shhhseqscommand.cpp

   1 /*
   2  *  shhhseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/8/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "shhhseqscommand.h"
  11
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> ShhhSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23                 CommandParameter psigma("sigma", "Number", "", "0.01", "", "", "",false,false); parameters.push_back(psigma);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ShhhSeqsCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ShhhSeqsCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The shhh.seqs command reads a fasta and name file and ....\n";
  39                 helpString += "The shhh.seqs command parameters are fasta, name, group, sigma and processors.\n";
  40                 helpString += "The fasta parameter allows you to enter the fasta file containing your sequences, and is required, unless you have a valid current fasta file. \n";
  41                 helpString += "The name parameter allows you to provide a name file associated with your fasta file. It is required. \n";
  42                 helpString += "The group parameter allows you to provide a group file.  When checking sequences, only sequences from the same group as the query sequence will be used as the reference. \n";
  43                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  44                 helpString += "The sigma parameter ....  The default is 0.01. \n";
  45                 helpString += "The shhh.seqs command should be in the following format: \n";
  46                 helpString += "shhh.seqs(fasta=yourFastaFile, name=yourNameFile) \n";
  47                 helpString += "Example: shhh.seqs(fasta=AD.align, name=AD.names) \n";
  48                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  49                 return helpString;
  50
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "ShhhSeqsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58
  59 ShhhSeqsCommand::ShhhSeqsCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63                 vector<string> tempOutNames;
  64                 outputTypes["fasta"] = tempOutNames;
  65                 outputTypes["name"] = tempOutNames;
  66                 outputTypes["map"] = tempOutNames;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "ShhhSeqsCommand", "ShhhSeqsCommand");
  70                 exit(1);
  71         }
  72 }
  73
  74 //**********************************************************************************************************************
  75 ShhhSeqsCommand::ShhhSeqsCommand(string option) {
  76         try {
  77                 abort = false; calledHelp = false;
  78
  79                 //allow user to run help
  80                 if(option == "help") { help(); abort = true; calledHelp = true; }
  81                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  82
  83                 else {
  84                         vector<string> myArray = setParameters();
  85
  86                         OptionParser parser(option);
  87                         map<string, string> parameters = parser.getParameters();
  88
  89                         ValidParameters validParameter;
  90                         map<string, string>::iterator it;
  91
  92                         //check to make sure all parameters are valid for command
  93                         for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  94                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  95                         }
  96
  97                         //initialize outputTypes
  98                         vector<string> tempOutNames;
  99                         outputTypes["fasta"] = tempOutNames;
 100                         outputTypes["name"] = tempOutNames;
 101                         outputTypes["map"] = tempOutNames;
 102
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 string path;
 108                                 it = parameters.find("fasta");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 114                                 }
 115
 116                                 it = parameters.find("name");
 117                                 //user has given a template file
 118                                 if(it != parameters.end()){
 119                                         path = m->hasPath(it->second);
 120                                         //if the user has not given a path then, add inputdir. else leave path alone.
 121                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 122                                 }
 123
 124                                 it = parameters.find("group");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 130                                 }
 131                         }
 132
 133                         //check for required parameters
 134                         fastafile = validParameter.validFile(parameters, "fasta", true);
 135                         if (fastafile == "not found") {
 136                                 fastafile = m->getFastaFile();
 137                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 138                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 139                         }
 140                         else if (fastafile == "not open") { abort = true; }
 141                         else { m->setFastaFile(fastafile); }
 142
 143                         //if the user changes the output directory command factory will send this info to us in the output parameter
 144                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 145
 146                         //check for optional parameter and set defaults
 147                         // ...at some point should added some additional type checking...
