flowFilePos[(i+1)]));
 266                 }
 267         #else
 268                 ifstream in; m->openInputFile(flowFileName, in); in >> numFlows; in.close();
 269         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 270                 processors = 1;
 271         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 272                 if (processors == 1) {
 273                         lines.push_back(new linePair(0, 1000));
 274                 }else {
 275                         int numFlowLines;
 276                         flowFilePos = m->setFilePosEachLine(flowFileName, numFlowLines);
 277                         flowFilePos.erase(flowFilePos.begin() + 1); numFlowLines--;
 278
 279                         //figure out how many sequences you have to process
 280                         int numSeqsPerProcessor = numFlowLines / processors;
 281                         cout << numSeqsPerProcessor << '\t' << numFlowLines << endl;
 282                         for (int i = 0; i < processors; i++) {
 283                                 int startIndex =  i * numSeqsPerProcessor;
 284                                 if(i == (processors - 1)){      numSeqsPerProcessor = numFlowLines - i * numSeqsPerProcessor;   }
 285                                 lines.push_back(new linePair(flowFilePos[startIndex], numSeqsPerProcessor));
 286                                 cout << flowFilePos[startIndex] << '\t' << numSeqsPerProcessor << endl;
 287                         }
 288                 }
 289         #endif
 290
 291                 vector<vector<string> > barcodePrimerComboFileNames;
 292                 if(oligoFileName != ""){
 293                         getOligos(barcodePrimerComboFileNames);
 294                 }
 295
 296                 if(processors == 1){
 297                         driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 298                 }else{
 299                         createProcessesCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames);
 300                 }
 301
 302                 if (m->control_pressed) {  return 0; }
 303
 304                 string flowFilesFileName;
 305                 ofstream output;
 306
 307                 if(allFiles){
 308                         set<string> namesAlreadyProcessed;
 309             variables["[tag]"] = "";
 310                         flowFilesFileName = getOutputFileName("file",variables);
 311                         m->openOutputFile(flowFilesFileName, output);
 312
 313                         for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 314                                 for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 315                                         if (namesAlreadyProcessed.count(barcodePrimerComboFileNames[i][j]) == 0) {
 316                                                 FILE * pFile;
 317                                                 unsigned long long size;
 318
 319                                                 //get num bytes in file
 320                                                 pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
 321                                                 if (pFile==NULL) perror ("Error opening file");
 322                                                 else{
 323                                                         fseek (pFile, 0, SEEK_END);
 324                                                         size=ftell(pFile);
 325                                                         fclose (pFile);
 326                                                 }
 327
 328                                                 if(size < 10){
 329                                                         m->mothurRemove(barcodePrimerComboFileNames[i][j]);
 330                                                 }
 331                                                 else{
 332                                                         output << m->getFullPathName(barcodePrimerComboFileNames[i][j]) << endl;
 333                                                         outputNames.push_back(barcodePrimerComboFileNames[i][j]);
 334                                                         outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
 335                                                 }
 336                                                 namesAlreadyProcessed.insert(barcodePrimerComboFileNames[i][j]);
 337                                         }
 338                                 }
 339                         }
 340                         output.close();
 341                 }
 342                 else{
 343             variables["[tag]"] = "";
 344                         flowFilesFileName = getOutputFileName("file",variables);
 345                         m->openOutputFile(flowFilesFileName, output);
 346
 347                         output << m->getFullPathName(trimFlowFileName) << endl;
 348
 349                         output.close();
 350                 }
 351                 outputTypes["file"].push_back(flowFilesFileName);
 352                 outputNames.push_back(flowFilesFileName);
