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             numLinkers = 0;
 211             numSpacers = 0;
 212                 }
 213
 214         }
 215         catch(exception& e) {
 216                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
 217                 exit(1);
 218         }
 219 }
 220
 221 //***************************************************************************************************************
 222
 223 int TrimFlowsCommand::execute(){
 224         try{
 225
 226                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 227
 228                 string trimFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "trim.flow";
 229                 outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
 230
 231                 string scrapFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "scrap.flow";
 232                 outputNames.push_back(scrapFlowFileName); outputTypes["flow"].push_back(scrapFlowFileName);
 233
 234                 string fastaFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.fasta";
 235                 if(fasta){
 236                         outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName);
 237                 }
 238
 239                 vector<unsigned long long> flowFilePos;
 240         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 241                 flowFilePos = getFlowFileBreaks();
 242                 for (int i = 0; i < (flowFilePos.size()-1); i++) {
 243                         lines.push_back(new linePair(flowFilePos[i], flowFilePos[(i+1)]));
 244                 }
 245         #else
 246                 ifstream in; m->openInputFile(flowFileName, in); in >> numFlows; in.close();
 247         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 248                 processors = 1;
 249         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 250                 if (processors == 1) {
 251                         lines.push_back(new linePair(0, 1000));
 252                 }else {
 253                         int numFlowLines;
 254                         flowFilePos = m->setFilePosEachLine(flowFileName, numFlowLines);
 255                         flowFilePos.erase(flowFilePos.begin() + 1); numFlowLines--;
 256
 257                         //figure out how many sequences you have to process
 258                         int numSeqsPerProcessor = numFlowLines / processors;
 259                         cout << numSeqsPerProcessor << '\t' << numFlowLines << endl;
 260                         for (int i = 0; i < processors; i++) {
 261                                 int startIndex =  i * numSeqsPerProcessor;
 262                                 if(i == (processors - 1)){      numSeqsPerProcessor = numFlowLines - i * numSeqsPerProcessor;   }
 263                                 lines.push_back(new linePair(flowFilePos[startIndex], numSeqsPerProcessor));
 264                                 cout << flowFilePos[startIndex] << '\t' << numSeqsPerProcessor << endl;
 265                         }
 266                 }
 267         #endif
 268
 269                 vector<vector<string> > barcodePrimerComboFileNames;
 270                 if(oligoFileName != ""){
 271                         getOligos(barcodePrimerComboFileNames);
 272                 }
 273
 274                 if(processors == 1){
 275                         driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 276                 }else{
 277                         createProcessesCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames);
 278                 }
 279
 280                 if (m->control_pressed) {  return 0; }
 281
 282                 string flowFilesFileName;
 283                 ofstream output;
 284
 285                 if(allFiles){
 286                         set<string> namesAlreadyProcessed;
 287                         flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
 288                         m->openOutputFile(flowFilesFileName, output);
 289
 290                         for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 291                                 for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 292                                         if (namesAlreadyProcessed.count(barcodePrimerComboFileNames[i][j]) == 0) {
 293                                                 FILE * pFile;
 294                                                 unsigned long long size;
 295
 296                                                 //get num bytes in file
 297                                                 pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
 298                                                 if (pFile==NULL) perror ("Error opening file");
 299                                                 else{
 300                                                         fseek (pFile, 0, SEEK_END);
 301                                                         size=ftell(pFile);
 302                                                         fclose (pFile);
 303                                                 }
 304
 305                                                 if(size < 10){
 306                                                         m->mothurRemove(barcodePrimerComboFileNames[i][j]);
 307                                                 }
 308                                                 else{
 309                                                         output << m->getFullPathName(barcodePrimerComboFileNames[i][j]) << endl;
