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 string ShhhSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  59         try {
  60         string outputFileName = "";
  61                 map<string, vector<string> >::iterator it;
  62
  63         //is this a type this command creates
  64         it = outputTypes.find(type);
  65         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  66         else {
  67             if (type == "fasta")            {   outputFileName =  "shhh_seqs.fasta";   }
  68             else if (type == "name")    {   outputFileName =  "shhh_seqs.names";   }
  69             else if (type == "map")        {   outputFileName =  "shhh_seqs.map";   }
  70             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  71         }
  72         return outputFileName;
  73         }
  74         catch(exception& e) {
  75                 m->errorOut(e, "ShhhSeqsCommand", "getOutputFileNameTag");
  76                 exit(1);
  77         }
  78 }
  79
  80 //**********************************************************************************************************************
  81
  82 ShhhSeqsCommand::ShhhSeqsCommand(){
  83         try {
  84                 abort = true; calledHelp = true;
  85                 setParameters();
  86                 vector<string> tempOutNames;
  87                 outputTypes["fasta"] = tempOutNames;
  88                 outputTypes["name"] = tempOutNames;
  89                 outputTypes["map"] = tempOutNames;
  90         }
  91         catch(exception& e) {
  92                 m->errorOut(e, "ShhhSeqsCommand", "ShhhSeqsCommand");
  93                 exit(1);
  94         }
  95 }
  96
  97 //**********************************************************************************************************************
  98 ShhhSeqsCommand::ShhhSeqsCommand(string option) {
  99         try {
 100                 abort = false; calledHelp = false;
 101
 102                 //allow user to run help
 103                 if(option == "help") { help(); abort = true; calledHelp = true; }
 104                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 105
 106                 else {
 107                         vector<string> myArray = setParameters();
 108
 109                         OptionParser parser(option);
 110                         map<string, string> parameters = parser.getParameters();
 111
 112                         ValidParameters validParameter;
 113                         map<string, string>::iterator it;
 114
 115                         //check to make sure all parameters are valid for command
 116                         for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
 117                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
 118                         }
 119
 120                         //initialize outputTypes
 121                         vector<string> tempOutNames;
 122                         outputTypes["fasta"] = tempOutNames;
 123                         outputTypes["name"] = tempOutNames;
 124                         outputTypes["map"] = tempOutNames;
 125
 126                         //if the user changes the input directory command factory will send this info to us in the output parameter
 127                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 128                         if (inputDir == "not found"){   inputDir = "";          }
 129                         else {
 130                                 string path;
 131                                 it = parameters.find("fasta");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 137                                 }
 138
 139                                 it = parameters.find("name");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 145                                 }
 146
 147                                 it = parameters.find("group");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 153                                 }
 154                         }
 155
 156                         //check for required parameters
 157                         fastafile = validParameter.validFile(parameters, "fasta", true);
 158                         if (fastafile == "not found") {
 159                                 fastafile = m->getFastaFile();
 160                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 161                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 162                         }
 163                         else if (fastafile == "not open") { abort = true; }
 164                         else { m->setFastaFile(fastafile); }
 165
 166                         //if the user changes the output directory command factory will send this info to us in the output parameter
 167                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 168
 169                         //check for optional parameter and set defaults
 170                         // ...at some point should added some additional type checking...