 148                         namefile = validParameter.validFile(parameters, "name", true);
 149                         if (namefile == "not found") {
 150                                 namefile = m->getNameFile();
 151                                 if (namefile != "") { m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 152                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 153                         }
 154                         else if (namefile == "not open") { namefile =  ""; abort = true; }
 155                         else {  m->setNameFile(namefile); }
 156
 157                         groupfile = validParameter.validFile(parameters, "group", true);
 158                         if (groupfile == "not found") { groupfile =  "";   }
 159                         else if (groupfile == "not open") { abort = true; groupfile =  ""; }
 160                         else {   m->setGroupFile(groupfile);  }
 161
 162                         string temp     = validParameter.validFile(parameters, "sigma", false);         if(temp == "not found"){        temp = "0.01"; }
 163                         m->mothurConvert(temp, sigma);
 164
 165                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 166                         m->setProcessors(temp);
 167                         m->mothurConvert(temp, processors);
 168
 169                         if (namefile == "") {
 170                                 vector<string> files; files.push_back(fastafile);
 171                                 parser.getNameFile(files);
 172                         }
 173                 }
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "ShhhSeqsCommand", "ShhhSeqsCommand");
 177                 exit(1);
 178         }
 179 }
 180 //**********************************************************************************************************************
 181 int ShhhSeqsCommand::execute() {
 182         try {
 183
 184                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 185
 186                 if (outputDir == "") { outputDir = m->hasPath(fastafile);  }//if user entered a file with a path then preserve it
 187                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "shhh.fasta";
 188                 string nameFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.names";
 189                 string mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.map";
 190
 191                 if (groupfile != "") {
 192                         //Parse sequences by group
 193                         SequenceParser parser(groupfile, fastafile, namefile);
 194                         vector<string> groups = parser.getNamesOfGroups();
 195
 196                         if (m->control_pressed) {  return 0; }
 197
 198                         //clears files
 199                         ofstream out, out1, out2;
 200                         m->openOutputFile(outputFileName, out); out.close();
 201                         m->openOutputFile(nameFileName, out1); out1.close();
 202                         mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.";
 203
 204                         vector<string> mapFileNames;
 205                         if(processors == 1)     {       mapFileNames = driverGroups(parser, outputFileName, nameFileName, mapFileName, 0, groups.size(), groups);       }
 206                         else                            {       mapFileNames = createProcessesGroups(parser, outputFileName, nameFileName, mapFileName, groups);                        }
 207
 208                         if (m->control_pressed) {    return 0;  }
 209
 210                         for (int j = 0; j < mapFileNames.size(); j++) { outputNames.push_back(mapFileNames[j]); outputTypes["map"].push_back(mapFileNames[j]); }
 211
 212                         //deconvolute results by running unique.seqs
 213                         deconvoluteResults(outputFileName, nameFileName);
 214
 215                         if (m->control_pressed) {   return 0;   }
 216
 217                 }else{
 218                         vector<string> sequences;
 219                         vector<string> uniqueNames;
 220                         vector<string> redundantNames;
 221                         vector<int> seqFreq;
 222
 223                         seqNoise noise;
 224                         correctDist* correct = new correctDist(processors);
 225
 226                         //reads fasta and name file and loads them in order
 227                         readData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq);
 228                         if (m->control_pressed) { return 0; }
 229
 230                         //calc distances for cluster
 231                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "shhh.dist";
 232                         correct->execute(distFileName);
 233                         delete correct;
 234
 235                         if (m->control_pressed) { m->mothurRemove(distFileName); return 0; }
 236
 237                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, outputFileName, nameFileName, mapFileName);
 238                         outputNames.push_back(mapFileName); outputTypes["map"].push_back(mapFileName);
 239                 }
 240
 241                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 242
 243                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 244                 outputNames.push_back(nameFileName); outputTypes["name"].push_back(nameFileName);
 245
 246                 m->mothurOutEndLine();
 247                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 248                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 249                 m->mothurOutEndLine();
 250
 251                 //set accnos file as new current accnosfile
 252                 string current = "";
 253                 itTypes = outputTypes.find("fasta");
 254                 if (itTypes != outputTypes.end()) {
 255                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 256                 }
 257
 258                 itTypes = outputTypes.find("name");
 259                 if (itTypes != outputTypes.end()) {
 260                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 261                 }
 262
 263
 264                 return 0;
 265         }
 266         catch(exception& e) {
 267                 m->errorOut(e, "ShhhSeqsCommand", "execute");
 268                 exit(1);
 269         }
 270 }
 271 //**********************************************************************************************************************
 272 int ShhhSeqsCommand::readData(correctDist* correct, seqNoise& noise, vector<string>& seqs, vector<string>& uNames, vector<string>& rNames, vector<int>& freq) {
 273         try {
 274                 map<string, string> nameMap;
 275                 map<string, string>::iterator it;
 276                 m->readNames(namefile, nameMap);
 277                 bool error = false;
 278
 279                 ifstream in;
 280                 m->openInputFile(fastafile, in);
 281
 282                 while (!in.eof()) {