 310                                                         outputNames.push_back(barcodePrimerComboFileNames[i][j]);
 311                                                         outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
 312                                                 }
 313                                                 namesAlreadyProcessed.insert(barcodePrimerComboFileNames[i][j]);
 314                                         }
 315                                 }
 316                         }
 317                         output.close();
 318                 }
 319                 else{
 320                         flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
 321                         m->openOutputFile(flowFilesFileName, output);
 322
 323                         output << m->getFullPathName(trimFlowFileName) << endl;
 324
 325                         output.close();
 326                 }
 327                 outputTypes["flow.files"].push_back(flowFilesFileName);
 328                 outputNames.push_back(flowFilesFileName);
 329
 330 //              set fasta file as new current fastafile
 331 //              string current = "";
 332 //              itTypes = outputTypes.find("fasta");
 333 //              if (itTypes != outputTypes.end()) {
 334 //                      if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 335 //              }
 336
 337                 m->mothurOutEndLine();
 338                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 339                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 340                 m->mothurOutEndLine();
 341
 342                 return 0;
 343         }
 344         catch(exception& e) {
 345                 m->errorOut(e, "TrimSeqsCommand", "execute");
 346                 exit(1);
 347         }
 348 }
 349
 350 //***************************************************************************************************************
 351
 352 int TrimFlowsCommand::driverCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > thisBarcodePrimerComboFileNames, linePair* line){
 353
 354         try {
 355                 ofstream trimFlowFile;
 356                 m->openOutputFile(trimFlowFileName, trimFlowFile);
 357                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 358
 359                 ofstream scrapFlowFile;
 360                 m->openOutputFile(scrapFlowFileName, scrapFlowFile);
 361                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 362
 363                 ofstream fastaFile;
 364                 if(fasta){      m->openOutputFile(fastaFileName, fastaFile);    }
 365
 366                 ifstream flowFile;
 367                 m->openInputFile(flowFileName, flowFile);
 368
 369                 flowFile.seekg(line->start);
 370
 371                 if(line->start == 0){
 372                         flowFile >> numFlows; m->gobble(flowFile);
 373                         scrapFlowFile << maxFlows << endl;
 374                         trimFlowFile << maxFlows << endl;
 375                         if(allFiles){
 376                                 for(int i=0;i<thisBarcodePrimerComboFileNames.size();i++){
 377                                         for(int j=0;j<thisBarcodePrimerComboFileNames[0].size();j++){
 378                                                 ofstream temp;
 379                                                 m->openOutputFile(thisBarcodePrimerComboFileNames[i][j], temp);
 380                                                 temp << maxFlows << endl;
 381                                                 temp.close();
 382                                         }
 383                                 }
 384                         }
 385                 }
 386
 387                 FlowData flowData(numFlows, signal, noise, maxHomoP, flowOrder);
 388                 //cout << " driver flowdata address " <<  &flowData  << &flowFile << endl;
 389                 int count = 0;
 390                 bool moreSeqs = 1;
 391
 392                 TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer);
 393
 394                 while(moreSeqs) {
 395
 396                         if (m->control_pressed) { break; }
 397
 398                         int success = 1;
 399                         int currentSeqDiffs = 0;
 400                         string trashCode = "";
 401
 402                         flowData.getNext(flowFile);
 403                         //cout << "driver good bit " << flowFile.good() << endl;
 404                         flowData.capFlows(maxFlows);
 405
 406                         Sequence currSeq = flowData.getSequence();
 407
 408                         if(!flowData.hasMinFlows(minFlows)){    //screen to see if sequence is of a minimum number of flows
 409                                 success = 0;
 410                                 trashCode += 'l';
 411                         }
 412
 413                         int primerIndex = 0;
 414                         int barcodeIndex = 0;
 415
 416             if(numLinkers != 0){
 417                 success = trimOligos.stripLinker(currSeq);
 418                 if(success > ldiffs)            {       trashCode += 'k';       }
 419                 else{ currentSeqDiffs += success;  }
 420
 421             }
 422
 423                         if(barcodes.size() != 0){
 424                                 success = trimOligos.stripBarcode(currSeq, barcodeIndex);
 425                                 if(success > bdiffs)            {       trashCode += 'b';       }
 426                                 else{ currentSeqDiffs += success;  }