 171                         namefile = validParameter.validFile(parameters, "name", true);
 172                         if (namefile == "not found") {
 173                                 namefile = m->getNameFile();
 174                                 if (namefile != "") { m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 175                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 176                         }
 177                         else if (namefile == "not open") { namefile =  ""; abort = true; }
 178                         else {  m->setNameFile(namefile); }
 179
 180                         groupfile = validParameter.validFile(parameters, "group", true);
 181                         if (groupfile == "not found") { groupfile =  "";   }
 182                         else if (groupfile == "not open") { abort = true; groupfile =  ""; }
 183                         else {   m->setGroupFile(groupfile);  }
 184
 185                         string temp     = validParameter.validFile(parameters, "sigma", false);         if(temp == "not found"){        temp = "0.01"; }
 186                         m->mothurConvert(temp, sigma);
 187                         sigma = 1/sigma;
 188
 189                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 190                         m->setProcessors(temp);
 191                         m->mothurConvert(temp, processors);
 192
 193                         if (namefile == "") {
 194                                 vector<string> files; files.push_back(fastafile);
 195                                 parser.getNameFile(files);
 196                         }
 197                 }
 198         }
 199         catch(exception& e) {
 200                 m->errorOut(e, "ShhhSeqsCommand", "ShhhSeqsCommand");
 201                 exit(1);
 202         }
 203 }
 204 //**********************************************************************************************************************
 205 int ShhhSeqsCommand::execute() {
 206         try {
 207
 208                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 209
 210                 if (outputDir == "") { outputDir = m->hasPath(fastafile);  }//if user entered a file with a path then preserve it
 211                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta");
 212                 string nameFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + getOutputFileNameTag("name");
 213                 string mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + getOutputFileNameTag("map");
 214
 215                 if (groupfile != "") {
 216                         //Parse sequences by group
 217                         SequenceParser parser(groupfile, fastafile, namefile);
 218                         vector<string> groups = parser.getNamesOfGroups();
 219
 220                         if (m->control_pressed) {  return 0; }
 221
 222                         //clears files
 223                         ofstream out, out1, out2;
 224                         m->openOutputFile(outputFileName, out); out.close();
 225                         m->openOutputFile(nameFileName, out1); out1.close();
 226                         mapFileName = outputDir + m->getRootName(m->getSimpleName(fastafile))  + "shhh.";
 227
 228                         vector<string> mapFileNames;
 229                         if(processors == 1)     {       mapFileNames = driverGroups(parser, outputFileName, nameFileName, mapFileName, 0, groups.size(), groups);       }
 230                         else                            {       mapFileNames = createProcessesGroups(parser, outputFileName, nameFileName, mapFileName, groups);                        }
 231
 232                         if (m->control_pressed) {    return 0;  }
 233
 234                         for (int j = 0; j < mapFileNames.size(); j++) { outputNames.push_back(mapFileNames[j]); outputTypes["map"].push_back(mapFileNames[j]); }
 235
 236                         //deconvolute results by running unique.seqs
 237                         deconvoluteResults(outputFileName, nameFileName);
 238
 239                         if (m->control_pressed) {   return 0;   }
 240
 241                 }else{
 242                         vector<string> sequences;
 243                         vector<string> uniqueNames;
 244                         vector<string> redundantNames;
 245                         vector<int> seqFreq;
 246
 247                         seqNoise noise;
 248                         correctDist* correct = new correctDist(processors);
 249
 250                         //reads fasta and name file and loads them in order
 251                         readData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq);
 252                         if (m->control_pressed) { return 0; }
 253
 254                         //calc distances for cluster
 255                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "shhh.dist";
 256                         correct->execute(distFileName);
 257                         delete correct;
 258
 259                         if (m->control_pressed) { m->mothurRemove(distFileName); return 0; }
 260
 261                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, outputFileName, nameFileName, mapFileName);
 262                         outputNames.push_back(mapFileName); outputTypes["map"].push_back(mapFileName);
 263                 }
 264
 265                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 266
 267                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 268                 outputNames.push_back(nameFileName); outputTypes["name"].push_back(nameFileName);
 269
 270                 m->mothurOutEndLine();
 271                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 272                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 273                 m->mothurOutEndLine();
 274
 275                 //set accnos file as new current accnosfile
 276                 string current = "";
 277                 itTypes = outputTypes.find("fasta");
 278                 if (itTypes != outputTypes.end()) {
 279                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 280                 }
 281
 282                 itTypes = outputTypes.find("name");
 283                 if (itTypes != outputTypes.end()) {
 284                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 285                 }
 286
 287
 288                 return 0;
 289         }
 290         catch(exception& e) {
 291                 m->errorOut(e, "ShhhSeqsCommand", "execute");
 292                 exit(1);
 293         }
 294 }
 295 //**********************************************************************************************************************
 296 int ShhhSeqsCommand::readData(correctDist* correct, seqNoise& noise, vector<string>& seqs, vector<string>& uNames, vector<string>& rNames, vector<int>& freq) {
 297         try {
 298                 map<string, string> nameMap;
 299                 map<string, string>::iterator it;
 300                 m->readNames(namefile, nameMap);
 301                 bool error = false;
 302
