trimflowscommand.cpp

   1 /*
   2  *  trimflowscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 12/22/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "trimflowscommand.h"
  11 #include "needlemanoverlap.hpp"
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> TrimFlowsCommand::setParameters(){
  16         try {
  17                 CommandParameter pflow("flow", "InputTypes", "", "", "none", "none", "none","flow",false,true,true); parameters.push_back(pflow);
  18                 CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none","",false,false,true); parameters.push_back(poligos);
  19                 CommandParameter pmaxhomop("maxhomop", "Number", "", "9", "", "", "","",false,false); parameters.push_back(pmaxhomop);
  20                 CommandParameter pmaxflows("maxflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pmaxflows);
  21                 CommandParameter pminflows("minflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pminflows);
  22                 CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(ppdiffs);
  23                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(pbdiffs);
  24         CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pldiffs);
  25                 CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psdiffs);
  26         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  28                 CommandParameter psignal("signal", "Number", "", "0.50", "", "", "","",false,false); parameters.push_back(psignal);
  29                 CommandParameter pnoise("noise", "Number", "", "0.70", "", "", "","",false,false); parameters.push_back(pnoise);
  30                 CommandParameter pallfiles("allfiles", "Boolean", "", "t", "", "", "","",false,false); parameters.push_back(pallfiles);
  31                 CommandParameter porder("order", "String", "", "TACG", "", "", "","",false,false); parameters.push_back(porder);
  32                 CommandParameter pfasta("fasta", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pfasta);
  33                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  34                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  35
  36                 vector<string> myArray;
  37                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  38                 return myArray;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "TrimFlowsCommand", "setParameters");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 string TrimFlowsCommand::getHelpString(){
  47         try {
  48                 string helpString = "";
  49                 helpString += "The trim.flows command reads a flowgram file and creates .....\n";
  50                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  51                 helpString += "For more details please check out the wiki http://www.mothur.org/wiki/Trim.flows.\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "TrimFlowsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 string TrimFlowsCommand::getOutputPattern(string type) {
  61     try {
  62         string pattern = "";
  63
  64         if (type == "flow") {  pattern = "[filename],[tag],flow"; }
  65         else if (type == "fasta") {  pattern = "[filename],flow.fasta"; }
  66         else if (type == "file") {  pattern = "[filename],[tag],flow.files"; }
  67         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  68
  69         return pattern;
  70     }
  71     catch(exception& e) {
  72         m->errorOut(e, "TrimFlowsCommand", "getOutputPattern");
  73         exit(1);
  74     }
  75 }
  76 //**********************************************************************************************************************
  77
  78 TrimFlowsCommand::TrimFlowsCommand(){
  79         try {
  80                 abort = true; calledHelp = true;
  81                 setParameters();
  82                 vector<string> tempOutNames;
  83                 outputTypes["flow"] = tempOutNames;
  84                 outputTypes["fasta"] = tempOutNames;
  85         outputTypes["file"] = tempOutNames;
  86         }
  87         catch(exception& e) {
  88                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
  89                 exit(1);
  90         }
  91 }
  92 //**********************************************************************************************************************
  93
  94 TrimFlowsCommand::TrimFlowsCommand(string option)  {
  95         try {
  96
  97                 abort = false; calledHelp = false;
  98                 comboStarts = 0;
  99
 100                 //allow user to run help
 101                 if(option == "help") { help(); abort = true; calledHelp = true; }
 102                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 103
 104                 else {
 105
 106                         vector<string> myArray = setParameters();
 107
 108                         OptionParser parser(option);
 109                         map<string,string> parameters = parser.getParameters();
 110
 111                         ValidParameters validParameter;
 112                         map<string,string>::iterator it;
 113
 114                         //check to make sure all parameters are valid for command
 115                         for (it = parameters.begin(); it != parameters.end(); it++) {
 116                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 117                         }
 118
 119                         //initialize outputTypes
 120                         vector<string> tempOutNames;
 121                         outputTypes["flow"] = tempOutNames;
 122                         outputTypes["fasta"] = tempOutNames;
 123             outputTypes["file"] = tempOutNames;
 124
 125                         //if the user changes the input directory command factory will send this info to us in the output parameter
 126                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 127                         if (inputDir == "not found"){   inputDir = "";          }
 128                         else {
 129                                 string path;
 130                                 it = parameters.find("flow");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["flow"] = inputDir + it->second;             }
 136                                 }
 137
 138                                 it = parameters.find("oligos");
 139                                 //user has given a template file
 140                                 if(it != parameters.end()){
 141                                         path = m->hasPath(it->second);
 142                                         //if the user has not given a path then, add inputdir. else leave path alone.
