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         }
 170         catch(exception& e) {
 171                 m->errorOut(e, "ShhhSeqsCommand", "ShhhSeqsCommand");
 172                 exit(1);
 173         }
 174 }
 175 //**********************************************************************************************************************
 176 int ShhhSeqsCommand::execute() {
 177         try {
 178
 179                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 180
 181                 if (outputDir == "") { outputDir = m->hasPath(fastafile);  }//if user entered a file with a path then preserve it
 182                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "shhh.fasta";
 183                 string nameFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.names";
 184                 string mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.map";
 185
 186                 if (groupfile != "") {
 187                         //Parse sequences by group
 188                         SequenceParser parser(groupfile, fastafile, namefile);
 189                         vector<string> groups = parser.getNamesOfGroups();
 190
 191                         if (m->control_pressed) {  return 0; }
 192
 193                         //clears files
 194                         ofstream out, out1, out2;
 195                         m->openOutputFile(outputFileName, out); out.close();
 196                         m->openOutputFile(nameFileName, out1); out1.close();
 197                         mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.";
 198
 199                         vector<string> mapFileNames;
 200                         if(processors == 1)     {       mapFileNames = driverGroups(parser, outputFileName, nameFileName, mapFileName, 0, groups.size(), groups);       }
 201                         else                            {       mapFileNames = createProcessesGroups(parser, outputFileName, nameFileName, mapFileName, groups);                        }
 202
 203                         if (m->control_pressed) {    return 0;  }
 204
 205                         for (int j = 0; j < mapFileNames.size(); j++) { outputNames.push_back(mapFileNames[j]); outputTypes["map"].push_back(mapFileNames[j]); }
 206
 207                         //deconvolute results by running unique.seqs
 208                         deconvoluteResults(outputFileName, nameFileName);
 209
 210                         if (m->control_pressed) {   return 0;   }
 211
 212                 }else{
 213                         vector<string> sequences;
 214                         vector<string> uniqueNames;
 215                         vector<string> redundantNames;
 216                         vector<int> seqFreq;
 217
 218                         seqNoise noise;
 219                         correctDist* correct = new correctDist(processors);
 220
 221                         //reads fasta and name file and loads them in order
 222                         readData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq);
 223                         if (m->control_pressed) { return 0; }
 224
 225                         //calc distances for cluster
 226                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "shhh.dist";
 227                         correct->execute(distFileName);
 228                         delete correct;
 229
 230                         if (m->control_pressed) { m->mothurRemove(distFileName); return 0; }
 231
 232                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, outputFileName, nameFileName, mapFileName);
 233                         outputNames.push_back(mapFileName); outputTypes["map"].push_back(mapFileName);
 234                 }
 235
 236                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 237
 238                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 239                 outputNames.push_back(nameFileName); outputTypes["name"].push_back(nameFileName);
 240
 241                 m->mothurOutEndLine();
 242                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 243                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 244                 m->mothurOutEndLine();
 245
 246                 //set accnos file as new current accnosfile
 247                 string current = "";
 248                 itTypes = outputTypes.find("fasta");
 249                 if (itTypes != outputTypes.end()) {
 250                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 251                 }
 252
 253                 itTypes = outputTypes.find("name");
 254                 if (itTypes != outputTypes.end()) {
 255                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 256                 }
 257
 258
 259                 return 0;
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "ShhhSeqsCommand", "execute");
 263                 exit(1);
 264         }
 265 }
 266 //**********************************************************************************************************************
 267 int ShhhSeqsCommand::readData(correctDist* correct, seqNoise& noise, vector<string>& seqs, vector<string>& uNames, vector<string>& rNames, vector<int>& freq) {
 268         try {
 269                 map<string, string> nameMap;
 270                 map<string, string>::iterator it;
 271                 m->readNames(namefile, nameMap);
 272                 bool error = false;
 273
 274                 ifstream in;
 275                 m->openInputFile(fastafile, in);
 276
 277                 while (!in.eof()) {
 278
 279                         if (m->control_pressed) { in.close(); return 0; }
 280
 281                         Sequence seq(in); m->gobble(in);
 282
 283                         if (seq.getName() != "") {
 284                                 correct->addSeq(seq.getName(), seq.getAligned());
 285
 286                                 it = nameMap.find(seq.getName());
 287                                 if (it != nameMap.end()) {
 288                                         noise.addSeq(seq.getAligned(), seqs);
 289                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 290                                 }else {
 291                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your namefile, please correct.");
