matrixoutputcommand.cpp

   1 /*
   2  *  matrixoutputcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/20/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "matrixoutputcommand.h"
  11 #include "subsample.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> MatrixOutputCommand::setParameters(){
  15         try {
  16                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pshared);
  17                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  18         CommandParameter psubsample("subsample", "String", "", "", "", "", "",false,false); parameters.push_back(psubsample);
  19                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  20                 CommandParameter pcalc("calc", "Multiple", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-whittaker-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-hamming-structchi2-gower-memchi2-memchord-memeuclidean-mempearson", "jclass-thetayc", "", "", "",true,false); parameters.push_back(pcalc);
  21                 CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput);
  22         CommandParameter pmode("mode", "Multiple", "average-median", "average", "", "", "",false,false); parameters.push_back(pmode);
  23                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  24         CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  25         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  26                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  27
  28                 vector<string> myArray;
  29                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "MatrixOutputCommand", "setParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 string MatrixOutputCommand::getHelpString(){
  39         try {
  40                 string helpString = "";
  41                 ValidCalculators validCalculator;
  42                 helpString += "The dist.shared command parameters are shared, groups, calc, output, processors, subsample, iters, mode, and label.  shared is a required, unless you have a valid current file.\n";
  43                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n";
  44                 helpString += "The group names are separated by dashes. The label parameter allows you to select what distance levels you would like distance matrices created for, and is also separated by dashes.\n";
  45         helpString += "The iters parameter allows you to choose the number of times you would like to run the subsample.\n";
  46         helpString += "The subsample parameter allows you to enter the size pergroup of the sample or you can set subsample=T and mothur will use the size of your smallest group.\n";
  47                 helpString += "The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n";
  48                 helpString += "The output parameter allows you to specify format of your distance matrix. Options are lt, and square. The default is lt.\n";
  49         helpString += "The mode parameter allows you to specify if you want the average or the median values reported when subsampling. Options are average, and median. The default is average.\n";
  50                 helpString += "Example dist.shared(groups=A-B-C, calc=jabund-sorabund).\n";
  51                 helpString += "The default value for groups is all the groups in your groupfile.\n";
  52                 helpString += "The default value for calc is jclass and thetayc.\n";
  53                 helpString += validCalculator.printCalc("matrix");
  54                 helpString += "The dist.shared command outputs a .dist file for each calculator you specify at each distance you choose.\n";
  55                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  56                 return helpString;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "MatrixOutputCommand", "getHelpString");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 MatrixOutputCommand::MatrixOutputCommand(){
  65         try {
  66                 abort = true; calledHelp = true;
  67                 setParameters();
  68                 vector<string> tempOutNames;
  69                 outputTypes["phylip"] = tempOutNames;
  70         }
  71         catch(exception& e) {
  72                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
  73                 exit(1);
  74         }
  75 }
  76 //**********************************************************************************************************************
  77
  78 MatrixOutputCommand::MatrixOutputCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81                 allLines = 1;
  82
  83                 //allow user to run help
  84                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  85                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  86
  87                 else {
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters  = parser.getParameters();
  92                         map<string,string>::iterator it;
  93
  94                         ValidParameters validParameter;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         //initialize outputTypes
 102                         vector<string> tempOutNames;
 103                         outputTypes["phylip"] = tempOutNames;
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("shared");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 116                                 }
 117                         }
 118
 119                         sharedfile = validParameter.validFile(parameters, "shared", true);
 120                         if (sharedfile == "not found") {
 121                                 //if there is a current shared file, use it
 122                                 sharedfile = m->getSharedFile();
 123                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 124                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 125                         }else if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 126                         else { m->setSharedFile(sharedfile); }
 127
 128                         //if the user changes the output directory command factory will send this info to us in the output parameter
 129                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 130                                 outputDir = "";
 131                                 outputDir += m->hasPath(sharedfile); //if user entered a file with a path then preserve it
 132                         }
 133
 134                         //check for optional parameter and set defaults
 135                         // ...at some point should added some additional type checking...