 143                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 144                                 }
 145
 146                         }
 147
 148
 149                         //check for required parameters
 150                         flowFileName = validParameter.validFile(parameters, "flow", true);
 151                         if (flowFileName == "not found") {
 152                                 flowFileName = m->getFlowFile();
 153                                 if (flowFileName != "") {  m->mothurOut("Using " + flowFileName + " as input file for the flow parameter."); m->mothurOutEndLine(); }
 154                                 else {
 155                                         m->mothurOut("No valid current flow file. You must provide a flow file."); m->mothurOutEndLine();
 156                                         abort = true;
 157                                 }
 158                         }else if (flowFileName == "not open") { flowFileName = ""; abort = true; }
 159
 160                         //if the user changes the output directory command factory will send this info to us in the output parameter
 161                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 162                                 outputDir = "";
 163                                 outputDir += m->hasPath(flowFileName); //if user entered a file with a path then preserve it
 164                         }
 165
 166
 167                         //check for optional parameter and set defaults
 168                         // ...at some point should added some additional type checking...
 169
 170                         string temp;
 171                         temp = validParameter.validFile(parameters, "minflows", false); if (temp == "not found") { temp = "450"; }
 172                         m->mothurConvert(temp, minFlows);
 173
 174                         temp = validParameter.validFile(parameters, "maxflows", false); if (temp == "not found") { temp = "450"; }
 175                         m->mothurConvert(temp, maxFlows);
 176
 177
 178                         temp = validParameter.validFile(parameters, "oligos", true);
 179                         if (temp == "not found")        {       oligoFileName = "";             }
 180                         else if(temp == "not open")     {       abort = true;                   }
 181                         else                                            {       oligoFileName = temp;   m->setOligosFile(oligoFileName); }
 182
 183                         temp = validParameter.validFile(parameters, "fasta", false);            if (temp == "not found"){       fasta = 0;              }
 184                         else if(m->isTrue(temp))        {       fasta = 1;      }
 185
 186                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found"){       temp = "9";             }
 187                         m->mothurConvert(temp, maxHomoP);
 188
 189                         temp = validParameter.validFile(parameters, "signal", false);           if (temp == "not found"){       temp = "0.50";  }
 190                         m->mothurConvert(temp, signal);
 191
 192                         temp = validParameter.validFile(parameters, "noise", false);            if (temp == "not found"){       temp = "0.70";  }
 193                         m->mothurConvert(temp, noise);
 194
 195                         temp = validParameter.validFile(parameters, "bdiffs", false);           if (temp == "not found"){       temp = "0";             }
 196                         m->mothurConvert(temp, bdiffs);
 197
 198                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found"){       temp = "0";             }
 199                         m->mothurConvert(temp, pdiffs);
 200
 201             temp = validParameter.validFile(parameters, "ldiffs", false);               if (temp == "not found") { temp = "0"; }
 202                         m->mothurConvert(temp, ldiffs);
 203
 204             temp = validParameter.validFile(parameters, "sdiffs", false);               if (temp == "not found") { temp = "0"; }
 205                         m->mothurConvert(temp, sdiffs);
 206
 207                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs;  temp = toString(tempTotal); }
 208                         m->mothurConvert(temp, tdiffs);
 209
 210                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs + ldiffs + sdiffs;     }
 211
 212
 213                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 214                         m->setProcessors(temp);
 215                         m->mothurConvert(temp, processors);
 216
 217                         flowOrder = validParameter.validFile(parameters, "order", false);
 218                         if (flowOrder == "not found"){ flowOrder = "TACG";              }
 219                         else if(flowOrder.length() != 4){
 220                                 m->mothurOut("The value of the order option must be four bases long\n");
 221                         }
 222
 223                         if(oligoFileName == "") {       allFiles = 0;           }
 224                         else                                    {       allFiles = 1;           }
 225
 226                         numFPrimers = 0;
 227                         numRPrimers = 0;
 228             numLinkers = 0;
 229             numSpacers = 0;
 230                 }
 231
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
 235                 exit(1);
 236         }
 237 }
 238
