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