 427                         }
 428
 429             if(numSpacers != 0){
 430                 success = trimOligos.stripSpacer(currSeq);
 431                 if(success > sdiffs)            {       trashCode += 's';       }
 432                 else{ currentSeqDiffs += success;  }
 433
 434             }
 435
 436                         if(numFPrimers != 0){
 437                                 success = trimOligos.stripForward(currSeq, primerIndex);
 438                                 if(success > pdiffs)            {       trashCode += 'f';       }
 439                                 else{ currentSeqDiffs += success;  }
 440                         }
 441
 442                         if (currentSeqDiffs > tdiffs)   {       trashCode += 't';   }
 443
 444                         if(numRPrimers != 0){
 445                                 success = trimOligos.stripReverse(currSeq);
 446                                 if(!success)                            {       trashCode += 'r';       }
 447                         }
 448
 449                         if(trashCode.length() == 0){
 450
 451                                 flowData.printFlows(trimFlowFile);
 452
 453                                 if(fasta)       {       currSeq.printSequence(fastaFile);       }
 454
 455                                 if(allFiles){
 456                                         ofstream output;
 457                                         m->openOutputFileAppend(thisBarcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 458                                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 459
 460                                         flowData.printFlows(output);
 461                                         output.close();
 462                                 }
 463                         }
 464                         else{
 465                                 flowData.printFlows(scrapFlowFile, trashCode);
 466                         }
 467
 468                         count++;
 469                         //cout << "driver" << '\t' << currSeq.getName() << endl;
 470                         //report progress
 471                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 472
 473 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 474                         unsigned long long pos = flowFile.tellg();
 475
 476                         if ((pos == -1) || (pos >= line->end)) { break; }
 477 #else
 478                         if (flowFile.eof()) { break; }
 479 #endif
 480
 481                 }
 482                 //report progress
 483                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 484
 485                 trimFlowFile.close();
 486                 scrapFlowFile.close();
 487                 flowFile.close();
 488                 if(fasta){      fastaFile.close();      }
 489
 490                 return count;
 491         }
 492         catch(exception& e) {
 493                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 494                 exit(1);
 495         }
 496 }
 497
 498 //***************************************************************************************************************
 499
 500 void TrimFlowsCommand::getOligos(vector<vector<string> >& outFlowFileNames){
 501         try {
 502                 ifstream oligosFile;
 503                 m->openInputFile(oligoFileName, oligosFile);
 504
 505                 string type, oligo, group;
 506
 507                 int indexPrimer = 0;
 508                 int indexBarcode = 0;
 509
 510                 while(!oligosFile.eof()){
 511
 512                         oligosFile >> type; m->gobble(oligosFile);      //get the first column value of the row - is it a comment or a feature we are interested in?
 513
 514                         if(type[0] == '#'){     //igore the line because there's a comment
 515                                 while (!oligosFile.eof())       {       char c = oligosFile.get(); if (c == 10 || c == 13){     break;  }       } // get rest of line if there's any crap there
 516                         }
 517                         else{                           //there's a feature we're interested in
 518
 519                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }                                  //make type case insensitive
 520
 521                                 oligosFile >> oligo;    //get the DNA sequence for the feature
 522
 523                                 for(int i=0;i<oligo.length();i++){      //make type case insensitive and change any U's to T's
 524                                         oligo[i] = toupper(oligo[i]);
 525                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 526                                 }
 527
 528                                 if(type == "FORWARD"){  //if the feature is a forward primer...
 529                                         group = "";
 530
 531                                         while (!oligosFile.eof())       {       // get rest of line in case there is a primer name = will have the name of the primer
 532                                                 char c = oligosFile.get();
 533                                                 if (c == 10 || c == 13){        break;  }
 534                                                 else if (c == 32 || c == 9){;} //space or tab
 535                                                 else {  group += c;  }
 536                                         }
 537
 538                                         //have we seen this primer already?