 239 //***************************************************************************************************************
 240
 241 int TrimFlowsCommand::execute(){
 242         try{
 243
 244                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 245
 246         map<string, string> variables;
 247                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
 248         string fastaFileName = getOutputFileName("fasta",variables);
 249                 if(fasta){ outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName); }
 250
 251         variables["[tag]"] = "trim";
 252                 string trimFlowFileName = getOutputFileName("flow",variables);
 253                 outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
 254
 255         variables["[tag]"] = "scrap";
 256                 string scrapFlowFileName = getOutputFileName("flow",variables);
 257                 outputNames.push_back(scrapFlowFileName); outputTypes["flow"].push_back(scrapFlowFileName);
 258
 259
 260
 261                 vector<unsigned long long> flowFilePos;
 262         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 263                 flowFilePos = getFlowFileBreaks();
 264                 for (int i = 0; i < (flowFilePos.size()-1); i++) {
 265                         lines.push_back(new linePair(flowFilePos[i], flowFilePos[(i+1)]));
 266                 }
 267         #else
 268                 ifstream in; m->openInputFile(flowFileName, in); in >> numFlows; in.close();
 269         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 270                 processors = 1;
 271         ///////////////////////////////////////// until I fix multiple processors for windows //////////////////
 272                 if (processors == 1) {
 273                         lines.push_back(new linePair(0, 1000));
 274                 }else {
 275                         int numFlowLines;
 276                         flowFilePos = m->setFilePosEachLine(flowFileName, numFlowLines);
 277                         flowFilePos.erase(flowFilePos.begin() + 1); numFlowLines--;
 278
 279                         //figure out how many sequences you have to process
 280                         int numSeqsPerProcessor = numFlowLines / processors;
 281                         cout << numSeqsPerProcessor << '\t' << numFlowLines << endl;
 282                         for (int i = 0; i < processors; i++) {
 283                                 int startIndex =  i * numSeqsPerProcessor;
 284                                 if(i == (processors - 1)){      numSeqsPerProcessor = numFlowLines - i * numSeqsPerProcessor;   }
 285                                 lines.push_back(new linePair(flowFilePos[startIndex], numSeqsPerProcessor));
 286                                 cout << flowFilePos[startIndex] << '\t' << numSeqsPerProcessor << endl;
 287                         }
 288                 }
 289         #endif
 290
 291                 vector<vector<string> > barcodePrimerComboFileNames;
 292                 if(oligoFileName != ""){
 293                         getOligos(barcodePrimerComboFileNames);
 294                 }
 295
 296                 if(processors == 1){
 297                         driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 298                 }else{
 299                         createProcessesCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames);
 300                 }
 301
 302                 if (m->control_pressed) {  return 0; }
 303
 304                 string flowFilesFileName;
 305                 ofstream output;
 306
 307                 if(allFiles){
 308                         set<string> namesAlreadyProcessed;
 309             variables["[tag]"] = "";
 310                         flowFilesFileName = getOutputFileName("file",variables);
 311                         m->openOutputFile(flowFilesFileName, output);
 312
 313                         for(int i=0;i<barcodePrimerComboFileNames.size();i++){
 314                                 for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
 315                                         if (namesAlreadyProcessed.count(barcodePrimerComboFileNames[i][j]) == 0) {
 316                         if (barcodePrimerComboFileNames[i][j] != "") {
 317                             FILE * pFile;
 318                             unsigned long long size;
 319
 320                             //get num bytes in file
 321                             pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
 322                             if (pFile==NULL) perror ("Error opening file");
 323                             else{
 324                                 fseek (pFile, 0, SEEK_END);
 325                                 size=ftell(pFile);
 326                                 fclose (pFile);
 327                             }
 328
 329                             if(size < 10){
 330                                 m->mothurRemove(barcodePrimerComboFileNames[i][j]);
 331                             }
 332                             else{
 333                                 output << m->getFullPathName(barcodePrimerComboFileNames[i][j]) << endl;
 334                                 outputNames.push_back(barcodePrimerComboFileNames[i][j]);
 335                                 outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
 336                             }
 337                             namesAlreadyProcessed.insert(barcodePrimerComboFileNames[i][j]);
 338                         }
 339                                         }
 340                                 }