 292                                         error = true;
 293                                 }
 294                         }
 295                 }
 296                 in.close();
 297
 298                 if (error) { m->control_pressed = true; }
 299
 300                 return seqs.size();
 301
 302         }catch(exception& e) {
 303                 m->errorOut(e, "ShhhSeqsCommand", "readData");
 304                 exit(1);
 305         }
 306 }
 307 //**********************************************************************************************************************
 308 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) {
 309         try {
 310                 bool error = false;
 311                 map<string, string>::iterator it;
 312
 313                 for (int i = 0; i < sequences.size(); i++) {
 314
 315                         if (m->control_pressed) { return 0; }
 316
 317                         if (sequences[i].getName() != "") {
 318                                 correct->addSeq(sequences[i].getName(), sequences[i].getAligned());
 319
 320                                 it = nameMap.find(sequences[i].getName());
 321                                 if (it != nameMap.end()) {
 322                                         noise.addSeq(sequences[i].getAligned(), seqs);
 323                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 324                                 }else {
 325                                         m->mothurOut("[ERROR]: " + sequences[i].getName() + " is in your fasta file and not in your namefile, please correct.");
 326                                         error = true;
 327                                 }
 328                         }
 329                 }
 330
 331                 if (error) { m->control_pressed = true; }
 332
 333                 return seqs.size();
 334
 335         }catch(exception& e) {
 336                 m->errorOut(e, "ShhhSeqsCommand", "loadData");
 337                 exit(1);
 338         }
 339 }
 340 /**************************************************************************************************/
 341 vector<string> ShhhSeqsCommand::createProcessesGroups(SequenceParser& parser, string newFName, string newNName, string newMName, vector<string> groups) {
 342         try {
 343
 344                 vector<int> processIDS;
 345                 int process = 1;
 346                 vector<string> mapfileNames;
 347
 348                 //sanity check
 349                 if (groups.size() < processors) { processors = groups.size(); }
 350
 351                 //divide the groups between the processors
 352                 vector<linePair> lines;
 353                 int numGroupsPerProcessor = groups.size() / processors;
 354                 for (int i = 0; i < processors; i++) {
 355                         int startIndex =  i * numGroupsPerProcessor;
 356                         int endIndex = (i+1) * numGroupsPerProcessor;
 357                         if(i == (processors - 1)){      endIndex = groups.size();       }
 358                         lines.push_back(linePair(startIndex, endIndex));
 359                 }
 360
 361 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 362
 363                 //loop through and create all the processes you want
 364                 while (process != processors) {
 365                         int pid = fork();
 366
 367                         if (pid > 0) {
 368                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 369                                 process++;
 370                         }else if (pid == 0){
 371                                 mapfileNames = driverGroups(parser, newFName + toString(getpid()) + ".temp", newNName + toString(getpid()) + ".temp", newMName, lines[process].start, lines[process].end, groups);
 372
 373                                 //pass filenames to parent
 374                                 ofstream out;
 375                                 string tempFile = newMName + toString(getpid()) + ".temp";
 376                                 m->openOutputFile(tempFile, out);
 377                                 out << mapfileNames.size() << endl;
 378                                 for (int i = 0; i < mapfileNames.size(); i++) {
 379                                         out << mapfileNames[i] << endl;
 380                                 }
 381                                 out.close();
 382
 383                                 exit(0);
 384                         }else {
 385                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 386                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 387                                 exit(0);
 388                         }
 389                 }
 390
 391                 //do my part
 392                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 393
 394                 //force parent to wait until all the processes are done
 395                 for (int i=0;i<processIDS.size();i++) {
 396                         int temp = processIDS[i];
 397                         wait(&temp);
 398                 }
 399
 400                 //append output files
 401                 for(int i=0;i<processIDS.size();i++){
 402                         ifstream in;
 403                         string tempFile =  newMName + toString(processIDS[i]) + ".temp";
 404                         m->openInputFile(tempFile, in);
 405                         if (!in.eof()) {
 406                                 int tempNum = 0; in >> tempNum;  m->gobble(in);
 407                                 for (int j = 0; j < tempNum; j++) {
 408                                         string filename;
 409                                         in >> filename; m->gobble(in);
 410                                         mapfileNames.push_back(filename);
 411                                 }
 412                         }
 413                         in.close(); m->mothurRemove(tempFile);
 414
 415                 }
 416 #else
 417
 418                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 419                 //Windows version shared memory, so be careful when passing variables through the shhhseqsData struct.