 303                 ifstream in;
 304                 m->openInputFile(fastafile, in);
 305
 306                 while (!in.eof()) {
 307
 308                         if (m->control_pressed) { in.close(); return 0; }
 309
 310                         Sequence seq(in); m->gobble(in);
 311
 312                         if (seq.getName() != "") {
 313                                 correct->addSeq(seq.getName(), seq.getAligned());
 314
 315                                 it = nameMap.find(seq.getName());
 316                                 if (it != nameMap.end()) {
 317                                         noise.addSeq(seq.getAligned(), seqs);
 318                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 319                                 }else {
 320                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your namefile, please correct.");
 321                                         error = true;
 322                                 }
 323                         }
 324                 }
 325                 in.close();
 326
 327                 if (error) { m->control_pressed = true; }
 328
 329                 return seqs.size();
 330
 331         }catch(exception& e) {
 332                 m->errorOut(e, "ShhhSeqsCommand", "readData");
 333                 exit(1);
 334         }
 335 }
 336 //**********************************************************************************************************************
 337 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) {
 338         try {
 339                 bool error = false;
 340                 map<string, string>::iterator it;
 341
 342                 for (int i = 0; i < sequences.size(); i++) {
 343
 344                         if (m->control_pressed) { return 0; }
 345
 346                         if (sequences[i].getName() != "") {
 347                                 correct->addSeq(sequences[i].getName(), sequences[i].getAligned());
 348
 349                                 it = nameMap.find(sequences[i].getName());
 350                                 if (it != nameMap.end()) {
 351                                         noise.addSeq(sequences[i].getAligned(), seqs);
 352                                         noise.addRedundantName(it->first, it->second, uNames, rNames, freq);
 353                                 }else {
 354                                         m->mothurOut("[ERROR]: " + sequences[i].getName() + " is in your fasta file and not in your namefile, please correct.");
 355                                         error = true;
 356                                 }
 357                         }
 358                 }
 359
 360                 if (error) { m->control_pressed = true; }
 361
 362                 return seqs.size();
 363
 364         }catch(exception& e) {
 365                 m->errorOut(e, "ShhhSeqsCommand", "loadData");
 366                 exit(1);
 367         }
 368 }
 369 /**************************************************************************************************/
 370 vector<string> ShhhSeqsCommand::createProcessesGroups(SequenceParser& parser, string newFName, string newNName, string newMName, vector<string> groups) {
 371         try {
 372
 373                 vector<int> processIDS;
 374                 int process = 1;
 375                 vector<string> mapfileNames;
 376
 377                 //sanity check
 378                 if (groups.size() < processors) { processors = groups.size(); }
 379
 380                 //divide the groups between the processors
 381                 vector<linePair> lines;
 382                 int numGroupsPerProcessor = groups.size() / processors;
 383                 for (int i = 0; i < processors; i++) {
 384                         int startIndex =  i * numGroupsPerProcessor;
 385                         int endIndex = (i+1) * numGroupsPerProcessor;
 386                         if(i == (processors - 1)){      endIndex = groups.size();       }
 387                         lines.push_back(linePair(startIndex, endIndex));
 388                 }
 389
 390 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 391
 392                 //loop through and create all the processes you want
 393                 while (process != processors) {
 394                         int pid = fork();
 395
 396                         if (pid > 0) {
 397                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 398                                 process++;
 399                         }else if (pid == 0){
 400                                 mapfileNames = driverGroups(parser, newFName + toString(getpid()) + ".temp", newNName + toString(getpid()) + ".temp", newMName, lines[process].start, lines[process].end, groups);
 401
 402                                 //pass filenames to parent
 403                                 ofstream out;
 404                                 string tempFile = newMName + toString(getpid()) + ".temp";
 405                                 m->openOutputFile(tempFile, out);
 406                                 out << mapfileNames.size() << endl;
 407                                 for (int i = 0; i < mapfileNames.size(); i++) {
 408                                         out << mapfileNames[i] << endl;
 409                                 }
 410                                 out.close();
 411
 412                                 exit(0);
 413                         }else {
 414                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 415                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 416                                 exit(0);
 417                         }
 418                 }
 419
 420                 //do my part
 421                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 422
 423                 //force parent to wait until all the processes are done
 424                 for (int i=0;i<processIDS.size();i++) {
 425                         int temp = processIDS[i];
 426                         wait(&temp);
 427                 }
 428
 429                 //append output files
 430                 for(int i=0;i<processIDS.size();i++){
 431                         ifstream in;
 432                         string tempFile =  newMName + toString(processIDS[i]) + ".temp";
 433                         m->openInputFile(tempFile, in);
 434                         if (!in.eof()) {
 435                                 int tempNum = 0; in >> tempNum;  m->gobble(in);
 436                                 for (int j = 0; j < tempNum; j++) {
 437                                         string filename;
 438                                         in >> filename; m->gobble(in);
 439                                         mapfileNames.push_back(filename);
 440                                 }
 441                         }
 442                         in.close(); m->mothurRemove(tempFile);
 443
 444                 }
 445 #else
 446
 447                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 448                 //Windows version shared memory, so be careful when passing variables through the shhhseqsData struct.