 341                         }
 342                         output.close();
 343                 }
 344                 else{
 345             variables["[tag]"] = "";
 346                         flowFilesFileName = getOutputFileName("file",variables);
 347                         m->openOutputFile(flowFilesFileName, output);
 348
 349                         output << m->getFullPathName(trimFlowFileName) << endl;
 350
 351                         output.close();
 352                 }
 353                 outputTypes["file"].push_back(flowFilesFileName);
 354                 outputNames.push_back(flowFilesFileName);
 355
 356                 m->mothurOutEndLine();
 357                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 358                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 359                 m->mothurOutEndLine();
 360
 361                 return 0;
 362         }
 363         catch(exception& e) {
 364                 m->errorOut(e, "TrimSeqsCommand", "execute");
 365                 exit(1);
 366         }
 367 }
 368
 369 //***************************************************************************************************************
 370
 371 int TrimFlowsCommand::driverCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > thisBarcodePrimerComboFileNames, linePair* line){
 372
 373         try {
 374                 ofstream trimFlowFile;
 375                 m->openOutputFile(trimFlowFileName, trimFlowFile);
 376                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 377
 378                 ofstream scrapFlowFile;
 379                 m->openOutputFile(scrapFlowFileName, scrapFlowFile);
 380                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 381
 382                 ofstream fastaFile;
 383                 if(fasta){      m->openOutputFile(fastaFileName, fastaFile);    }
 384
 385                 ifstream flowFile;
 386                 m->openInputFile(flowFileName, flowFile);
 387
 388                 flowFile.seekg(line->start);
 389
 390                 if(line->start == 0){
 391                         flowFile >> numFlows; m->gobble(flowFile);
 392                         scrapFlowFile << maxFlows << endl;
 393                         trimFlowFile << maxFlows << endl;
 394                         if(allFiles){
 395                                 for(int i=0;i<thisBarcodePrimerComboFileNames.size();i++){
 396                                         for(int j=0;j<thisBarcodePrimerComboFileNames[0].size();j++){
 397                         if (thisBarcodePrimerComboFileNames[i][j] != "") {
 398                             ofstream temp;
 399                             m->openOutputFile(thisBarcodePrimerComboFileNames[i][j], temp);
 400                             temp << maxFlows << endl;
 401                             temp.close();
 402                         }
 403                                         }
 404                                 }
 405                         }
 406                 }
 407
 408                 FlowData flowData(numFlows, signal, noise, maxHomoP, flowOrder);
 409                 //cout << " driver flowdata address " <<  &flowData  << &flowFile << endl;
 410                 int count = 0;
 411                 bool moreSeqs = 1;
 412
 413                 TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer);
 414
 415                 while(moreSeqs) {
 416
 417                         if (m->control_pressed) { break; }
 418
 419                         int success = 1;
 420                         int currentSeqDiffs = 0;
 421                         string trashCode = "";
 422
 423                         flowData.getNext(flowFile);
 424                         flowData.capFlows(maxFlows);
 425
 426                         Sequence currSeq = flowData.getSequence();
 427                         if(!flowData.hasMinFlows(minFlows)){    //screen to see if sequence is of a minimum number of flows
 428                                 success = 0;
 429                                 trashCode += 'l';
 430                         }
 431
 432                         int primerIndex = 0;
 433                         int barcodeIndex = 0;
 434
 435             if(numLinkers != 0){
 436                 success = trimOligos.stripLinker(currSeq);
 437                 if(success > ldiffs)            {       trashCode += 'k';       }
 438                 else{ currentSeqDiffs += success;  }
 439
 440             }
 441
 442             if (m->debug) { m->mothurOut("[DEBUG]: " + currSeq.getName() + " " + currSeq.getUnaligned() + "\n"); }
 443
 444                         if(barcodes.size() != 0){
 445                                 success = trimOligos.stripBarcode(currSeq, barcodeIndex);
 446                                 if(success > bdiffs)            {       trashCode += 'b';       }
 447                                 else{ currentSeqDiffs += success;  }
 448                         }
 449
 450             if(numSpacers != 0){
 451                 success = trimOligos.stripSpacer(currSeq);
 452                 if(success > sdiffs)            {       trashCode += 's';       }
 453                 else{ currentSeqDiffs += success;  }
 454
 455             }
 456
 457                         if(numFPrimers != 0){
 458                                 success = trimOligos.stripForward(currSeq, primerIndex);
 459                                 if(success > pdiffs)            {       trashCode += 'f';       }
 460                                 else{ currentSeqDiffs += success;  }
 461                         }
 462