 539                                         map<string, int>::iterator itPrimer = primers.find(oligo);
 540                                         if (itPrimer != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 541
 542                                         primers[oligo]=indexPrimer; indexPrimer++;
 543                                         primerNameVector.push_back(group);
 544
 545                                 }
 546                                 else if(type == "REVERSE"){
 547                                         string oligoRC = reverseOligo(oligo);
 548                                         revPrimer.push_back(oligoRC);
 549                                 }
 550                                 else if(type == "BARCODE"){
 551                                         oligosFile >> group;
 552
 553                                         //check for repeat barcodes
 554                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 555                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 556
 557                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 558                                         barcodeNameVector.push_back(group);
 559                                 }else if(type == "LINKER"){
 560                                         linker.push_back(oligo);
 561                                 }else if(type == "SPACER"){
 562                                         spacer.push_back(oligo);
 563                                 }
 564                                 else{
 565                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 566                                 }
 567                         }
 568
 569                         m->gobble(oligosFile);
 570                 }
 571                 oligosFile.close();
 572
 573                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 574
 575                 //add in potential combos
 576                 if(barcodeNameVector.size() == 0){
 577                         barcodes[""] = 0;
 578                         barcodeNameVector.push_back("");
 579                 }
 580
 581                 if(primerNameVector.size() == 0){
 582                         primers[""] = 0;
 583                         primerNameVector.push_back("");
 584                 }
 585
 586
 587                 outFlowFileNames.resize(barcodeNameVector.size());
 588                 for(int i=0;i<outFlowFileNames.size();i++){
 589                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 590                 }
 591
 592                 if(allFiles){
 593
 594                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 595                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 596
 597                                         string primerName = primerNameVector[itPrimer->second];
 598                                         string barcodeName = barcodeNameVector[itBar->second];
 599
 600                                         string comboGroupName = "";
 601                                         string fileName = "";
 602
 603                                         if(primerName == ""){
 604                                                 comboGroupName = barcodeNameVector[itBar->second];
 605                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 606                                         }
 607                                         else{
 608                                                 if(barcodeName == ""){
 609                                                         comboGroupName = primerNameVector[itPrimer->second];
 610                                                 }
 611                                                 else{
 612                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 613                                                 }
 614                                                 fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
 615                                         }
 616
 617                                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 618
 619                                         ofstream temp;
 620                                         m->openOutputFile(fileName, temp);
 621                                         temp.close();
 622                                 }
 623                         }
 624                 }
 625
 626                 numFPrimers = primers.size();
 627                 numRPrimers = revPrimer.size();
 628         numLinkers = linker.size();
 629         numSpacers = spacer.size();
 630
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 634                 exit(1);
 635         }
 636 }
 637 //********************************************************************/
 638 string TrimFlowsCommand::reverseOligo(string oligo){
 639         try {
 640         string reverse = "";
 641
 642         for(int i=oligo.length()-1;i>=0;i--){
 643
 644             if(oligo[i] == 'A')         {       reverse += 'T'; }
 645             else if(oligo[i] == 'T'){   reverse += 'A'; }
 646             else if(oligo[i] == 'U'){   reverse += 'A'; }
 647
 648             else if(oligo[i] == 'G'){   reverse += 'C'; }
 649             else if(oligo[i] == 'C'){   reverse += 'G'; }
 650
 651             else if(oligo[i] == 'R'){   reverse += 'Y'; }
 652             else if(oligo[i] == 'Y'){   reverse += 'R'; }
 653
 654             else if(oligo[i] == 'M'){   reverse += 'K'; }
 655             else if(oligo[i] == 'K'){   reverse += 'M'; }
 656
 657             else if(oligo[i] == 'W'){   reverse += 'W'; }
 658             else if(oligo[i] == 'S'){   reverse += 'S'; }
 659
 660             else if(oligo[i] == 'B'){   reverse += 'V'; }
 661             else if(oligo[i] == 'V'){   reverse += 'B'; }
 662
 663             else if(oligo[i] == 'D'){   reverse += 'H'; }
 664             else if(oligo[i] == 'H'){   reverse += 'D'; }
 665
 666             else                                                {       reverse += 'N'; }
 667         }
 668
 669
 670         return reverse;
 671     }
 672         catch(exception& e) {
 673                 m->errorOut(e, "TrimFlowsCommand", "reverseOligo");
 674                 exit(1);
