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, primers, barcodes, revPrimer);
 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
 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                                         Sequence oligoRC("reverse", oligo);
 546                                         oligoRC.reverseComplement();
 547                                         revPrimer.push_back(oligoRC.getUnaligned());
 548                                 }
 549                                 else if(type == "BARCODE"){
 550                                         oligosFile >> group;
 551
 552                                         //check for repeat barcodes
 553                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 554                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 555
 556                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 557                                         barcodeNameVector.push_back(group);
 558                                 }else if(type == "LINKER"){
 559                                         linker.push_back(oligo);
 560                                 }else if(type == "SPACER"){
 561                                         spacer.push_back(oligo);
 562                                 }
 563                                 else{
 564                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 565                                 }
 566                         }
 567
 568                         m->gobble(oligosFile);
 569                 }
 570                 oligosFile.close();
 571
 572                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 573
 574                 //add in potential combos
 575                 if(barcodeNameVector.size() == 0){
 576                         barcodes[""] = 0;
 577                         barcodeNameVector.push_back("");
 578                 }
 579
 580                 if(primerNameVector.size() == 0){
 581                         primers[""] = 0;
 582                         primerNameVector.push_back("");
 583                 }
 584
 585
 586                 outFlowFileNames.resize(barcodeNameVector.size());
 587                 for(int i=0;i<outFlowFileNames.size();i++){
 588                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 589                 }
 590
 591                 if(allFiles){
 592
 593                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 594                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 595
 596                                         string primerName = primerNameVector[itPrimer->second];
 597                                         string barcodeName = barcodeNameVector[itBar->second];
 598
 599                                         string comboGroupName = "";
 600                                         string fileName = "";
 601
 602                                         if(primerName == ""){
 603                                                 comboGroupName = barcodeNameVector[itBar->second];
 604                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 605                                         }
 606                                         else{
 607                                                 if(barcodeName == ""){
 608                                                         comboGroupName = primerNameVector[itPrimer->second];
 609                                                 }
 610                                                 else{
 611                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 612                                                 }
 613                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 614                                         }
 615
 616                                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 617
 618                                         ofstream temp;
 619                                         m->openOutputFile(fileName, temp);
 620                                         temp.close();
 621                                 }
 622                         }
 623                 }
 624
 625                 numFPrimers = primers.size();
 626                 numRPrimers = revPrimer.size();
 627         numLinkers = linker.size();
 628         numSpacers = spacer.size();
 629
 630         }
 631         catch(exception& e) {
 632                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 633                 exit(1);
 634         }
 635 }
 636 /**************************************************************************************************/
 637 vector<unsigned long long> TrimFlowsCommand::getFlowFileBreaks() {
 638
 639         try{
 640
 641                 vector<unsigned long long> filePos;
 642                 filePos.push_back(0);
 643
 644                 FILE * pFile;
 645                 unsigned long long size;
 646
 647                 //get num bytes in file
 648                 pFile = fopen (flowFileName.c_str(),"rb");
 649                 if (pFile==NULL) perror ("Error opening file");
 650                 else{
 651                         fseek (pFile, 0, SEEK_END);
 652                         size=ftell (pFile);
 653                         fclose (pFile);
 654                 }
 655
 656                 //estimate file breaks
 657                 unsigned long long chunkSize = 0;
 658                 chunkSize = size / processors;
 659
 660                 //file too small to divide by processors
 661                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 662
 663                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 664                 for (int i = 0; i < processors; i++) {
 665                         unsigned long long spot = (i+1) * chunkSize;
 666
 667                         ifstream in;
 668                         m->openInputFile(flowFileName, in);
 669                         in.seekg(spot);
 670
 671                         string dummy = m->getline(in);
 672
 673                         //there was not another sequence before the end of the file
 674                         unsigned long long sanityPos = in.tellg();
 675
 676 //                      if (sanityPos == -1) {  break;  }
 677 //                      else {  filePos.push_back(newSpot);  }
 678                         if (sanityPos == -1) {  break;  }
 679                         else {  filePos.push_back(sanityPos);  }
 680
 681                         in.close();
 682                 }
 683
 684                 //save end pos
 685                 filePos.push_back(size);
 686
 687                 //sanity check filePos
 688                 for (int i = 0; i < (filePos.size()-1); i++) {
 689                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
 690                 }
 691
 692                 ifstream in;
 693                 m->openInputFile(flowFileName, in);
 694                 in >> numFlows;
 695                 m->gobble(in);
 696                 //unsigned long long spot = in.tellg();
 697                 //filePos[0] = spot;
 698                 in.close();
 699
 700                 processors = (filePos.size() - 1);
 701
 702                 return filePos;
 703         }
 704         catch(exception& e) {
 705                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
 706                 exit(1);
 707         }
 708 }
 709
 710 /**************************************************************************************************/
 711
 712 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
 713
 714         try {
 715                 processIDS.clear();
 716                 int exitCommand = 1;
 717
 718 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 719                 int process = 1;
 720
 721                 //loop through and create all the processes you want
 722                 while (process != processors) {
 723                         int pid = fork();
 724
 725                         if (pid > 0) {
 726                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 727                                 process++;
 728                         }else if (pid == 0){
 729
 730                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 731                                 if(allFiles){