 463                         if (currentSeqDiffs > tdiffs)   {       trashCode += 't';   }
 464
 465                         if(numRPrimers != 0){
 466                                 success = trimOligos.stripReverse(currSeq);
 467                                 if(!success)                            {       trashCode += 'r';       }
 468                         }
 469
 470                         if(trashCode.length() == 0){
 471                 string thisGroup = "";
 472                 if(barcodes.size() != 0){
 473                     thisGroup = barcodeNameVector[barcodeIndex];
 474                     if (primers.size() != 0) {
 475                         if (primerNameVector[primerIndex] != "") {
 476                             if(thisGroup != "") {
 477                                 thisGroup += "." + primerNameVector[primerIndex];
 478                             }else {
 479                                 thisGroup = primerNameVector[primerIndex];
 480                             }
 481                         }
 482                     }
 483                 }
 484
 485                 int pos = thisGroup.find("ignore");
 486                 if (pos == string::npos) {
 487                     flowData.printFlows(trimFlowFile);
 488
 489                     if(fasta)   {       currSeq.printSequence(fastaFile);       }
 490
 491                     if(allFiles){
 492                         ofstream output;
 493                         m->openOutputFileAppend(thisBarcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 494                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 495
 496                         flowData.printFlows(output);
 497                         output.close();
 498                     }
 499                 }
 500                         }
 501                         else{
 502                                 flowData.printFlows(scrapFlowFile, trashCode);
 503                         }
 504
 505                         count++;
 506                         //cout << "driver" << '\t' << currSeq.getName() << endl;
 507                         //report progress
 508                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 509
 510 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 511                         unsigned long long pos = flowFile.tellg();
 512
 513                         if ((pos == -1) || (pos >= line->end)) { break; }
 514 #else
 515                         if (flowFile.eof()) { break; }
 516 #endif
 517
 518                 }
 519                 //report progress
 520                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 521
 522                 trimFlowFile.close();
 523                 scrapFlowFile.close();
 524                 flowFile.close();
 525                 if(fasta){      fastaFile.close();      }
 526
 527                 return count;
 528         }
 529         catch(exception& e) {
 530                 m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
 531                 exit(1);
 532         }
 533 }
 534
 535 //***************************************************************************************************************
 536
 537 void TrimFlowsCommand::getOligos(vector<vector<string> >& outFlowFileNames){
 538         try {
 539                 ifstream oligosFile;
 540                 m->openInputFile(oligoFileName, oligosFile);
 541
 542                 string type, oligo, group;
 543
 544                 int indexPrimer = 0;
 545                 int indexBarcode = 0;
 546
 547                 while(!oligosFile.eof()){
 548
 549                         oligosFile >> type; m->gobble(oligosFile);      //get the first column value of the row - is it a comment or a feature we are interested in?
 550
 551                         if(type[0] == '#'){     //igore the line because there's a comment
 552                                 while (!oligosFile.eof())       {       char c = oligosFile.get(); if (c == 10 || c == 13){     break;  }       } // get rest of line if there's any crap there
 553                         }
 554                         else{                           //there's a feature we're interested in
 555
 556                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }                                  //make type case insensitive
 557
 558                                 oligosFile >> oligo;    //get the DNA sequence for the feature
 559
 560                                 for(int i=0;i<oligo.length();i++){      //make type case insensitive and change any U's to T's
 561                                         oligo[i] = toupper(oligo[i]);
 562                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 563                                 }
 564
 565                                 if(type == "FORWARD"){  //if the feature is a forward primer...
 566                                         group = "";
 567
 568                                         while (!oligosFile.eof())       {       // get rest of line in case there is a primer name = will have the name of the primer
 569                                                 char c = oligosFile.get();
 570                                                 if (c == 10 || c == 13){        break;  }
 571                                                 else if (c == 32 || c == 9){;} //space or tab
 572                                                 else {  group += c;  }
 573                                         }
 574
 575                                         //have we seen this primer already?