 675         }
 676 }
 677
 678 /**************************************************************************************************/
 679 vector<unsigned long long> TrimFlowsCommand::getFlowFileBreaks() {
 680
 681         try{
 682
 683                 vector<unsigned long long> filePos;
 684                 filePos.push_back(0);
 685
 686                 FILE * pFile;
 687                 unsigned long long size;
 688
 689                 //get num bytes in file
 690                 pFile = fopen (flowFileName.c_str(),"rb");
 691                 if (pFile==NULL) perror ("Error opening file");
 692                 else{
 693                         fseek (pFile, 0, SEEK_END);
 694                         size=ftell (pFile);
 695                         fclose (pFile);
 696                 }
 697
 698                 //estimate file breaks
 699                 unsigned long long chunkSize = 0;
 700                 chunkSize = size / processors;
 701
 702                 //file too small to divide by processors
 703                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 704
 705                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 706                 for (int i = 0; i < processors; i++) {
 707                         unsigned long long spot = (i+1) * chunkSize;
 708
 709                         ifstream in;
 710                         m->openInputFile(flowFileName, in);
 711                         in.seekg(spot);
 712
 713                         string dummy = m->getline(in);
 714
 715                         //there was not another sequence before the end of the file
 716                         unsigned long long sanityPos = in.tellg();
 717
 718 //                      if (sanityPos == -1) {  break;  }
 719 //                      else {  filePos.push_back(newSpot);  }
 720                         if (sanityPos == -1) {  break;  }
 721                         else {  filePos.push_back(sanityPos);  }
 722
 723                         in.close();
 724                 }
 725
 726                 //save end pos
 727                 filePos.push_back(size);
 728
 729                 //sanity check filePos
 730                 for (int i = 0; i < (filePos.size()-1); i++) {
 731                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
 732                 }
 733
 734                 ifstream in;
 735                 m->openInputFile(flowFileName, in);
 736                 in >> numFlows;
 737                 m->gobble(in);
 738                 //unsigned long long spot = in.tellg();
 739                 //filePos[0] = spot;
 740                 in.close();
 741
 742                 processors = (filePos.size() - 1);
 743
 744                 return filePos;
 745         }
 746         catch(exception& e) {
 747                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
 748                 exit(1);
 749         }
 750 }
 751
 752 /**************************************************************************************************/
 753
 754 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
 755
 756         try {
 757                 processIDS.clear();
 758                 int exitCommand = 1;
 759
 760 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 761                 int process = 1;
 762
 763                 //loop through and create all the processes you want
 764                 while (process != processors) {
 765                         int pid = fork();
 766
 767                         if (pid > 0) {
 768                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 769                                 process++;
 770                         }else if (pid == 0){
 771
 772                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 773                                 if(allFiles){
 774                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 775                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 776                                                         tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 777                                                         ofstream temp;
 778                                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 779                                                         temp.close();
 780
 781                                                 }
 782                                         }
 783                                 }
 784                                 driverCreateTrim(flowFileName,
 785                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
 786                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
 787                                                                  (fastaFileName + toString(getpid()) + ".temp"),
 788                                                                  tempBarcodePrimerComboFileNames, lines[process]);
 789
 790                                 exit(0);
 791                         }else {
 792                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 793                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 794                                 exit(0);
 795                         }
 796                 }
 797
 798                 //parent do my part
 799                 ofstream temp;
 800                 m->openOutputFile(trimFlowFileName, temp);
 801                 temp.close();
 802
 803                 m->openOutputFile(scrapFlowFileName, temp);
 804                 temp.close();
 805
 806                 if(fasta){
 807                         m->openOutputFile(fastaFileName, temp);
 808                         temp.close();
 809                 }
 810
 811                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 812
 813                 //force parent to wait until all the processes are done
 814                 for (int i=0;i<processIDS.size();i++) {
 815                         int temp = processIDS[i];
 816                         wait(&temp);
 817                 }
 818 #else
 819                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 820                 //Windows version shared memory, so be careful when passing variables through the trimFlowData struct.