 449                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 450                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 451
 452                 vector<shhhseqsData*> pDataArray;
 453                 DWORD   dwThreadIdArray[processors-1];
 454                 HANDLE  hThreadArray[processors-1];
 455
 456                 //Create processor worker threads.
 457                 for( int i=1; i<processors; i++ ){
 458                         // Allocate memory for thread data.
 459                         string extension = toString(i) + ".temp";
 460
 461                         shhhseqsData* tempShhhseqs = new shhhseqsData(fastafile, namefile, groupfile, (newFName+extension), (newNName+extension), newMName, groups, m, lines[i].start, lines[i].end, sigma, i);
 462                         pDataArray.push_back(tempShhhseqs);
 463                         processIDS.push_back(i);
 464
 465                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 466                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 467                         hThreadArray[i-1] = CreateThread(NULL, 0, MyShhhSeqsThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 468                 }
 469
 470
 471                 //using the main process as a worker saves time and memory
 472                 mapfileNames = driverGroups(parser, newFName, newNName, newMName, lines[0].start, lines[0].end, groups);
 473
 474                 //Wait until all threads have terminated.
 475                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 476
 477                 //Close all thread handles and free memory allocations.
 478                 for(int i=0; i < pDataArray.size(); i++){
 479                         for (int j = 0; j < pDataArray[i]->mapfileNames.size(); j++) {
 480                                 mapfileNames.push_back(pDataArray[i]->mapfileNames[j]);
 481                         }
 482                         CloseHandle(hThreadArray[i]);
 483                         delete pDataArray[i];
 484                 }
 485
 486 #endif
 487
 488                 //append output files
 489                 for(int i=0;i<processIDS.size();i++){
 490                         m->appendFiles((newFName + toString(processIDS[i]) + ".temp"), newFName);
 491                         m->mothurRemove((newFName + toString(processIDS[i]) + ".temp"));
 492
 493                         m->appendFiles((newNName + toString(processIDS[i]) + ".temp"), newNName);
 494                         m->mothurRemove((newNName + toString(processIDS[i]) + ".temp"));
 495                 }
 496
 497                 return mapfileNames;
 498
 499         }
 500         catch(exception& e) {
 501                 m->errorOut(e, "ShhhSeqsCommand", "createProcessesGroups");
 502                 exit(1);
 503         }
 504 }
 505 /**************************************************************************************************/
 506 vector<string> ShhhSeqsCommand::driverGroups(SequenceParser& parser, string newFFile, string newNFile, string newMFile, int start, int end, vector<string> groups){
 507         try {
 508
 509                 vector<string> mapFileNames;
 510
 511                 for (int i = start; i < end; i++) {
 512
 513                         start = time(NULL);
 514
 515                         if (m->control_pressed) {  return mapFileNames; }
 516
 517                         m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[i] + ":"); m->mothurOutEndLine();
 518
 519                         map<string, string> thisNameMap;
 520                         thisNameMap = parser.getNameMap(groups[i]);
 521                         vector<Sequence> thisSeqs = parser.getSeqs(groups[i]);
 522
 523                         vector<string> sequences;
 524                         vector<string> uniqueNames;
 525                         vector<string> redundantNames;
 526                         vector<int> seqFreq;
 527
 528                         seqNoise noise;
 529                         correctDist* correct = new correctDist(1); //we use one processor since we already split up the work load.