 136                         label = validParameter.validFile(parameters, "label", false);
 137                         if (label == "not found") { label = ""; }
 138                         else {
 139                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 140                                 else { allLines = 1;  }
 141                         }
 142
 143                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
 144                         if ((output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are lt and square. I will use lt."); m->mothurOutEndLine(); output = "lt"; }
 145
 146             mode = validParameter.validFile(parameters, "mode", false);         if(mode == "not found"){        mode = "average"; }
 147                         if ((mode != "average") && (mode != "median")) { m->mothurOut(mode + " is not a valid mode. Options are average and medina. I will use average."); m->mothurOutEndLine(); output = "average"; }
 148
 149                         groups = validParameter.validFile(parameters, "groups", false);
 150                         if (groups == "not found") { groups = ""; }
 151                         else {
 152                                 m->splitAtDash(groups, Groups);
 153                                 m->setGroups(Groups);
 154                         }
 155
 156                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 157                         m->setProcessors(temp);
 158                         m->mothurConvert(temp, processors);
 159
 160                         calc = validParameter.validFile(parameters, "calc", false);
 161                         if (calc == "not found") { calc = "jclass-thetayc";  }
 162                         else {
 163                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 164                         }
 165                         m->splitAtDash(calc, Estimators);
 166                         if (m->inUsersGroups("citation", Estimators)) {
 167                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 168                                 //remove citation from list of calcs
 169                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 170                         }
 171
 172             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 173                         m->mothurConvert(temp, iters);
 174
 175             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 176                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 177             else {
 178                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 179                 else { subsample = false; }
 180             }
 181
 182             if (subsample == false) { iters = 1; }
 183
 184                         if (abort == false) {
 185
 186                                 ValidCalculators validCalculator;
 187
 188                                 int i;
 189                                 for (i=0; i<Estimators.size(); i++) {
 190                                         if (validCalculator.isValidCalculator("matrix", Estimators[i]) == true) {
 191                                                 if (Estimators[i] == "sharedsobs") {
 192                                                         matrixCalculators.push_back(new SharedSobsCS());
 193                                                 }else if (Estimators[i] == "sharedchao") {
 194                                                         matrixCalculators.push_back(new SharedChao1());
 195                                                 }else if (Estimators[i] == "sharedace") {
 196                                                         matrixCalculators.push_back(new SharedAce());
 197                                                 }else if (Estimators[i] == "jabund") {
 198                                                         matrixCalculators.push_back(new JAbund());
 199                                                 }else if (Estimators[i] == "sorabund") {
 200                                                         matrixCalculators.push_back(new SorAbund());
 201                                                 }else if (Estimators[i] == "jclass") {
 202                                                         matrixCalculators.push_back(new Jclass());
 203                                                 }else if (Estimators[i] == "sorclass") {
 204                                                         matrixCalculators.push_back(new SorClass());
 205                                                 }else if (Estimators[i] == "jest") {
 206                                                         matrixCalculators.push_back(new Jest());
 207                                                 }else if (Estimators[i] == "sorest") {
 208                                                         matrixCalculators.push_back(new SorEst());
 209                                                 }else if (Estimators[i] == "thetayc") {
 210                                                         matrixCalculators.push_back(new ThetaYC());
 211                                                 }else if (Estimators[i] == "thetan") {
 212                                                         matrixCalculators.push_back(new ThetaN());
 213                                                 }else if (Estimators[i] == "kstest") {
 214                                                         matrixCalculators.push_back(new KSTest());
 215                                                 }else if (Estimators[i] == "sharednseqs") {
 216                                                         matrixCalculators.push_back(new SharedNSeqs());
 217                                                 }else if (Estimators[i] == "ochiai") {
 218                                                         matrixCalculators.push_back(new Ochiai());
 219                                                 }else if (Estimators[i] == "anderberg") {
 220                                                         matrixCalculators.push_back(new Anderberg());
 221                                                 }else if (Estimators[i] == "kulczynski") {
 222                                                         matrixCalculators.push_back(new Kulczynski());
 223                                                 }else if (Estimators[i] == "kulczynskicody") {
 224                                                         matrixCalculators.push_back(new KulczynskiCody());
 225                                                 }else if (Estimators[i] == "lennon") {
 226                                                         matrixCalculators.push_back(new Lennon());