 420                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 421                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 422
 423                 vector<shhhseqsData*> pDataArray;
 424                 DWORD   dwThreadIdArray[processors-1];
 425                 HANDLE  hThreadArray[processors-1];
 426
 427                 //Create processor worker threads.
 428                 for( int i=1; i<processors; i++ ){
 429                         // Allocate memory for thread data.
 430                         string extension = toString(i) + ".temp";
 431
 432                         shhhseqsData* tempShhhseqs = new shhhseqsData(fastafile, namefile, groupfile, (newFName+extension), (newNName+extension), newMName, groups, m, lines[i].start, lines[i].end, sigma, i);
 433                         pDataArray.push_back(tempShhhseqs);
 434                         processIDS.push_back(i);
 435
 436                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 437                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 438                         hThreadArray[i-1] = CreateThread(NULL, 0, MyShhhSeqsThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 439                 }
 440
 441
 442                 //using the main process as a worker saves time and memory
 443                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 444
 445                 //Wait until all threads have terminated.
 446                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 447
 448                 //Close all thread handles and free memory allocations.
 449                 for(int i=0; i < pDataArray.size(); i++){
 450                         for (int j = 0; j < pDataArray[i]->mapfileNames.size(); j++) {
 451                                 mapfileNames.push_back(pDataArray[i]->mapfileNames[j]);
 452                         }
 453                         CloseHandle(hThreadArray[i]);
 454                         delete pDataArray[i];
 455                 }
 456
 457 #endif
 458
 459                 //append output files
 460                 for(int i=0;i<processIDS.size();i++){
 461                         m->appendFiles((newFName + toString(processIDS[i]) + ".temp"), newFName);
 462                         m->mothurRemove((newFName + toString(processIDS[i]) + ".temp"));
 463
 464                         m->appendFiles((newNName + toString(processIDS[i]) + ".temp"), newNName);
 465                         m->mothurRemove((newNName + toString(processIDS[i]) + ".temp"));
 466                 }
 467
 468                 return mapfileNames;
 469
 470         }
 471         catch(exception& e) {
 472                 m->errorOut(e, "ShhhSeqsCommand", "createProcessesGroups");
 473                 exit(1);
 474         }
 475 }
 476 /**************************************************************************************************/
 477 vector<string> ShhhSeqsCommand::driverGroups(SequenceParser& parser, string newFFile, string newNFile, string newMFile, int start, int end, vector<string> groups){
 478         try {
 479
 480                 vector<string> mapFileNames;
 481
 482                 for (int i = start; i < end; i++) {
 483
 484                         start = time(NULL);
 485
 486                         if (m->control_pressed) {  return mapFileNames; }
 487
 488                         m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[i] + ":"); m->mothurOutEndLine();
 489
 490                         map<string, string> thisNameMap;
 491                         thisNameMap = parser.getNameMap(groups[i]);
 492                         vector<Sequence> thisSeqs = parser.getSeqs(groups[i]);
 493
 494                         vector<string> sequences;
 495                         vector<string> uniqueNames;
 496                         vector<string> redundantNames;
 497                         vector<int> seqFreq;
 498
 499                         seqNoise noise;
 500                         correctDist* correct = new correctDist(1); //we use one processor since we already split up the work load.