 353
 354 //              set fasta file as new current fastafile
 355 //              string current = "";
 356 //              itTypes = outputTypes.find("fasta");
 357 //              if (itTypes != outputTypes.end()) {
 358 //                      if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 359 //              }
 360
 361                 m->mothurOutEndLine();
 362                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 363                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 364                 m->mothurOutEndLine();
 365
 366                 return 0;
 367         }
 368         catch(exception& e) {
 369                 m->errorOut(e, "TrimSeqsCommand", "execute");
 370                 exit(1);
 371         }
 372 }
 373
 374 //***************************************************************************************************************
 375
 376 int TrimFlowsCommand::driverCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > thisBarcodePrimerComboFileNames, linePair* line){
 377
 378         try {
 379                 ofstream trimFlowFile;
 380                 m->openOutputFile(trimFlowFileName, trimFlowFile);
 381                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 382
 383                 ofstream scrapFlowFile;
 384                 m->openOutputFile(scrapFlowFileName, scrapFlowFile);
 385                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 386
 387                 ofstream fastaFile;
 388                 if(fasta){      m->openOutputFile(fastaFileName, fastaFile);    }
 389
 390                 ifstream flowFile;
 391                 m->openInputFile(flowFileName, flowFile);
 392
 393                 flowFile.seekg(line->start);
 394
 395                 if(line->start == 0){
 396                         flowFile >> numFlows; m->gobble(flowFile);
 397                         scrapFlowFile << maxFlows << endl;
 398                         trimFlowFile << maxFlows << endl;
 399                         if(allFiles){
 400                                 for(int i=0;i<thisBarcodePrimerComboFileNames.size();i++){
 401                                         for(int j=0;j<thisBarcodePrimerComboFileNames[0].size();j++){
 402                                                 ofstream temp;
 403                                                 m->openOutputFile(thisBarcodePrimerComboFileNames[i][j], temp);
 404                                                 temp << maxFlows << endl;
 405                                                 temp.close();
 406                                         }
 407                                 }
 408                         }
 409                 }
 410
 411                 FlowData flowData(numFlows, signal, noise, maxHomoP, flowOrder);
 412                 //cout << " driver flowdata address " <<  &flowData  << &flowFile << endl;
 413                 int count = 0;
 414                 bool moreSeqs = 1;
 415
 416                 TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer);
 417
 418                 while(moreSeqs) {
 419
 420                         if (m->control_pressed) { break; }
 421
 422                         int success = 1;
 423                         int currentSeqDiffs = 0;
 424                         string trashCode = "";
 425
 426                         flowData.getNext(flowFile);
 427                         flowData.capFlows(maxFlows);
 428
 429                         Sequence currSeq = flowData.getSequence();
 430                         if(!flowData.hasMinFlows(minFlows)){    //screen to see if sequence is of a minimum number of flows
 431                                 success = 0;
 432                                 trashCode += 'l';
 433                         }
 434
 435                         int primerIndex = 0;
 436                         int barcodeIndex = 0;
 437
 438             if(numLinkers != 0){
 439                 success = trimOligos.stripLinker(currSeq);
 440                 if(success > ldiffs)            {       trashCode += 'k';       }
 441                 else{ currentSeqDiffs += success;  }
 442
 443             }
 444
 445             if (m->debug) { m->mothurOut("[DEBUG]: " + currSeq.getName() + " " + currSeq.getUnaligned() + "\n"); }
 446
 447                         if(barcodes.size() != 0){
 448                                 success = trimOligos.stripBarcode(currSeq, barcodeIndex);
 449                                 if(success > bdiffs)            {       trashCode += 'b';       }
 450                                 else{ currentSeqDiffs += success;  }
 451                         }
 452
 453             if(numSpacers != 0){
 454                 success = trimOligos.stripSpacer(currSeq);
 455                 if(success > sdiffs)            {       trashCode += 's';       }
 456                 else{ currentSeqDiffs += success;  }
 457
 458             }
 459
 460                         if(numFPrimers != 0){
 461                                 success = trimOligos.stripForward(currSeq, primerIndex);