 227                                                 }else if (Estimators[i] == "morisitahorn") {
 228                                                         matrixCalculators.push_back(new MorHorn());
 229                                                 }else if (Estimators[i] == "braycurtis") {
 230                                                         matrixCalculators.push_back(new BrayCurtis());
 231                                                 }else if (Estimators[i] == "whittaker") {
 232                                                         matrixCalculators.push_back(new Whittaker());
 233                                                 }else if (Estimators[i] == "odum") {
 234                                                         matrixCalculators.push_back(new Odum());
 235                                                 }else if (Estimators[i] == "canberra") {
 236                                                         matrixCalculators.push_back(new Canberra());
 237                                                 }else if (Estimators[i] == "structeuclidean") {
 238                                                         matrixCalculators.push_back(new StructEuclidean());
 239                                                 }else if (Estimators[i] == "structchord") {
 240                                                         matrixCalculators.push_back(new StructChord());
 241                                                 }else if (Estimators[i] == "hellinger") {
 242                                                         matrixCalculators.push_back(new Hellinger());
 243                                                 }else if (Estimators[i] == "manhattan") {
 244                                                         matrixCalculators.push_back(new Manhattan());
 245                                                 }else if (Estimators[i] == "structpearson") {
 246                                                         matrixCalculators.push_back(new StructPearson());
 247                                                 }else if (Estimators[i] == "soergel") {
 248                                                         matrixCalculators.push_back(new Soergel());
 249                                                 }else if (Estimators[i] == "spearman") {
 250                                                         matrixCalculators.push_back(new Spearman());
 251                                                 }else if (Estimators[i] == "structkulczynski") {
 252                                                         matrixCalculators.push_back(new StructKulczynski());
 253                                                 }else if (Estimators[i] == "speciesprofile") {
 254                                                         matrixCalculators.push_back(new SpeciesProfile());
 255                                                 }else if (Estimators[i] == "hamming") {
 256                                                         matrixCalculators.push_back(new Hamming());
 257                                                 }else if (Estimators[i] == "structchi2") {
 258                                                         matrixCalculators.push_back(new StructChi2());
 259                                                 }else if (Estimators[i] == "gower") {
 260                                                         matrixCalculators.push_back(new Gower());
 261                                                 }else if (Estimators[i] == "memchi2") {
 262                                                         matrixCalculators.push_back(new MemChi2());
 263                                                 }else if (Estimators[i] == "memchord") {
 264                                                         matrixCalculators.push_back(new MemChord());
 265                                                 }else if (Estimators[i] == "memeuclidean") {
 266                                                         matrixCalculators.push_back(new MemEuclidean());
 267                                                 }else if (Estimators[i] == "mempearson") {
 268                                                         matrixCalculators.push_back(new MemPearson());
 269                                                 }
 270                                         }
 271                                 }
 272
 273                         }
 274                 }
 275
 276         }
 277         catch(exception& e) {
 278                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
 279                 exit(1);
 280         }
 281 }
 282
 283 //**********************************************************************************************************************
 284
 285 MatrixOutputCommand::~MatrixOutputCommand(){}
 286
 287 //**********************************************************************************************************************
 288
 289 int MatrixOutputCommand::execute(){
 290         try {
 291
 292                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 293
 294                 //if the users entered no valid calculators don't execute command
 295                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
 296
 297                 input = new InputData(sharedfile, "sharedfile");
 298                 lookup = input->getSharedRAbundVectors();
 299                 string lastLabel = lookup[0]->getLabel();
 300
 301                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 302                 set<string> processedLabels;
 303                 set<string> userLabels = labels;
 304
 305                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } return 0;}
 306
 307         if (subsample) {
 308             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 309                 subsampleSize = lookup[0]->getNumSeqs();
 310                 for (int i = 1; i < lookup.size(); i++) {
 311                     int thisSize = lookup[i]->getNumSeqs();
 312
 313                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 314                 }
 315             }else {
 316                 m->clearGroups();
 317                 Groups.clear();
 318                 vector<SharedRAbundVector*> temp;
 319                 for (int i = 0; i < lookup.size(); i++) {
 320                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 321                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 322                         delete lookup[i];
 323                     }else {
 324                         Groups.push_back(lookup[i]->getGroup());
 325                         temp.push_back(lookup[i]);