 501
 502                         //load this groups info in order
 503                         loadData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq, thisNameMap, thisSeqs);
 504                         if (m->control_pressed) { return mapFileNames; }
 505
 506                         //calc distances for cluster
 507                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + groups[i] + ".shhh.dist";
 508                         correct->execute(distFileName);
 509                         delete correct;
 510
 511                         if (m->control_pressed) { m->mothurRemove(distFileName); return mapFileNames; }
 512
 513                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, newFFile+groups[i], newNFile+groups[i], newMFile+groups[i]+".map");
 514
 515                         if (m->control_pressed) { return mapFileNames; }
 516
 517                         m->appendFiles(newFFile+groups[i], newFFile); m->mothurRemove(newFFile+groups[i]);
 518                         m->appendFiles(newNFile+groups[i], newNFile); m->mothurRemove(newNFile+groups[i]);
 519                         mapFileNames.push_back(newMFile+groups[i]+".map");
 520
 521                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to process group " + groups[i] + "."); m->mothurOutEndLine();
 522                 }
 523
 524                 return mapFileNames;
 525         }
 526         catch(exception& e) {
 527                 m->errorOut(e, "ShhhSeqsCommand", "driverGroups");
 528                 exit(1);
 529         }
 530 }
 531 //**********************************************************************************************************************
 532 int ShhhSeqsCommand::driver(seqNoise& noise,
 533                                                         vector<string>& sequences,
 534                                                         vector<string>& uniqueNames,
 535                                                         vector<string>& redundantNames,
 536                                                         vector<int>& seqFreq,
 537                                                         string distFileName, string outputFileName, string nameFileName, string mapFileName) {
 538         try {
 539                 double cutOff = 0.08;
 540                 int minIter = 10;
 541                 int maxIter = 1000;
 542                 double minDelta = 1e-6;
 543                 int numIters = 0;
 544                 double maxDelta = 1e6;
 545                 int numSeqs = sequences.size();
 546
 547                 //run cluster command
 548                 string inputString = "phylip=" + distFileName + ", method=furthest, cutoff=0.08";
 549                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 550                 m->mothurOut("Running command: cluster(" + inputString + ")"); m->mothurOutEndLine();
 551
 552                 Command* clusterCommand = new ClusterCommand(inputString);
 553                 clusterCommand->execute();
 554
 555                 map<string, vector<string> > filenames = clusterCommand->getOutputFiles();
 556                 string listFileName = filenames["list"][0];
 557                 string rabundFileName = filenames["rabund"][0]; m->mothurRemove(rabundFileName);
 558                 string sabundFileName = filenames["sabund"][0]; m->mothurRemove(sabundFileName);
 559
 560                 delete clusterCommand;
 561                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 562
 563                 if (m->control_pressed) { m->mothurRemove(distFileName); m->mothurRemove(listFileName); return 0; }
 564
 565                 vector<double> distances(numSeqs * numSeqs);
 566                 noise.getDistanceData(distFileName, distances);
 567                 m->mothurRemove(distFileName);
 568                 if (m->control_pressed) { m->mothurRemove(listFileName); return 0; }
 569
 570                 vector<int> otuData(numSeqs);
 571                 vector<int> otuFreq;
 572                 vector<vector<int> > otuBySeqLookUp;
 573                 noise.getListData(listFileName, cutOff, otuData, otuFreq, otuBySeqLookUp);
 574                 m->mothurRemove(listFileName);
 575                 if (m->control_pressed) { return 0; }
 576
 577                 int numOTUs = otuFreq.size();
 578
 579                 vector<double> weights(numOTUs, 0);
 580                 vector<int> change(numOTUs, 1);
 581                 vector<int> centroids(numOTUs, -1);
 582                 vector<int> cumCount(numOTUs, 0);
 583
 584                 vector<double> tau(numSeqs, 1);
 585                 vector<int> anP(numSeqs, 0);
 586                 vector<int> anI(numSeqs, 0);
 587                 vector<int> anN(numSeqs, 0);
 588                 vector<vector<int> > aanI = otuBySeqLookUp;
 589
 590                 while(numIters < minIter || ((maxDelta > minDelta) && (numIters < maxIter))){
 591
 592                         if (m->control_pressed) { return 0; }
 593
 594                         noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount); if (m->control_pressed) { return 0; }
 595                         maxDelta = noise.calcNewWeights(weights, seqFreq, anI, cumCount, anP, otuFreq, tau);  if (m->control_pressed) { return 0; }
 596
 597                         noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 598                         noise.checkCentroids(weights, centroids); if (m->control_pressed) { return 0; }
 599
 600                         otuFreq.assign(numOTUs, 0);
 601
 602                         int total = 0;
 603
 604                         for(int i=0;i<numSeqs;i++){
 605                                 if (m->control_pressed) { return 0; }