 462                                 if(success > pdiffs)            {       trashCode += 'f';       }
 463                                 else{ currentSeqDiffs += success;  }
 464                         }
 465
 466                         if (currentSeqDiffs > tdiffs)   {       trashCode += 't';   }
 467
 468                         if(numRPrimers != 0){
 469                                 success = trimOligos.stripReverse(currSeq);
 470                                 if(!success)                            {       trashCode += 'r';       }
 471                         }
 472
 473                         if(trashCode.length() == 0){
 474
 475                                 flowData.printFlows(trimFlowFile);
 476
 477                                 if(fasta)       {       currSeq.printSequence(fastaFile);       }
 478
 479                                 if(allFiles){
 480                                         ofstream output;
 481                                         m->openOutputFileAppend(thisBarcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 482                                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 483
 484                                         flowData.printFlows(output);
 485                                         output.close();
 486                                 }
 487                         }
 488                         else{
 489                                 flowData.printFlows(scrapFlowFile, trashCode);
 490                         }
 491
 492                         count++;
 493                         //cout << "driver" << '\t' << currSeq.getName() << endl;
 494                         //report progress
 495                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 496
 497 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 498                         unsigned long long pos = flowFile.tellg();
 499
 500                         if ((pos == -1) || (pos >= line->end)) { break; }
 501 #else
 502                         if (flowFile.eof()) { break; }
 503 #endif
 504
 505                 }
 506                 //report progress
 507                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 508
 509                 trimFlowFile.close();
 510                 scrapFlowFile.close();
 511                 flowFile.close();
 512                 if(fasta){      fastaFile.close();      }
 513
 514                 return count;
 515         }
 516         catch(exception& e) {
 517                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 518                 exit(1);
 519         }
 520 }
 521
 522 //***************************************************************************************************************
 523
 524 void TrimFlowsCommand::getOligos(vector<vector<string> >& outFlowFileNames){
 525         try {
 526                 ifstream oligosFile;
 527                 m->openInputFile(oligoFileName, oligosFile);
 528
 529                 string type, oligo, group;
 530
 531                 int indexPrimer = 0;
 532                 int indexBarcode = 0;
 533
 534                 while(!oligosFile.eof()){
 535
 536                         oligosFile >> type; m->gobble(oligosFile);      //get the first column value of the row - is it a comment or a feature we are interested in?
 537
 538                         if(type[0] == '#'){     //igore the line because there's a comment
 539                                 while (!oligosFile.eof())       {       char c = oligosFile.get(); if (c == 10 || c == 13){     break;  }       } // get rest of line if there's any crap there
 540                         }
 541                         else{                           //there's a feature we're interested in
 542
 543                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }                                  //make type case insensitive
 544
 545                                 oligosFile >> oligo;    //get the DNA sequence for the feature
 546
 547                                 for(int i=0;i<oligo.length();i++){      //make type case insensitive and change any U's to T's
 548                                         oligo[i] = toupper(oligo[i]);
 549                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 550                                 }
 551
 552                                 if(type == "FORWARD"){  //if the feature is a forward primer...
 553                                         group = "";
 554
 555                                         while (!oligosFile.eof())       {       // get rest of line in case there is a primer name = will have the name of the primer
 556                                                 char c = oligosFile.get();
 557                                                 if (c == 10 || c == 13){        break;  }
 558                                                 else if (c == 32 || c == 9){;} //space or tab
 559                                                 else {  group += c;  }
 560                                         }
 561
 562                                         //have we seen this primer already?