 283
 284                         if (m->control_pressed) { in.close(); return 0; }
 285
 286                         Sequence seq(in); m->gobble(in);
 287
 288                         if (seq.getName() != "") {
 289                                 correct->addSeq(seq.getName(), seq.getAligned());
 290
 291                                 it = nameMap.find(seq.getName());
 292                                 if (it != nameMap.end()) {
 293                                         noise.addSeq(seq.getAligned(), seqs);
 294                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 295                                 }else {
 296                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your namefile, please correct.");
 297                                         error = true;
 298                                 }
 299                         }
 300                 }
 301                 in.close();
 302
 303                 if (error) { m->control_pressed = true; }
 304
 305                 return seqs.size();
 306
 307         }catch(exception& e) {
 308                 m->errorOut(e, "ShhhSeqsCommand", "readData");
 309                 exit(1);
 310         }
 311 }
 312 //**********************************************************************************************************************
 313 int ShhhSeqsCommand::loadData(correctDist* correct, seqNoise& noise, vector<string>& seqs, vector<string>& uNames, vector<string>& rNames, vector<int>& freq, map<string, string>& nameMap, vector<Sequence>& sequences) {
 314         try {
 315                 bool error = false;
 316                 map<string, string>::iterator it;
 317
 318                 for (int i = 0; i < sequences.size(); i++) {
 319
 320                         if (m->control_pressed) { return 0; }
 321
 322                         if (sequences[i].getName() != "") {
 323                                 correct->addSeq(sequences[i].getName(), sequences[i].getAligned());
 324
 325                                 it = nameMap.find(sequences[i].getName());
 326                                 if (it != nameMap.end()) {
 327                                         noise.addSeq(sequences[i].getAligned(), seqs);
 328                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 329                                 }else {
 330                                         m->mothurOut("[ERROR]: " + sequences[i].getName() + " is in your fasta file and not in your namefile, please correct.");
 331                                         error = true;
 332                                 }
 333                         }
 334                 }
 335
 336                 if (error) { m->control_pressed = true; }
 337
 338                 return seqs.size();
 339
 340         }catch(exception& e) {
 341                 m->errorOut(e, "ShhhSeqsCommand", "loadData");
 342                 exit(1);
 343         }
 344 }
 345 /**************************************************************************************************/
 346 vector<string> ShhhSeqsCommand::createProcessesGroups(SequenceParser& parser, string newFName, string newNName, string newMName, vector<string> groups) {
 347         try {
 348
 349                 vector<int> processIDS;
 350                 int process = 1;
 351                 vector<string> mapfileNames;
 352
 353                 //sanity check
 354                 if (groups.size() < processors) { processors = groups.size(); }
 355
 356                 //divide the groups between the processors
 357                 vector<linePair> lines;
 358                 int numGroupsPerProcessor = groups.size() / processors;
 359                 for (int i = 0; i < processors; i++) {
 360                         int startIndex =  i * numGroupsPerProcessor;
 361                         int endIndex = (i+1) * numGroupsPerProcessor;
 362                         if(i == (processors - 1)){      endIndex = groups.size();       }
 363                         lines.push_back(linePair(startIndex, endIndex));
 364                 }
 365
 366 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 367
 368                 //loop through and create all the processes you want
 369                 while (process != processors) {
 370                         int pid = fork();
 371
 372                         if (pid > 0) {
 373                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 374                                 process++;
 375                         }else if (pid == 0){
 376                                 mapfileNames = driverGroups(parser, newFName + toString(getpid()) + ".temp", newNName + toString(getpid()) + ".temp", newMName, lines[process].start, lines[process].end, groups);
 377
 378                                 //pass filenames to parent
 379                                 ofstream out;
 380                                 string tempFile = newMName + toString(getpid()) + ".temp";
 381                                 m->openOutputFile(tempFile, out);
 382                                 out << mapfileNames.size() << endl;
 383                                 for (int i = 0; i < mapfileNames.size(); i++) {
 384                                         out << mapfileNames[i] << endl;
 385                                 }
 386                                 out.close();
 387
 388                                 exit(0);
 389                         }else {
 390                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 391                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 392                                 exit(0);
 393                         }
 394                 }
 395
 396                 //do my part
 397                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 398
 399                 //force parent to wait until all the processes are done
 400                 for (int i=0;i<processIDS.size();i++) {
 401                         int temp = processIDS[i];
 402                         wait(&temp);
 403                 }
 404
 405                 //append output files
 406                 for(int i=0;i<processIDS.size();i++){
 407                         ifstream in;
 408                         string tempFile =  newMName + toString(processIDS[i]) + ".temp";
 409                         m->openInputFile(tempFile, in);
 410                         if (!in.eof()) {
 411                                 int tempNum = 0; in >> tempNum;  m->gobble(in);
 412                                 for (int j = 0; j < tempNum; j++) {
 413                                         string filename;
 414                                         in >> filename; m->gobble(in);
 415                                         mapfileNames.push_back(filename);
 416                                 }
 417                         }
 418                         in.close(); m->mothurRemove(tempFile);
 419
 420                 }
 421 #else
 422
 423                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 424                 //Windows version shared memory, so be careful when passing variables through the shhhseqsData struct.