 606
 607                                 double offset = 1e6;
 608                                 double norm = 0.0000;
 609                                 double minWeight = 0.1;
 610                                 vector<double> currentTau(numOTUs);
 611
 612                                 for(int j=0;j<numOTUs;j++){
 613                                         if (m->control_pressed) { return 0; }
 614                                         if(weights[j] > minWeight && distances[i * numSeqs+centroids[j]] < offset){
 615                                                 offset = distances[i * numSeqs+centroids[j]];
 616                                         }
 617                                 }
 618
 619                                 for(int j=0;j<numOTUs;j++){
 620                                         if (m->control_pressed) { return 0; }
 621                                         if(weights[j] > minWeight){
 622                                                 currentTau[j] = exp(sigma * (-distances[(i * numSeqs + centroids[j])] + offset)) * weights[j];
 623                                                 norm += currentTau[j];
 624                                         }
 625                                         else{
 626                                                 currentTau[j] = 0.0000;
 627                                         }
 628                                 }
 629
 630                                 for(int j=0;j<numOTUs;j++){
 631                                         if (m->control_pressed) { return 0; }
 632                                         currentTau[j] /= norm;
 633                                 }
 634
 635                                 for(int j=0;j<numOTUs;j++){
 636                                         if (m->control_pressed) { return 0; }
 637
 638                                         if(currentTau[j] > 1.0e-4){
 639                                                 int oldTotal = total;
 640                                                 total++;
 641
 642                                                 tau.resize(oldTotal+1);
 643                                                 tau[oldTotal] = currentTau[j];
 644                                                 otuBySeqLookUp[j][otuFreq[j]] = oldTotal;
 645                                                 aanI[j][otuFreq[j]] = i;
 646                                                 otuFreq[j]++;
 647
 648                                         }
 649                                 }
 650
 651                                 anP.resize(total);
 652                                 anI.resize(total);
 653                         }
 654
 655                         numIters++;
 656                 }
 657
 658                 noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount);  if (m->control_pressed) { return 0; }
 659
 660                 vector<double> percentage(numSeqs);
 661                 noise.setUpOTUData(otuData, percentage, cumCount, tau, otuFreq, anP, anI);  if (m->control_pressed) { return 0; }
 662                 noise.finishOTUData(otuData, otuFreq, anP, anI, cumCount, otuBySeqLookUp, aanI, tau);  if (m->control_pressed) { return 0; }
 663
 664                 change.assign(numOTUs, 1);
 665                 noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 666
 667
 668                 vector<int> finalTau(numOTUs, 0);
 669                 for(int i=0;i<numSeqs;i++){
 670                         if (m->control_pressed) { return 0; }
 671                         finalTau[otuData[i]] += int(seqFreq[i]);
 672                 }
 673
 674                 noise.writeOutput(outputFileName, nameFileName, mapFileName, finalTau, centroids, otuData, sequences, uniqueNames, redundantNames, seqFreq, distances);
 675
 676                 return 0;
 677
 678         }catch(exception& e) {
 679                 m->errorOut(e, "ShhhSeqsCommand", "driver");
 680                 exit(1);
 681         }
 682 }
 683 //**********************************************************************************************************************
 684 int ShhhSeqsCommand::deconvoluteResults(string fastaFile, string nameFile){
 685         try {
 686                 m->mothurOutEndLine(); m->mothurOut("Deconvoluting results:"); m->mothurOutEndLine(); m->mothurOutEndLine();
 687
 688                 //use unique.seqs to create new name and fastafile
 689                 string inputString = "fasta=" + fastaFile + ", name=" + nameFile;
 690                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 691                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 692
 693                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 694                 uniqueCommand->execute();
 695
 696                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 697
 698                 delete uniqueCommand;
 699
 700                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 701
 702                 string newnameFile = filenames["name"][0];
 703                 string newfastaFile = filenames["fasta"][0];
 704
 705                 m->mothurRemove(fastaFile); rename(newfastaFile.c_str(), fastaFile.c_str());
 706                 if (nameFile != newnameFile) { m->mothurRemove(nameFile); rename(newnameFile.c_str(), nameFile.c_str()); }
 707
 708                 return 0;
 709         }
 710         catch(exception& e) {
 711                 m->errorOut(e, "ShhhSeqsCommand", "deconvoluteResults");
 712                 exit(1);
 713         }
 714 }
 715 //**********************************************************************************************************************
 716
 717
 718