 821                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 822                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 823
 824                 vector<trimFlowData*> pDataArray;
 825                 DWORD   dwThreadIdArray[processors-1];
 826                 HANDLE  hThreadArray[processors-1];
 827
 828                 //Create processor worker threads.
 829                 for( int i=0; i<processors-1; i++ ){
 830                         // Allocate memory for thread data.
 831                         string extension = "";
 832                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
 833
 834                         vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 835                         if(allFiles){
 836                                 for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 837                                         for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 838                                                 tempBarcodePrimerComboFileNames[i][j] += extension;
 839                                                 ofstream temp;
 840                                                 m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 841                                                 temp.close();
 842
 843                                         }
 844                                 }
 845                         }
 846
 847                         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);
 848                         pDataArray.push_back(tempflow);
 849
 850                         //MyTrimFlowThreadFunction is in header. It must be global or static to work with the threads.
 851                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 852                         hThreadArray[i] = CreateThread(NULL, 0, MyTrimFlowThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 853                 }
 854
 855                 //using the main process as a worker saves time and memory
 856                 ofstream temp;
 857                 m->openOutputFile(trimFlowFileName, temp);
 858                 temp.close();
 859
 860                 m->openOutputFile(scrapFlowFileName, temp);
 861                 temp.close();
 862
 863                 if(fasta){
 864                         m->openOutputFile(fastaFileName, temp);
 865                         temp.close();
 866                 }
 867
 868                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 869                 if(allFiles){
 870                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 871                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 872                                         tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
 873                                         ofstream temp;
 874                                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 875                                         temp.close();
 876
 877                                 }
 878                         }
 879                 }
 880
 881                 //do my part - do last piece because windows is looking for eof
 882                 int num = driverCreateTrim(flowFileName, (trimFlowFileName  + toString(processors-1) + ".temp"), (scrapFlowFileName  + toString(processors-1) + ".temp"), (fastaFileName + toString(processors-1) + ".temp"), tempBarcodePrimerComboFileNames, lines[processors-1]);
 883                 processIDS.push_back((processors-1));
 884
 885                 //Wait until all threads have terminated.
 886                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 887
 888                 //Close all thread handles and free memory allocations.
 889                 for(int i=0; i < pDataArray.size(); i++){
 890                         num += pDataArray[i]->count;
 891                         CloseHandle(hThreadArray[i]);
 892                         delete pDataArray[i];
 893                 }
 894
 895
 896 #endif
 897                 //append files
 898                 m->mothurOutEndLine();
 899                 for(int i=0;i<processIDS.size();i++){
 900
 901                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 902
 903                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
 904                         m->mothurRemove((trimFlowFileName + toString(processIDS[i]) + ".temp"));
 905 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
 906
 907                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
 908                         m->mothurRemove((scrapFlowFileName + toString(processIDS[i]) + ".temp"));
 909 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
 910
 911                         if(fasta){
 912                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
 913                                 m->mothurRemove((fastaFileName + toString(processIDS[i]) + ".temp"));
 914 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
 915                         }
 916                         if(allFiles){
 917                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
 918                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
 919                                                 m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
 920                                                 m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
 921                                         }
 922                                 }
 923                         }
 924                 }
 925
 926                 return exitCommand;
 927
 928         }
 929         catch(exception& e) {
 930                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
 931                 exit(1);
 932         }
 933 }
 934
 935 //***************************************************************************************************************