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