 563                                         map<string, int>::iterator itPrimer = primers.find(oligo);
 564                                         if (itPrimer != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 565
 566                                         primers[oligo]=indexPrimer; indexPrimer++;
 567                                         primerNameVector.push_back(group);
 568
 569                                 }
 570                                 else if(type == "REVERSE"){
 571                                         string oligoRC = reverseOligo(oligo);
 572                                         revPrimer.push_back(oligoRC);
 573                                 }
 574                                 else if(type == "BARCODE"){
 575                                         oligosFile >> group;
 576
 577                                         //check for repeat barcodes
 578                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 579                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 580
 581                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 582                                         barcodeNameVector.push_back(group);
 583                                 }else if(type == "LINKER"){
 584                                         linker.push_back(oligo);
 585                                 }else if(type == "SPACER"){
 586                                         spacer.push_back(oligo);
 587                                 }
 588                                 else{
 589                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 590                                 }
 591                         }
 592
 593                         m->gobble(oligosFile);
 594                 }
 595                 oligosFile.close();
 596
 597                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 598
 599                 //add in potential combos
 600                 if(barcodeNameVector.size() == 0){
 601                         barcodes[""] = 0;
 602                         barcodeNameVector.push_back("");
 603                 }
 604
 605                 if(primerNameVector.size() == 0){
 606                         primers[""] = 0;
 607                         primerNameVector.push_back("");
 608                 }
 609
 610
 611                 outFlowFileNames.resize(barcodeNameVector.size());
 612                 for(int i=0;i<outFlowFileNames.size();i++){
 613                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 614                 }
 615
 616                 if(allFiles){
 617
 618                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 619                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 620
 621                                         string primerName = primerNameVector[itPrimer->second];
 622                                         string barcodeName = barcodeNameVector[itBar->second];
 623
 624                                         string comboGroupName = "";
 625                                         string fileName = "";
 626
 627                     map<string, string> variables;
 628                     variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
 629
 630                                         if(primerName == ""){
 631                                                 comboGroupName = barcodeNameVector[itBar->second];
 632                         variables["[tag]"] = comboGroupName;
 633                                                 fileName = getOutputFileName("flow", variables);
 634                                         }
 635                                         else{
 636                                                 if(barcodeName == ""){
 637                                                         comboGroupName = primerNameVector[itPrimer->second];
 638                                                 }
 639                                                 else{
 640                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 641                                                 }
 642                         variables["[tag]"] = comboGroupName;
 643                                                 fileName = getOutputFileName("flow", variables);
 644                                         }
 645
 646                                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 647
 648                                         ofstream temp;
 649                                         m->openOutputFile(fileName, temp);
 650                                         temp.close();
 651                                 }
 652                         }
 653                 }
 654
 655                 numFPrimers = primers.size();
 656                 numRPrimers = revPrimer.size();
 657         numLinkers = linker.size();
 658         numSpacers = spacer.size();
 659
 660         }
 661         catch(exception& e) {
 662                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 663                 exit(1);
 664         }
 665 }
 666 //********************************************************************/
 667 string TrimFlowsCommand::reverseOligo(string oligo){
 668         try {
 669         string reverse = "";
 670
 671         for(int i=oligo.length()-1;i>=0;i--){
 672
 673             if(oligo[i] == 'A')         {       reverse += 'T'; }
 674             else if(oligo[i] == 'T'){   reverse += 'A'; }
 675             else if(oligo[i] == 'U'){   reverse += 'A'; }
 676
 677             else if(oligo[i] == 'G'){   reverse += 'C'; }
 678             else if(oligo[i] == 'C'){   reverse += 'G'; }
 679
 680             else if(oligo[i] == 'R'){   reverse += 'Y'; }
 681             else if(oligo[i] == 'Y'){   reverse += 'R'; }
 682
 683             else if(oligo[i] == 'M'){   reverse += 'K'; }
 684             else if(oligo[i] == 'K'){   reverse += 'M'; }
 685
 686             else if(oligo[i] == 'W'){   reverse += 'W'; }
 687             else if(oligo[i] == 'S'){   reverse += 'S'; }
 688
 689             else if(oligo[i] == 'B'){   reverse += 'V'; }
 690             else if(oligo[i] == 'V'){   reverse += 'B'; }
 691
 692             else if(oligo[i] == 'D'){   reverse += 'H'; }
 693             else if(oligo[i] == 'H'){   reverse += 'D'; }
 694
 695             else                                                {       reverse += 'N'; }
 696         }