 576                                         map<string, int>::iterator itPrimer = primers.find(oligo);
 577                                         if (itPrimer != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 578
 579                                         primers[oligo]=indexPrimer; indexPrimer++;
 580                                         primerNameVector.push_back(group);
 581
 582                                 }
 583                                 else if(type == "REVERSE"){
 584                                         string oligoRC = reverseOligo(oligo);
 585                                         revPrimer.push_back(oligoRC);
 586                                 }
 587                                 else if(type == "BARCODE"){
 588                                         oligosFile >> group;
 589
 590                                         //check for repeat barcodes
 591                                         map<string, int>::iterator itBar = barcodes.find(oligo);
 592                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
 593
 594                                         barcodes[oligo]=indexBarcode; indexBarcode++;
 595                                         barcodeNameVector.push_back(group);
 596                                 }else if(type == "LINKER"){
 597                                         linker.push_back(oligo);
 598                                 }else if(type == "SPACER"){
 599                                         spacer.push_back(oligo);
 600                                 }
 601                                 else{
 602                                         m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
 603                                 }
 604                         }
 605
 606                         m->gobble(oligosFile);
 607                 }
 608                 oligosFile.close();
 609
 610                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0;   }
 611
 612                 //add in potential combos
 613                 if(barcodeNameVector.size() == 0){
 614                         barcodes[""] = 0;
 615                         barcodeNameVector.push_back("");
 616                 }
 617
 618                 if(primerNameVector.size() == 0){
 619                         primers[""] = 0;
 620                         primerNameVector.push_back("");
 621                 }
 622
 623
 624                 outFlowFileNames.resize(barcodeNameVector.size());
 625                 for(int i=0;i<outFlowFileNames.size();i++){
 626                         outFlowFileNames[i].assign(primerNameVector.size(), "");
 627                 }
 628
 629                 if(allFiles){
 630
 631                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
 632                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
 633
 634                                         string primerName = primerNameVector[itPrimer->second];
 635                                         string barcodeName = barcodeNameVector[itBar->second];
 636
 637                                         if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
 638                                         else {
 639                         string comboGroupName = "";
 640                         string fileName = "";
 641
 642                         map<string, string> variables;
 643                         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
 644
 645                         if(primerName == ""){
 646                             comboGroupName = barcodeNameVector[itBar->second];
 647                             variables["[tag]"] = comboGroupName;
 648                             fileName = getOutputFileName("flow", variables);
 649                         }
 650                         else{
 651                             if(barcodeName == ""){
 652                                 comboGroupName = primerNameVector[itPrimer->second];
 653                             }
 654                             else{
 655                                 comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
 656                             }
 657                             variables["[tag]"] = comboGroupName;
 658                             fileName = getOutputFileName("flow", variables);
 659                         }
 660
 661                         outFlowFileNames[itBar->second][itPrimer->second] = fileName;
 662
 663                         ofstream temp;
 664                         m->openOutputFile(fileName, temp);
 665                         temp.close();
 666                     }
 667                                 }
 668                         }
 669                 }
 670
 671                 numFPrimers = primers.size();
 672                 numRPrimers = revPrimer.size();
 673         numLinkers = linker.size();
 674         numSpacers = spacer.size();
 675
 676         }
 677         catch(exception& e) {
 678                 m->errorOut(e, "TrimSeqsCommand", "getOligos");
 679                 exit(1);
 680         }
 681 }
 682 //********************************************************************/
 683 string TrimFlowsCommand::reverseOligo(string oligo){
 684         try {
 685         string reverse = "";
 686
 687         for(int i=oligo.length()-1;i>=0;i--){
 688
 689             if(oligo[i] == 'A')         {       reverse += 'T'; }
 690             else if(oligo[i] == 'T'){   reverse += 'A'; }
 691             else if(oligo[i] == 'U'){   reverse += 'A'; }
 692
 693             else if(oligo[i] == 'G'){   reverse += 'C'; }
 694             else if(oligo[i] == 'C'){   reverse += 'G'; }
 695
 696             else if(oligo[i] == 'R'){   reverse += 'Y'; }
 697             else if(oligo[i] == 'Y'){   reverse += 'R'; }
 698
 699             else if(oligo[i] == 'M'){   reverse += 'K'; }
 700             else if(oligo[i] == 'K'){   reverse += 'M'; }
 701
 702             else if(oligo[i] == 'W'){   reverse += 'W'; }
 703             else if(oligo[i] == 'S'){   reverse += 'S'; }
 704
 705             else if(oligo[i] == 'B'){   reverse += 'V'; }
 706             else if(oligo[i] == 'V'){   reverse += 'B'; }
 707
 708             else if(oligo[i] == 'D'){   reverse += 'H'; }
 709             else if(oligo[i] == 'H'){   reverse += 'D'; }
 710
 711             else                                                {       reverse += 'N'; }
 712         }
 713
 714
 715         return reverse;
 716     }
 717         catch(exception& e) {
 718                 m->errorOut(e, "TrimFlowsCommand", "reverseOligo");
 719                 exit(1);
 720         }
 721 }
 722