 425                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 426                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 427
 428                 vector<shhhseqsData*> pDataArray;
 429                 DWORD   dwThreadIdArray[processors-1];
 430                 HANDLE  hThreadArray[processors-1];
 431
 432                 //Create processor worker threads.
 433                 for( int i=1; i<processors; i++ ){
 434                         // Allocate memory for thread data.
 435                         string extension = toString(i) + ".temp";
 436
 437                         shhhseqsData* tempShhhseqs = new shhhseqsData(fastafile, namefile, groupfile, (newFName+extension), (newNName+extension), newMName, groups, m, lines[i].start, lines[i].end, sigma, i);
 438                         pDataArray.push_back(tempShhhseqs);
 439                         processIDS.push_back(i);
 440
 441                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 442                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 443                         hThreadArray[i-1] = CreateThread(NULL, 0, MyShhhSeqsThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 444                 }
 445
 446
 447                 //using the main process as a worker saves time and memory
 448                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 449
 450                 //Wait until all threads have terminated.
 451                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 452
 453                 //Close all thread handles and free memory allocations.
 454                 for(int i=0; i < pDataArray.size(); i++){
 455                         for (int j = 0; j < pDataArray[i]->mapfileNames.size(); j++) {
 456                                 mapfileNames.push_back(pDataArray[i]->mapfileNames[j]);
 457                         }
 458                         CloseHandle(hThreadArray[i]);
 459                         delete pDataArray[i];
 460                 }
 461
 462 #endif
 463
 464                 //append output files
 465                 for(int i=0;i<processIDS.size();i++){
 466                         m->appendFiles((newFName + toString(processIDS[i]) + ".temp"), newFName);
 467                         m->mothurRemove((newFName + toString(processIDS[i]) + ".temp"));
 468
 469                         m->appendFiles((newNName + toString(processIDS[i]) + ".temp"), newNName);
 470                         m->mothurRemove((newNName + toString(processIDS[i]) + ".temp"));
 471                 }
 472
 473                 return mapfileNames;
 474
 475         }
 476         catch(exception& e) {
 477                 m->errorOut(e, "ShhhSeqsCommand", "createProcessesGroups");
 478                 exit(1);
 479         }
 480 }
 481 /**************************************************************************************************/
 482 vector<string> ShhhSeqsCommand::driverGroups(SequenceParser& parser, string newFFile, string newNFile, string newMFile, int start, int end, vector<string> groups){
 483         try {
 484
 485                 vector<string> mapFileNames;
 486
 487                 for (int i = start; i < end; i++) {
 488
 489                         start = time(NULL);
 490
 491                         if (m->control_pressed) {  return mapFileNames; }
 492
 493                         m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[i] + ":"); m->mothurOutEndLine();
 494
 495                         map<string, string> thisNameMap;
 496                         thisNameMap = parser.getNameMap(groups[i]);
 497                         vector<Sequence> thisSeqs = parser.getSeqs(groups[i]);
 498
 499                         vector<string> sequences;
 500                         vector<string> uniqueNames;
 501                         vector<string> redundantNames;
 502                         vector<int> seqFreq;
 503
 504                         seqNoise noise;
 505                         correctDist* correct = new correctDist(1); //we use one processor since we already split up the work load.