 326                     }
 327                 }
 328                 lookup = temp;
 329                 m->setGroups(Groups);
 330             }
 331
 332             if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); m->control_pressed = true; delete input; return 0; }
 333         }
 334
 335                 numGroups = lookup.size();
 336         lines.resize(processors);
 337                 for (int i = 0; i < processors; i++) {
 338                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 339                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 340                 }
 341
 342                 if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->clearGroups(); return 0;  }
 343
 344                 //as long as you are not at the end of the file or done wih the lines you want
 345                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 346
 347                         if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 348
 349                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 350                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 351                                 process(lookup);
 352
 353                                 processedLabels.insert(lookup[0]->getLabel());
 354                                 userLabels.erase(lookup[0]->getLabel());
 355                         }
 356
 357                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 358                                 string saveLabel = lookup[0]->getLabel();
 359
 360                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 361                                 lookup = input->getSharedRAbundVectors(lastLabel);
 362
 363                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 364                                 process(lookup);
 365
 366                                 processedLabels.insert(lookup[0]->getLabel());
 367                                 userLabels.erase(lookup[0]->getLabel());
 368
 369                                 //restore real lastlabel to save below
 370                                 lookup[0]->setLabel(saveLabel);
 371                         }
 372
 373                         lastLabel = lookup[0]->getLabel();
 374
 375                         //get next line to process
 376                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 377                         lookup = input->getSharedRAbundVectors();
 378                 }
 379
 380                 if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 381
 382                 //output error messages about any remaining user labels
 383                 set<string>::iterator it;
 384                 bool needToRun = false;
 385                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 386                         m->mothurOut("Your file does not include the label " + *it);
 387                         if (processedLabels.count(lastLabel) != 1) {
 388                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 389                                 needToRun = true;
 390                         }else {
 391                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 392                         }
 393                 }
 394
 395                 if (m->control_pressed) { outputTypes.clear(); delete input;  for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 396
 397                 //run last label if you need to
 398                 if (needToRun == true)  {
 399                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  }
 400                         lookup = input->getSharedRAbundVectors(lastLabel);
 401
 402                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 403                         process(lookup);
 404                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 405                 }
 406
 407                 if (m->control_pressed) { outputTypes.clear();  delete input;  for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 408
 409                 //reset groups parameter
 410                 m->clearGroups();
 411
 412                 //set phylip file as new current phylipfile
 413                 string current = "";
 414                 itTypes = outputTypes.find("phylip");
 415                 if (itTypes != outputTypes.end()) {
 416                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 417                 }
 418
 419                 m->mothurOutEndLine();
 420                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 421                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 422                 m->mothurOutEndLine();
 423
 424
 425                 return 0;
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "MatrixOutputCommand", "execute");
 429                 exit(1);
 430         }
 431 }
 432 /***********************************************************/
 433 void MatrixOutputCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 434         try {
 435
 436                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 437
 438                 //output num seqs
 439                 out << simMatrix.size() << endl;
 440
 441                 if (output == "lt") {
 442                         for (int b = 0; b < simMatrix.size(); b++)      {
 443                                 out << lookup[b]->getGroup() << '\t';
 444                                 for (int n = 0; n < b; n++)     {
 445                                         out << simMatrix[b][n] << '\t';
 446                                 }
 447                                 out << endl;
 448                         }
 449                 }else{
 450                         for (int b = 0; b < simMatrix.size(); m++)      {
 451                                 out << lookup[b]->getGroup() << '\t';
 452                                 for (int n = 0; n < simMatrix[b].size(); n++)   {
 453                                         out << simMatrix[b][n] << '\t';
 454                                 }
 455                                 out << endl;
 456                         }
 457                 }
 458         }
 459         catch(exception& e) {
 460                 m->errorOut(e, "MatrixOutputCommand", "printSims");
 461                 exit(1);
 462         }
 463 }
 464 /***********************************************************/
 465 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
 466         try {
 467                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 468         vector< vector<seqDist>  > calcDists; calcDists.resize(matrixCalculators.size());
 469