 723 /**************************************************************************************************/
 724 vector<unsigned long long> TrimFlowsCommand::getFlowFileBreaks() {
 725
 726         try{
 727
 728                 vector<unsigned long long> filePos;
 729                 filePos.push_back(0);
 730
 731                 FILE * pFile;
 732                 unsigned long long size;
 733
 734                 //get num bytes in file
 735                 pFile = fopen (flowFileName.c_str(),"rb");
 736                 if (pFile==NULL) perror ("Error opening file");
 737                 else{
 738                         fseek (pFile, 0, SEEK_END);
 739                         size=ftell (pFile);
 740                         fclose (pFile);
 741                 }
 742
 743                 //estimate file breaks
 744                 unsigned long long chunkSize = 0;
 745                 chunkSize = size / processors;
 746
 747                 //file too small to divide by processors
 748                 if (chunkSize == 0)  {  processors = 1; filePos.push_back(size); return filePos;        }
 749
 750                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 751                 for (int i = 0; i < processors; i++) {
 752                         unsigned long long spot = (i+1) * chunkSize;
 753
 754                         ifstream in;
 755                         m->openInputFile(flowFileName, in);
 756                         in.seekg(spot);
 757
 758                         string dummy = m->getline(in);
 759
 760                         //there was not another sequence before the end of the file
 761                         unsigned long long sanityPos = in.tellg();
 762
 763 //                      if (sanityPos == -1) {  break;  }
 764 //                      else {  filePos.push_back(newSpot);  }
 765                         if (sanityPos == -1) {  break;  }
 766                         else {  filePos.push_back(sanityPos);  }
 767
 768                         in.close();
 769                 }
 770
 771                 //save end pos
 772                 filePos.push_back(size);
 773
 774                 //sanity check filePos
 775                 for (int i = 0; i < (filePos.size()-1); i++) {
 776                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
 777                 }
 778
 779                 ifstream in;
 780                 m->openInputFile(flowFileName, in);
 781                 in >> numFlows;
 782                 m->gobble(in);
 783                 //unsigned long long spot = in.tellg();
 784                 //filePos[0] = spot;
 785                 in.close();
 786
 787                 processors = (filePos.size() - 1);
 788
 789                 return filePos;
 790         }
 791         catch(exception& e) {
 792                 m->errorOut(e, "TrimSeqsCommand", "getFlowFileBreaks");
 793                 exit(1);
 794         }
 795 }
 796
 797 /**************************************************************************************************/
 798
 799 int TrimFlowsCommand::createProcessesCreateTrim(string flowFileName, string trimFlowFileName, string scrapFlowFileName, string fastaFileName, vector<vector<string> > barcodePrimerComboFileNames){
 800
 801         try {
 802                 processIDS.clear();
 803                 int exitCommand = 1;
 804
 805 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 806                 int process = 1;
 807
 808                 //loop through and create all the processes you want
 809                 while (process != processors) {
 810                         int pid = fork();
 811
 812                         if (pid > 0) {
 813                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 814                                 process++;
 815                         }else if (pid == 0){
 816
 817                                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 818                                 if(allFiles){
 819                                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 820                                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 821                             if (tempBarcodePrimerComboFileNames[i][j] != "") {
 822                                 tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
 823                                 ofstream temp;
 824                                 m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 825                                 temp.close();
 826                             }
 827                                                 }
 828                                         }
 829                                 }
 830                                 driverCreateTrim(flowFileName,
 831                                                                  (trimFlowFileName + toString(getpid()) + ".temp"),
 832                                                                  (scrapFlowFileName + toString(getpid()) + ".temp"),
 833                                                                  (fastaFileName + toString(getpid()) + ".temp"),
 834                                                                  tempBarcodePrimerComboFileNames, lines[process]);
 835
 836                                 exit(0);
 837                         }else {
 838                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 839                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 840                                 exit(0);
 841                         }
 842                 }
 843
 844                 //parent do my part
 845                 ofstream temp;
 846                 m->openOutputFile(trimFlowFileName, temp);
 847                 temp.close();
 848
 849                 m->openOutputFile(scrapFlowFileName, temp);
 850                 temp.close();
 851
 852                 if(fasta){
 853                         m->openOutputFile(fastaFileName, temp);
 854                         temp.close();
 855                 }
 856
 857                 driverCreateTrim(flowFileName, trimFlowFileName, scrapFlowFileName, fastaFileName, barcodePrimerComboFileNames, lines[0]);
 858
 859                 //force parent to wait until all the processes are done
 860                 for (int i=0;i<processIDS.size();i++) {
 861                         int temp = processIDS[i];
 862                         wait(&temp);
 863                 }
 864 #else
 865                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 866                 //Windows version shared memory, so be careful when passing variables through the trimFlowData struct.