 732                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 733                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 734                                                         tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 735                                                         ofstream temp;
 736                                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 737                                                         temp.close();
 738
 739                                                 }
 740                                         }
 741                                 }
 742                                 driverCreateTrim(flowFileName,
 743                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
 744                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
 745                                                                  (fastaFileName + toString(getpid()) + ".temp"),
 746                                                                  tempBarcodePrimerComboFileNames, lines[process]);
 747
 748                                 exit(0);
 749                         }else {
 750                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 751                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 752                                 exit(0);
 753                         }
 754                 }
 755
 756                 //parent do my part
 757                 ofstream temp;
 758                 m->openOutputFile(trimFlowFileName, temp);
 759                 temp.close();
 760
 761                 m->openOutputFile(scrapFlowFileName, temp);
 762                 temp.close();
 763
 764                 if(fasta){
 765                         m->openOutputFile(fastaFileName, temp);
 766                         temp.close();
 767                 }
 768
 769                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 770
 771                 //force parent to wait until all the processes are done
 772                 for (int i=0;i<processIDS.size();i++) {
 773                         int temp = processIDS[i];
 774                         wait(&temp);
 775                 }
 776 #else
 777                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 778                 //Windows version shared memory, so be careful when passing variables through the trimFlowData struct.
 779                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 780                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 781
 782                 vector<trimFlowData*> pDataArray;
 783                 DWORD   dwThreadIdArray[processors-1];
 784                 HANDLE  hThreadArray[processors-1];
 785
 786                 //Create processor worker threads.
 787                 for( int i=0; i<processors-1; i++ ){
 788                         // Allocate memory for thread data.
 789                         string extension = "";
 790                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
 791
 792                         vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 793                         if(allFiles){
 794                                 for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 795                                         for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 796                                                 tempBarcodePrimerComboFileNames[i][j] += extension;
 797                                                 ofstream temp;
 798                                                 m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 799                                                 temp.close();
 800
 801                                         }
 802                                 }
 803                         }
 804
 805                         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);
 806                         pDataArray.push_back(tempflow);
 807
 808                         //MyTrimFlowThreadFunction is in header. It must be global or static to work with the threads.
 809                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 810                         hThreadArray[i] = CreateThread(NULL, 0, MyTrimFlowThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 811                 }
 812
 813                 //using the main process as a worker saves time and memory
 814                 ofstream temp;
 815                 m->openOutputFile(trimFlowFileName, temp);
 816                 temp.close();
 817
 818                 m->openOutputFile(scrapFlowFileName, temp);
 819                 temp.close();
 820
 821                 if(fasta){
 822                         m->openOutputFile(fastaFileName, temp);
 823                         temp.close();
 824                 }
 825
 826                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 827                 if(allFiles){
 828                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 829                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 830                                         tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
 831                                         ofstream temp;
 832                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 833                                         temp.close();
 834
 835                                 }
 836                         }
 837                 }
 838
 839                 //do my part - do last piece because windows is looking for eof
 840                 int num = driverCreateTrim(flowFileName, (trimFlowFileName  + toString(processors-1) + ".temp"), (scrapFlowFileName  + toString(processors-1) + ".temp"), (fastaFileName + toString(processors-1) + ".temp"), tempBarcodePrimerComboFileNames, lines[processors-1]);
 841                 processIDS.push_back((processors-1));
 842
 843                 //Wait until all threads have terminated.
 844                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 845
 846                 //Close all thread handles and free memory allocations.
 847                 for(int i=0; i < pDataArray.size(); i++){
 848                         num += pDataArray[i]->count;
 849                         CloseHandle(hThreadArray[i]);
 850                         delete pDataArray[i];
 851                 }
 852
 853
 854 #endif
 855                 //append files
 856                 m->mothurOutEndLine();
 857                 for(int i=0;i<processIDS.size();i++){
 858
 859                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 860
 861                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
 862                         m->mothurRemove((trimFlowFileName + toString(processIDS[i]) + ".temp"));
 863 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
 864
 865                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
 866                         m->mothurRemove((scrapFlowFileName + toString(processIDS[i]) + ".temp"));
 867 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
 868
 869                         if(fasta){
 870                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
 871                                 m->mothurRemove((fastaFileName + toString(processIDS[i]) + ".temp"));
 872 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
 873                         }
 874                         if(allFiles){
 875                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
 876                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
 877                                                 m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
 878                                                 m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
 879                                         }
 880                                 }
 881                         }
 882                 }
 883
 884                 return exitCommand;
 885
 886         }
 887         catch(exception& e) {
 888                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
 889                 exit(1);
 890         }
 891 }
 892
 893 //***************************************************************************************************************