 530
 531                         //load this groups info in order
 532                         loadData(correct, noise, sequences, uniqueNames, redundantNames, seqFreq, thisNameMap, thisSeqs);
 533                         if (m->control_pressed) { return mapFileNames; }
 534
 535                         //calc distances for cluster
 536                         string distFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + groups[i] + ".shhh.dist";
 537                         correct->execute(distFileName);
 538                         delete correct;
 539
 540                         if (m->control_pressed) { m->mothurRemove(distFileName); return mapFileNames; }
 541
 542                         driver(noise, sequences, uniqueNames, redundantNames, seqFreq, distFileName, newFFile+groups[i], newNFile+groups[i], newMFile+groups[i]+".map");
 543
 544                         if (m->control_pressed) { return mapFileNames; }
 545
 546                         m->appendFiles(newFFile+groups[i], newFFile); m->mothurRemove(newFFile+groups[i]);
 547                         m->appendFiles(newNFile+groups[i], newNFile); m->mothurRemove(newNFile+groups[i]);
 548                         mapFileNames.push_back(newMFile+groups[i]+".map");
 549
 550                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to process group " + groups[i] + "."); m->mothurOutEndLine();
 551                 }
 552
 553                 return mapFileNames;
 554         }
 555         catch(exception& e) {
 556                 m->errorOut(e, "ShhhSeqsCommand", "driverGroups");
 557                 exit(1);
 558         }
 559 }
 560 //**********************************************************************************************************************
 561 int ShhhSeqsCommand::driver(seqNoise& noise,
 562                                                         vector<string>& sequences,
 563                                                         vector<string>& uniqueNames,
 564                                                         vector<string>& redundantNames,
 565                                                         vector<int>& seqFreq,
 566                                                         string distFileName, string outputFileName, string nameFileName, string mapFileName) {
 567         try {
 568                 double cutOff = 0.08;
 569                 int minIter = 10;
 570                 int maxIter = 1000;
 571                 double minDelta = 1e-6;
 572                 int numIters = 0;
 573                 double maxDelta = 1e6;
 574                 int numSeqs = sequences.size();
 575
 576                 //run cluster command
 577                 string inputString = "phylip=" + distFileName + ", method=furthest, cutoff=0.08";
 578                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 579                 m->mothurOut("Running command: cluster(" + inputString + ")"); m->mothurOutEndLine();
 580
 581                 Command* clusterCommand = new ClusterCommand(inputString);
 582                 clusterCommand->execute();
 583
 584                 map<string, vector<string> > filenames = clusterCommand->getOutputFiles();
 585                 string listFileName = filenames["list"][0];
 586                 string rabundFileName = filenames["rabund"][0]; m->mothurRemove(rabundFileName);
 587                 string sabundFileName = filenames["sabund"][0]; m->mothurRemove(sabundFileName);
 588
 589                 delete clusterCommand;
 590                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 591
 592                 if (m->control_pressed) { m->mothurRemove(distFileName); m->mothurRemove(listFileName); return 0; }
 593
 594                 vector<double> distances(numSeqs * numSeqs);
 595                 noise.getDistanceData(distFileName, distances);
 596                 m->mothurRemove(distFileName);
 597                 if (m->control_pressed) { m->mothurRemove(listFileName); return 0; }
 598
 599                 vector<int> otuData(numSeqs);
 600                 vector<int> otuFreq;
 601                 vector<vector<int> > otuBySeqLookUp;
 602                 noise.getListData(listFileName, cutOff, otuData, otuFreq, otuBySeqLookUp);
 603                 m->mothurRemove(listFileName);
 604                 if (m->control_pressed) { return 0; }
 605
 606                 int numOTUs = otuFreq.size();
 607
 608                 vector<double> weights(numOTUs, 0);
 609                 vector<int> change(numOTUs, 1);
 610                 vector<int> centroids(numOTUs, -1);
 611                 vector<int> cumCount(numOTUs, 0);
 612
 613                 vector<double> tau(numSeqs, 1);
 614                 vector<int> anP(numSeqs, 0);
 615                 vector<int> anI(numSeqs, 0);
 616                 vector<int> anN(numSeqs, 0);
 617                 vector<vector<int> > aanI = otuBySeqLookUp;
 618
 619                 while(numIters < minIter || ((maxDelta > minDelta) && (numIters < maxIter))){
 620
 621                         if (m->control_pressed) { return 0; }
 622
 623                         noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount); if (m->control_pressed) { return 0; }
 624                         maxDelta = noise.calcNewWeights(weights, seqFreq, anI, cumCount, anP, otuFreq, tau);  if (m->control_pressed) { return 0; }
 625
 626                         noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 627                         noise.checkCentroids(weights, centroids); if (m->control_pressed) { return 0; }
 628
 629                         otuFreq.assign(numOTUs, 0);
 630
 631                         int total = 0;
 632
 633                         for(int i=0;i<numSeqs;i++){
 634                                 if (m->control_pressed) { return 0; }
 635
 636                                 double offset = 1e6;