 470         for (int thisIter = 0; thisIter < iters+1; thisIter++) {
 471
 472             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 473
 474             if (subsample && (thisIter != 0)) {
 475                 SubSample sample;
 476                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 477
 478                 //make copy of lookup so we don't get access violations
 479                 vector<SharedRAbundVector*> newLookup;
 480                 for (int k = 0; k < thisItersLookup.size(); k++) {
 481                     SharedRAbundVector* temp = new SharedRAbundVector();
 482                     temp->setLabel(thisItersLookup[k]->getLabel());
 483                     temp->setGroup(thisItersLookup[k]->getGroup());
 484                     newLookup.push_back(temp);
 485                 }
 486
 487                 //for each bin
 488                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 489                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 490                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 491                 }
 492
 493                 tempLabels = sample.getSample(newLookup, subsampleSize);
 494                 thisItersLookup = newLookup;
 495             }
 496
 497             if(processors == 1){
 498                 driver(thisItersLookup, 0, numGroups, calcDists);
 499             }else{
 500                 int process = 1;
 501                 vector<int> processIDS;
 502
 503                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 504                 //loop through and create all the processes you want
 505                 while (process != processors) {
 506                     int pid = fork();
 507
 508                     if (pid > 0) {
 509                         processIDS.push_back(pid);
 510                         process++;
 511                     }else if (pid == 0){
 512
 513                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 514
 515                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 516                         ofstream outtemp;
 517                         m->openOutputFile(tempdistFileName, outtemp);
 518
 519                         for (int i = 0; i < calcDists.size(); i++) {
 520                             outtemp << calcDists[i].size() << endl;
 521
 522                             for (int j = 0; j < calcDists[i].size(); j++) {
 523                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 524                             }
 525                         }
 526                         outtemp.close();
 527
 528                         exit(0);
 529                     }else {
 530                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 531                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 532                         exit(0);
 533                     }
 534                 }
 535
 536                 //parent do your part
 537                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 538
 539                 //force parent to wait until all the processes are done
 540                 for (int i = 0; i < processIDS.size(); i++) {
 541                     int temp = processIDS[i];
 542                     wait(&temp);
 543                 }
 544
 545                 for (int i = 0; i < processIDS.size(); i++) {
 546                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 547                     ifstream intemp;
 548                     m->openInputFile(tempdistFileName, intemp);
 549
 550                     for (int k = 0; k < calcDists.size(); k++) {
 551                         int size = 0;
 552                         intemp >> size; m->gobble(intemp);
 553
 554                         for (int j = 0; j < size; j++) {
 555                             int seq1 = 0;
 556                             int seq2 = 0;
 557                             float dist = 1.0;
 558
 559                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 560
 561                             seqDist tempDist(seq1, seq2, dist);
 562                             calcDists[k].push_back(tempDist);
 563                         }
 564                     }
 565                     intemp.close();
 566                     m->mothurRemove(tempdistFileName);
 567                 }
 568                 #else
 569                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 570                 //Windows version shared memory, so be careful when passing variables through the distSharedData struct.
 571                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 572                 //Taking advantage of shared memory to pass results vectors.
 573                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 574
 575                 vector<distSharedData*> pDataArray;
 576                 DWORD   dwThreadIdArray[processors-1];
 577                 HANDLE  hThreadArray[processors-1];
 578
 579                 //Create processor worker threads.
 580                 for( int i=1; i<processors; i++ ){
 581
 582                     //make copy of lookup so we don't get access violations
 583                     vector<SharedRAbundVector*> newLookup;
 584                     for (int k = 0; k < thisItersLookup.size(); k++) {
 585                         SharedRAbundVector* temp = new SharedRAbundVector();
 586                         temp->setLabel(thisItersLookup[k]->getLabel());
 587                         temp->setGroup(thisItersLookup[k]->getGroup());
 588                         newLookup.push_back(temp);
 589                     }
 590
 591                     //for each bin
 592                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 593                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 594                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 595                     }
 596
 597                     // Allocate memory for thread data.
 598                     distSharedData* tempSum = new distSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 599                     pDataArray.push_back(tempSum);
 600                     processIDS.push_back(i);
 601
 602                     hThreadArray[i-1] = CreateThread(NULL, 0, MyDistSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 603                 }
 604
 605                 //parent do your part
 606                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 607
 608                 //Wait until all threads have terminated.
 609                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 610
 611                 //Close all thread handles and free memory allocations.