 697
 698
 699         return reverse;
 700     }
 701         catch(exception& e) {
 702                 m->errorOut(e, "TrimFlowsCommand", "reverseOligo");
 703                 exit(1);
 704         }
 705 }
 706
 707 /**************************************************************************************************/
 708 vector<unsigned long long> TrimFlowsCommand::getFlowFileBreaks() {
 709
 710         try{
 711
 712                 vector<unsigned long long> filePos;
 713                 filePos.push_back(0);
 714
 715                 FILE * pFile;
 716                 unsigned long long size;
 717
 718                 //get num bytes in file
 719                 pFile = fopen (flowFileName.c_str(),"rb");
 720                 if (pFile==NULL) perror ("Error opening file");
 721                 else{
 722                         fseek (pFile, 0, SEEK_END);
 723                         size=ftell (pFile);
 724                         fclose (pFile);
 725                 }
 726
 727                 //estimate file breaks
 728                 unsigned long long chunkSize = 0;
 729                 chunkSize = size / processors;
 730
 731                 //file too small to divide by processors
 732                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 733
 734                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 735                 for (int i = 0; i < processors; i++) {
 736                         unsigned long long spot = (i+1) * chunkSize;
 737
 738                         ifstream in;
 739                         m->openInputFile(flowFileName, in);
 740                         in.seekg(spot);
 741
 742                         string dummy = m->getline(in);
 743
 744                         //there was not another sequence before the end of the file
 745                         unsigned long long sanityPos = in.tellg();
 746
 747 //                      if (sanityPos == -1) {  break;  }
 748 //                      else {  filePos.push_back(newSpot);  }
 749                         if (sanityPos == -1) {  break;  }
 750                         else {  filePos.push_back(sanityPos);  }
 751
 752                         in.close();
 753                 }
 754
 755                 //save end pos
 756                 filePos.push_back(size);
 757
 758                 //sanity check filePos
 759                 for (int i = 0; i < (filePos.size()-1); i++) {
 760                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
 761                 }
 762
 763                 ifstream in;
 764                 m->openInputFile(flowFileName, in);
 765                 in >> numFlows;
 766                 m->gobble(in);
 767                 //unsigned long long spot = in.tellg();
 768                 //filePos[0] = spot;
 769                 in.close();
 770
 771                 processors = (filePos.size() - 1);
 772
 773                 return filePos;
 774         }
 775         catch(exception& e) {
 776                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
 777                 exit(1);
 778         }
 779 }
 780
 781 /**************************************************************************************************/
 782
 783 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
 784
 785         try {
 786                 processIDS.clear();
 787                 int exitCommand = 1;
 788
 789 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 790                 int process = 1;
 791
 792                 //loop through and create all the processes you want
 793                 while (process != processors) {
 794                         int pid = fork();
 795
 796                         if (pid > 0) {
 797                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 798                                 process++;
 799                         }else if (pid == 0){
 800
 801                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 802                                 if(allFiles){
 803                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 804                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 805                                                         tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 806                                                         ofstream temp;
 807                                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 808                                                         temp.close();
 809
 810                                                 }
 811                                         }
 812                                 }
 813                                 driverCreateTrim(flowFileName,
 814                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
 815                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
 816                                                                  (fastaFileName + toString(getpid()) + ".temp"),
 817                                                                  tempBarcodePrimerComboFileNames, lines[process]);
 818
 819                                 exit(0);
 820                         }else {
 821                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 822                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 823                                 exit(0);
 824                         }
 825                 }
 826
 827                 //parent do my part
 828                 ofstream temp;
 829                 m->openOutputFile(trimFlowFileName, temp);
 830                 temp.close();
 831
 832                 m->openOutputFile(scrapFlowFileName, temp);
 833                 temp.close();
 834
 835                 if(fasta){
 836                         m->openOutputFile(fastaFileName, temp);
 837                         temp.close();
 838                 }
 839
 840                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 841
 842                 //force parent to wait until all the processes are done
 843                 for (int i=0;i<processIDS.size();i++) {
 844                         int temp = processIDS[i];
 845                         wait(&temp);
 846                 }
 847 #else
 848                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 849                 //Windows version shared memory, so be careful when passing variables through the trimFlowData struct.