 637                                 double norm = 0.0000;
 638                                 double minWeight = 0.1;
 639                                 vector<double> currentTau(numOTUs);
 640
 641                                 for(int j=0;j<numOTUs;j++){
 642                                         if (m->control_pressed) { return 0; }
 643                                         if(weights[j] > minWeight && distances[i * numSeqs+centroids[j]] < offset){
 644                                                 offset = distances[i * numSeqs+centroids[j]];
 645                                         }
 646                                 }
 647
 648                                 for(int j=0;j<numOTUs;j++){
 649                                         if (m->control_pressed) { return 0; }
 650                                         if(weights[j] > minWeight){
 651                                                 currentTau[j] = exp(sigma * (-distances[(i * numSeqs + centroids[j])] + offset)) * weights[j];
 652                                                 norm += currentTau[j];
 653                                         }
 654                                         else{
 655                                                 currentTau[j] = 0.0000;
 656                                         }
 657                                 }
 658
 659                                 for(int j=0;j<numOTUs;j++){
 660                                         if (m->control_pressed) { return 0; }
 661                                         currentTau[j] /= norm;
 662                                 }
 663
 664                                 for(int j=0;j<numOTUs;j++){
 665                                         if (m->control_pressed) { return 0; }
 666
 667                                         if(currentTau[j] > 1.0e-4){
 668                                                 int oldTotal = total;
 669                                                 total++;
 670
 671                                                 tau.resize(oldTotal+1);
 672                                                 tau[oldTotal] = currentTau[j];
 673                                                 otuBySeqLookUp[j][otuFreq[j]] = oldTotal;
 674                                                 aanI[j][otuFreq[j]] = i;
 675                                                 otuFreq[j]++;
 676
 677                                         }
 678                                 }
 679
 680                                 anP.resize(total);
 681                                 anI.resize(total);
 682                         }
 683
 684                         numIters++;
 685                 }
 686
 687                 noise.updateOTUCountData(otuFreq, otuBySeqLookUp, aanI, anP, anI, cumCount);  if (m->control_pressed) { return 0; }
 688
 689                 vector<double> percentage(numSeqs);
 690                 noise.setUpOTUData(otuData, percentage, cumCount, tau, otuFreq, anP, anI);  if (m->control_pressed) { return 0; }
 691                 noise.finishOTUData(otuData, otuFreq, anP, anI, cumCount, otuBySeqLookUp, aanI, tau);  if (m->control_pressed) { return 0; }
 692
 693                 change.assign(numOTUs, 1);
 694                 noise.calcCentroids(anI, anP, change, centroids, cumCount, distances, seqFreq, otuFreq, tau); if (m->control_pressed) { return 0; }
 695
 696
 697                 vector<int> finalTau(numOTUs, 0);
 698                 for(int i=0;i<numSeqs;i++){
 699                         if (m->control_pressed) { return 0; }
 700                         finalTau[otuData[i]] += int(seqFreq[i]);
 701                 }
 702
 703                 noise.writeOutput(outputFileName, nameFileName, mapFileName, finalTau, centroids, otuData, sequences, uniqueNames, redundantNames, seqFreq, distances);
 704
 705                 return 0;
 706
 707         }catch(exception& e) {
 708                 m->errorOut(e, "ShhhSeqsCommand", "driver");
 709                 exit(1);
 710         }
 711 }
 712 //**********************************************************************************************************************
 713 int ShhhSeqsCommand::deconvoluteResults(string fastaFile, string nameFile){
 714         try {
 715                 m->mothurOutEndLine(); m->mothurOut("Deconvoluting results:"); m->mothurOutEndLine(); m->mothurOutEndLine();
 716
 717                 //use unique.seqs to create new name and fastafile
 718                 string inputString = "fasta=" + fastaFile + ", name=" + nameFile;
 719                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 720                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 721                 m->mothurCalling = true;
 722
 723                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 724                 uniqueCommand->execute();
 725
 726                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 727
 728                 delete uniqueCommand;
 729                 m->mothurCalling = false;
 730                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 731
 732                 string newnameFile = filenames["name"][0];
 733                 string newfastaFile = filenames["fasta"][0];
 734
 735                 m->mothurRemove(fastaFile); rename(newfastaFile.c_str(), fastaFile.c_str());
 736                 if (nameFile != newnameFile) { m->mothurRemove(nameFile); rename(newnameFile.c_str(), nameFile.c_str()); }
 737
 738                 return 0;
 739         }
 740         catch(exception& e) {
 741                 m->errorOut(e, "ShhhSeqsCommand", "deconvoluteResults");
 742                 exit(1);
 743         }
 744 }
 745 //**********************************************************************************************************************
 746
 747
 748