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