 867                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 868                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 869
 870                 vector<trimFlowData*> pDataArray;
 871                 DWORD   dwThreadIdArray[processors-1];
 872                 HANDLE  hThreadArray[processors-1];
 873
 874                 //Create processor worker threads.
 875                 for( int i=0; i<processors-1; i++ ){
 876                         // Allocate memory for thread data.
 877                         string extension = "";
 878                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
 879
 880                         vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 881                         if(allFiles){
 882                                 for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 883                                         for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 884                         if (tempBarcodePrimerComboFileNames[i][j] != "") {
 885                             tempBarcodePrimerComboFileNames[i][j] += extension;
 886                             ofstream temp;
 887                             m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 888                             temp.close();
 889                                                 }
 890                                         }
 891                                 }
 892                         }
 893
 894                         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);
 895                         pDataArray.push_back(tempflow);
 896
 897                         //MyTrimFlowThreadFunction is in header. It must be global or static to work with the threads.
 898                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 899                         hThreadArray[i] = CreateThread(NULL, 0, MyTrimFlowThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 900                 }
 901
 902                 //using the main process as a worker saves time and memory
 903                 ofstream temp;
 904                 m->openOutputFile(trimFlowFileName, temp);
 905                 temp.close();
 906
 907                 m->openOutputFile(scrapFlowFileName, temp);
 908                 temp.close();
 909
 910                 if(fasta){
 911                         m->openOutputFile(fastaFileName, temp);
 912                         temp.close();
 913                 }
 914
 915                 vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
 916                 if(allFiles){
 917                         for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
 918                                 for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
 919                     if (tempBarcodePrimerComboFileNames[i][j] != "") {
 920                         tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
 921                         ofstream temp;
 922                         m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
 923                         temp.close();
 924                     }
 925
 926                                 }
 927                         }
 928                 }
 929
 930                 //do my part - do last piece because windows is looking for eof
 931                 int num = driverCreateTrim(flowFileName, (trimFlowFileName  + toString(processors-1) + ".temp"), (scrapFlowFileName  + toString(processors-1) + ".temp"), (fastaFileName + toString(processors-1) + ".temp"), tempBarcodePrimerComboFileNames, lines[processors-1]);
 932                 processIDS.push_back((processors-1));
 933
 934                 //Wait until all threads have terminated.
 935                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 936
 937                 //Close all thread handles and free memory allocations.
 938                 for(int i=0; i < pDataArray.size(); i++){
 939                         num += pDataArray[i]->count;
 940                         CloseHandle(hThreadArray[i]);
 941                         delete pDataArray[i];
 942                 }
 943
 944
 945 #endif
 946                 //append files
 947                 m->mothurOutEndLine();
 948                 for(int i=0;i<processIDS.size();i++){
 949
 950                         m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
 951
 952                         m->appendFiles((trimFlowFileName + toString(processIDS[i]) + ".temp"), trimFlowFileName);
 953                         m->mothurRemove((trimFlowFileName + toString(processIDS[i]) + ".temp"));
 954 //                      m->mothurOut("\tDone with trim.flow file"); m->mothurOutEndLine();
 955
 956                         m->appendFiles((scrapFlowFileName + toString(processIDS[i]) + ".temp"), scrapFlowFileName);
 957                         m->mothurRemove((scrapFlowFileName + toString(processIDS[i]) + ".temp"));
 958 //                      m->mothurOut("\tDone with scrap.flow file"); m->mothurOutEndLine();
 959
 960                         if(fasta){
 961                                 m->appendFiles((fastaFileName + toString(processIDS[i]) + ".temp"), fastaFileName);
 962                                 m->mothurRemove((fastaFileName + toString(processIDS[i]) + ".temp"));
 963 //                              m->mothurOut("\tDone with flow.fasta file"); m->mothurOutEndLine();
 964                         }
 965                         if(allFiles){
 966                                 for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
 967                                         for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
 968                         if (barcodePrimerComboFileNames[j][k] != "") {
 969                             m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
 970                             m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
 971                         }
 972                                         }
 973                                 }
 974                         }
 975                 }
 976
 977                 return exitCommand;
 978
 979         }
 980         catch(exception& e) {
 981                 m->errorOut(e, "TrimFlowsCommand", "createProcessesCreateTrim");
 982                 exit(1);
 983         }
 984 }
 985
 986 //***************************************************************************************************************