 850                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 851                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 852
 853                 vector<trimFlowData*> pDataArray;
 854                 DWORD   dwThreadIdArray[processors-1];
 855                 HANDLE  hThreadArray[processors-1];
 856
 857                 //Create processor worker threads.
 858                 for( int i=0; i<processors-1; i++ ){
 859                         // Allocate memory for thread data.
 860                         string extension = "";
 861                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
 862
 863                         vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 864                         if(allFiles){
 865                                 for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 866                                         for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 867                                                 tempBarcodePrimerComboFileNames[i][j] += extension;
 868                                                 ofstream temp;
 869                                                 m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 870                                                 temp.close();
 871
 872                                         }
 873                                 }
 874                         }
 875
 876                         trimFlowData* tempflow = new trimFlowData(flowFileName, (trimFlowFileName + extension), (scrapFlowFileName + extension), fastaFileName, flowOrder, tempBarcodePrimerComboFileNames, barcodes, primers, revPrimer, fasta, allFiles, lines[i]->start, lines[i]->end, m, signal, noise, numFlows, maxFlows, minFlows, maxHomoP, tdiffs, bdiffs, pdiffs, i);
 877                         pDataArray.push_back(tempflow);
 878
 879                         //MyTrimFlowThreadFunction is in header. It must be global or static to work with the threads.
 880                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 881                         hThreadArray[i] = CreateThread(NULL, 0, MyTrimFlowThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 882                 }
 883
 884                 //using the main process as a worker saves time and memory
 885                 ofstream temp;
 886                 m->openOutputFile(trimFlowFileName, temp);
 887                 temp.close();
 888
 889                 m->openOutputFile(scrapFlowFileName, temp);
 890                 temp.close();
 891
 892                 if(fasta){
 893                         m->openOutputFile(fastaFileName, temp);
 894                         temp.close();
 895                 }
 896
 897                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 898                 if(allFiles){
 899                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 900                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 901                                         tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
 902                                         ofstream temp;
 903                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 904                                         temp.close();
 905
 906                                 }
 907                         }
 908                 }
 909
 910                 //do my part - do last piece because windows is looking for eof
 911                 int num = driverCreateTrim(flowFileName, (trimFlowFileName  + toString(processors-1) + ".temp"), (scrapFlowFileName  + toString(processors-1) + ".temp"), (fastaFileName + toString(processors-1) + ".temp"), tempBarcodePrimerComboFileNames, lines[processors-1]);
 912                 processIDS.push_back((processors-1));
 913
 914                 //Wait until all threads have terminated.
 915                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 916
 917                 //Close all thread handles and free memory allocations.
 918                 for(int i=0; i < pDataArray.size(); i++){
 919                         num += pDataArray[i]->count;
 920                         CloseHandle(hThreadArray[i]);
 921                         delete pDataArray[i];
 922                 }
 923
 924
 925 #endif
 926                 //append files
 927                 m->mothurOutEndLine();
 928                 for(int i=0;i<processIDS.size();i++){
 929
 930                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 931
 932                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
 933                         m->mothurRemove((trimFlowFileName + toString(processIDS[i]) + ".temp"));
 934 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
 935
 936                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
 937                         m->mothurRemove((scrapFlowFileName + toString(processIDS[i]) + ".temp"));
 938 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
 939
 940                         if(fasta){
 941                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
 942                                 m->mothurRemove((fastaFileName + toString(processIDS[i]) + ".temp"));
 943 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
 944                         }
 945                         if(allFiles){
 946                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
 947                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
 948                                                 m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
 949                                                 m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
 950                                         }
 951                                 }
 952                         }
 953                 }
 954
 955                 return exitCommand;
 956
 957         }
 958         catch(exception& e) {
 959                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
 960                 exit(1);
 961         }
 962 }
 963
 964 //***************************************************************************************************************