 506
 507                         //load this groups info in order
 508                         loadData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq, thisNameMap, thisSeqs);
 509                         if (m->control_pressed) { return mapFileNames; }
 510
 511                         //calc distances for cluster
 512                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + groups[i] + ".shhh.dist";
 513                         correct->execute(distFileName);
 514                         delete correct;
 515
 516                         if (m->control_pressed) { m->mothurRemove(distFileName); return mapFileNames; }
 517
 518                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, newFFile+groups[i], newNFile+groups[i], newMFile+groups[i]+".map");
 519
 520                         if (m->control_pressed) { return mapFileNames; }
 521
 522                         m->appendFiles(newFFile+groups[i], newFFile); m->mothurRemove(newFFile+groups[i]);
 523                         m->appendFiles(newNFile+groups[i], newNFile); m->mothurRemove(newNFile+groups[i]);
 524                         mapFileNames.push_back(newMFile+groups[i]+".map");
 525
 526                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to process group " + groups[i] + "."); m->mothurOutEndLine();
 527                 }
 528
 529                 return mapFileNames;
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "ShhhSeqsCommand", "driverGroups");
 533                 exit(1);
 534         }
 535 }
 536 //**********************************************************************************************************************
 537 int ShhhSeqsCommand::driver(seqNoise& noise,
 538                                                         vector<string>& sequences,
 539                                                         vector<string>& uniqueNames,
 540                                                         vector<string>& redundantNames,
 541                                                         vector<int>& seqFreq,
 542                                                         string distFileName, string outputFileName, string nameFileName, string mapFileName) {
 543         try {
 544                 double cutOff = 0.08;
 545                 int minIter = 10;
 546                 int maxIter = 1000;
 547                 double minDelta = 1e-6;
 548                 int numIters = 0;
 549                 double maxDelta = 1e6;
 550                 int numSeqs = sequences.size();
 551
 552                 //run cluster command
 553                 string inputString = "phylip=" + distFileName + ", method=furthest, cutoff=0.08";
 554                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 555                 m->mothurOut("Running command: cluster(" + inputString + ")"); m->mothurOutEndLine();
 556
 557                 Command* clusterCommand = new ClusterCommand(inputString);
 558                 clusterCommand->execute();
 559
 560                 map<string, vector<string> > filenames = clusterCommand->getOutputFiles();
 561                 string listFileName = filenames["list"][0];
 562                 string rabundFileName = filenames["rabund"][0]; m->mothurRemove(rabundFileName);
 563                 string sabundFileName = filenames["sabund"][0]; m->mothurRemove(sabundFileName);
 564
 565                 delete clusterCommand;
 566                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 567
 568                 if (m->control_pressed) { m->mothurRemove(distFileName); m->mothurRemove(listFileName); return 0; }
 569
 570                 vector<double> distances(numSeqs * numSeqs);
 571                 noise.getDistanceData(distFileName, distances);
 572                 m->mothurRemove(distFileName);
 573                 if (m->control_pressed) { m->mothurRemove(listFileName); return 0; }
 574
 575                 vector<int> otuData(numSeqs);
 576                 vector<int> otuFreq;
 577                 vector<vector<int> > otuBySeqLookUp;
 578                 noise.getListData(listFileName, cutOff, otuData, otuFreq, otuBySeqLookUp);
 579                 m->mothurRemove(listFileName);
 580                 if (m->control_pressed) { return 0; }
 581
 582                 int numOTUs = otuFreq.size();
 583
 584                 vector<double> weights(numOTUs, 0);
 585                 vector<int> change(numOTUs, 1);
 586                 vector<int> centroids(numOTUs, -1);
 587                 vector<int> cumCount(numOTUs, 0);
 588
 589                 vector<double> tau(numSeqs, 1);
 590                 vector<int> anP(numSeqs, 0);
 591                 vector<int> anI(numSeqs, 0);
 592                 vector<int> anN(numSeqs, 0);
 593                 vector<vector<int> > aanI = otuBySeqLookUp;
 594
 595                 while(numIters < minIter || ((maxDelta > minDelta) && (numIters < maxIter))){
 596
 597                         if (m->control_pressed) { return 0; }
 598
 599                         noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount); if (m->control_pressed) { return 0; }
 600                         maxDelta = noise.calcNewWeights(weights, seqFreq, anI, cumCount, anP, otuFreq, tau);  if (m->control_pressed) { return 0; }
 601
 602                         noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 603                         noise.checkCentroids(weights, centroids); if (m->control_pressed) { return 0; }
 604
 605                         otuFreq.assign(numOTUs, 0);
 606
 607                         int total = 0;
 608
 609                         for(int i=0;i<numSeqs;i++){
 610                                 if (m->control_pressed) { return 0; }