 612                 for(int i=0; i < pDataArray.size(); i++){
 613                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 614
 615                     for (int k = 0; k < calcDists.size(); k++) {
 616                         int size = pDataArray[i]->calcDists[k].size();
 617                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 618                     }
 619
 620                     CloseHandle(hThreadArray[i]);
 621                     delete pDataArray[i];
 622                 }
 623
 624                 #endif
 625             }
 626
 627             if (subsample && (thisIter != 0)) {
 628                 if((thisIter) % 100 == 0){      m->mothurOut(toString(thisIter)); m->mothurOutEndLine();                }
 629                 calcDistsTotals.push_back(calcDists);
 630                 for (int i = 0; i < calcDists.size(); i++) {
 631                     for (int j = 0; j < calcDists[i].size(); j++) {
 632                         if (m->debug) {  m->mothurOut("[DEBUG]: Results: iter = " + toString(thisIter) + ", " + thisLookup[calcDists[i][j].seq1]->getGroup() + " - " + thisLookup[calcDists[i][j].seq2]->getGroup() + " distance = " + toString(calcDists[i][j].dist) + ".\n");  }
 633                     }
 634                 }
 635                 //clean up memory
 636                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 637                 thisItersLookup.clear();
 638             }else { //print results for whole dataset
 639                 for (int i = 0; i < calcDists.size(); i++) {
 640                     if (m->control_pressed) { break; }
 641
 642                     //initialize matrix
 643                     vector< vector<double> > matrix; //square matrix to represent the distance
 644                     matrix.resize(thisLookup.size());
 645                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 646
 647                     for (int j = 0; j < calcDists[i].size(); j++) {
 648                         int row = calcDists[i][j].seq1;
 649                         int column = calcDists[i][j].seq2;
 650                         double dist = calcDists[i][j].dist;
 651
 652                         matrix[row][column] = dist;
 653                         matrix[column][row] = dist;
 654                     }
 655
 656                     string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + ".dist";
 657                     outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 658                     ofstream outDist;
 659                     m->openOutputFile(distFileName, outDist);
 660                     outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
 661
 662                     printSims(outDist, matrix);
 663
 664                     outDist.close();
 665                 }
 666             }
 667             for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 668                 }
 669
 670         if (iters != 1) {
 671             //we need to find the average distance and standard deviation for each groups distance
 672
 673             vector< vector<seqDist>  > calcAverages; calcAverages.resize(matrixCalculators.size());
 674             for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 675                 calcAverages[i].resize(calcDistsTotals[0][i].size());
 676
 677                 for (int j = 0; j < calcAverages[i].size(); j++) {
 678                     calcAverages[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 679                     calcAverages[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 680                     calcAverages[i][j].dist = 0.0;
 681                 }
 682             }
 683             if (mode == "average") {
 684                 for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 685                     for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 686                         for (int j = 0; j < calcAverages[i].size(); j++) {
 687                             calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
 688                             if (m->debug) {  m->mothurOut("[DEBUG]: Totaling for average calc: iter = " + toString(thisIter) + ", " + thisLookup[calcDistsTotals[thisIter][i][j].seq1]->getGroup() + " - " + thisLookup[calcDistsTotals[thisIter][i][j].seq2]->getGroup() + " distance = " + toString(calcDistsTotals[thisIter][i][j].dist) + ". New total = " + toString(calcAverages[i][j].dist) + ".\n");  }
 689                         }
 690                     }
 691                 }
 692
 693                 for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 694                     for (int j = 0; j < calcAverages[i].size(); j++) {
 695                         calcAverages[i][j].dist /= (float) iters;
 696                     }
 697                 }
 698             }else { //find median
 699                 for (int i = 0; i < calcAverages.size(); i++) { //for each calc
 700                     for (int j = 0; j < calcAverages[i].size(); j++) {  //for each comparison
 701                         vector<double> dists;
 702                         for (int thisIter = 0; thisIter < iters; thisIter++) { //for each subsample
 703                             dists.push_back(calcDistsTotals[thisIter][i][j].dist);
 704                         }
 705                         sort(dists.begin(), dists.end());
 706                         calcAverages[i][j].dist = dists[(iters/2)];
 707                     }
 708                 }
 709             }
 710             //find standard deviation
 711             vector< vector<seqDist>  > stdDev; stdDev.resize(matrixCalculators.size());
 712             for (int i = 0; i < stdDev.size(); i++) {  //initialize sums to zero.