 611
 612                                 double offset = 1e6;
 613                                 double norm = 0.0000;
 614                                 double minWeight = 0.1;
 615                                 vector<double> currentTau(numOTUs);
 616
 617                                 for(int j=0;j<numOTUs;j++){
 618                                         if (m->control_pressed) { return 0; }
 619                                         if(weights[j] > minWeight && distances[i * numSeqs+centroids[j]] < offset){
 620                                                 offset = distances[i * numSeqs+centroids[j]];
 621                                         }
 622                                 }
 623
 624                                 for(int j=0;j<numOTUs;j++){
 625                                         if (m->control_pressed) { return 0; }
 626                                         if(weights[j] > minWeight){
 627                                                 currentTau[j] = exp(sigma * (-distances[(i * numSeqs + centroids[j])] + offset)) * weights[j];
 628                                                 norm += currentTau[j];
 629                                         }
 630                                         else{
 631                                                 currentTau[j] = 0.0000;
 632                                         }
 633                                 }
 634
 635                                 for(int j=0;j<numOTUs;j++){
 636                                         if (m->control_pressed) { return 0; }
 637                                         currentTau[j] /= norm;
 638                                 }
 639
 640                                 for(int j=0;j<numOTUs;j++){
 641                                         if (m->control_pressed) { return 0; }
 642
 643                                         if(currentTau[j] > 1.0e-4){
 644                                                 int oldTotal = total;
 645                                                 total++;
 646
 647                                                 tau.resize(oldTotal+1);
 648                                                 tau[oldTotal] = currentTau[j];
 649                                                 otuBySeqLookUp[j][otuFreq[j]] = oldTotal;
 650                                                 aanI[j][otuFreq[j]] = i;
 651                                                 otuFreq[j]++;
 652
 653                                         }
 654                                 }
 655
 656                                 anP.resize(total);
 657                                 anI.resize(total);
 658                         }
 659
 660                         numIters++;
 661                 }
 662
 663                 noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount);  if (m->control_pressed) { return 0; }
 664
 665                 vector<double> percentage(numSeqs);
 666                 noise.setUpOTUData(otuData, percentage, cumCount, tau, otuFreq, anP, anI);  if (m->control_pressed) { return 0; }
 667                 noise.finishOTUData(otuData, otuFreq, anP, anI, cumCount, otuBySeqLookUp, aanI, tau);  if (m->control_pressed) { return 0; }
 668
 669                 change.assign(numOTUs, 1);
 670                 noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 671
 672
 673                 vector<int> finalTau(numOTUs, 0);
 674                 for(int i=0;i<numSeqs;i++){
 675                         if (m->control_pressed) { return 0; }
 676                         finalTau[otuData[i]] += int(seqFreq[i]);
 677                 }
 678
 679                 noise.writeOutput(outputFileName, nameFileName, mapFileName, finalTau, centroids, otuData, sequences, uniqueNames, redundantNames, seqFreq, distances);
 680
 681                 return 0;
 682
 683         }catch(exception& e) {
 684                 m->errorOut(e, "ShhhSeqsCommand", "driver");
 685                 exit(1);
 686         }
 687 }
 688 //**********************************************************************************************************************
 689 int ShhhSeqsCommand::deconvoluteResults(string fastaFile, string nameFile){
 690         try {
 691                 m->mothurOutEndLine(); m->mothurOut("Deconvoluting results:"); m->mothurOutEndLine(); m->mothurOutEndLine();
 692
 693                 //use unique.seqs to create new name and fastafile
 694                 string inputString = "fasta=" + fastaFile + ", name=" + nameFile;
 695                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 696                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 697
 698                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 699                 uniqueCommand->execute();
 700
 701                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 702
 703                 delete uniqueCommand;
 704
 705                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 706
 707                 string newnameFile = filenames["name"][0];
 708                 string newfastaFile = filenames["fasta"][0];
 709
 710                 m->mothurRemove(fastaFile); rename(newfastaFile.c_str(), fastaFile.c_str());
 711                 if (nameFile != newnameFile) { m->mothurRemove(nameFile); rename(newnameFile.c_str(), nameFile.c_str()); }
 712
 713                 return 0;
 714         }
 715         catch(exception& e) {
 716                 m->errorOut(e, "ShhhSeqsCommand", "deconvoluteResults");
 717                 exit(1);
 718         }
 719 }
 720 //**********************************************************************************************************************
 721
 722
 723