 713                 stdDev[i].resize(calcDistsTotals[0][i].size());
 714
 715                 for (int j = 0; j < stdDev[i].size(); j++) {
 716                     stdDev[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 717                     stdDev[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 718                     stdDev[i][j].dist = 0.0;
 719                 }
 720             }
 721
 722             for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
 723                 for (int i = 0; i < stdDev.size(); i++) {
 724                     for (int j = 0; j < stdDev[i].size(); j++) {
 725                         stdDev[i][j].dist += ((calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist) * (calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist));
 726                     }
 727                 }
 728             }
 729
 730             for (int i = 0; i < stdDev.size(); i++) {  //finds average.
 731                 for (int j = 0; j < stdDev[i].size(); j++) {
 732                     stdDev[i][j].dist /= (float) iters;
 733                     stdDev[i][j].dist = sqrt(stdDev[i][j].dist);
 734                 }
 735             }
 736
 737             //print results
 738             for (int i = 0; i < calcDists.size(); i++) {
 739                 vector< vector<double> > matrix; //square matrix to represent the distance
 740                 matrix.resize(thisLookup.size());
 741                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 742
 743                 vector< vector<double> > stdmatrix; //square matrix to represent the stdDev
 744                 stdmatrix.resize(thisLookup.size());
 745                 for (int k = 0; k < thisLookup.size(); k++) {  stdmatrix[k].resize(thisLookup.size(), 0.0); }
 746
 747
 748                 for (int j = 0; j < calcAverages[i].size(); j++) {
 749                     int row = calcAverages[i][j].seq1;
 750                     int column = calcAverages[i][j].seq2;
 751                     float dist = calcAverages[i][j].dist;
 752                     float stdDist = stdDev[i][j].dist;
 753
 754                     matrix[row][column] = dist;
 755                     matrix[column][row] = dist;
 756                     stdmatrix[row][column] = stdDist;
 757                     stdmatrix[column][row] = stdDist;
 758                 }
 759
 760                 string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + ".ave.dist";
 761                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 762                 ofstream outAve;
 763                 m->openOutputFile(distFileName, outAve);
 764                 outAve.setf(ios::fixed, ios::floatfield); outAve.setf(ios::showpoint);
 765
 766                 printSims(outAve, matrix);
 767
 768                 outAve.close();
 769
 770                 distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + ".std.dist";
 771                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 772                 ofstream outSTD;
 773                 m->openOutputFile(distFileName, outSTD);
 774                 outSTD.setf(ios::fixed, ios::floatfield); outSTD.setf(ios::showpoint);
 775
 776                 printSims(outSTD, stdmatrix);
 777
 778                 outSTD.close();
 779
 780             }
 781         }
 782
 783                 return 0;
 784         }
 785         catch(exception& e) {
 786                 m->errorOut(e, "MatrixOutputCommand", "process");
 787                 exit(1);
 788         }
 789 }
 790 /**************************************************************************************************/
 791 int MatrixOutputCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
 792         try {
 793                 vector<SharedRAbundVector*> subset;
 794
 795                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
 796
 797                         for (int l = 0; l < k; l++) {
 798
 799                                 if (k != l) { //we dont need to similiarity of a groups to itself
 800                                         subset.clear(); //clear out old pair of sharedrabunds
 801                                         //add new pair of sharedrabunds
 802                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 803
 804                                         for(int i=0;i<matrixCalculators.size();i++) {
 805
 806                                                 //if this calc needs all groups to calculate the pair load all groups
 807                                                 if (matrixCalculators[i]->getNeedsAll()) {
 808                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 809                                                         for (int w = 0; w < thisLookup.size(); w++) {
 810                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 811                                                         }
 812                                                 }
 813
 814                                                 vector<double> tempdata = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
 815
 816                                                 if (m->control_pressed) { return 1; }
 817
 818                                                 seqDist temp(l, k, tempdata[0]);
 819                                                 calcDists[i].push_back(temp);
 820                                         }
 821                                 }
 822                         }
 823                 }
 824
 825                 return 0;
 826         }
 827         catch(exception& e) {
 828                 m->errorOut(e, "MatrixOutputCommand", "driver");
 829                 exit(1);
 830         }
 831 }
 832 /***********************************************************/
 833
 834
 835