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 string MatrixOutputCommand::getOutputFileNameTag(string type, string inputName=""){
  65         try {
  66         string outputFileName = "";
  67                 map<string, vector<string> >::iterator it;
  68
  69         //is this a type this command creates
  70         it = outputTypes.find(type);
  71         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  72         else {
  73             if (type == "phylip")             {   outputFileName =  "dist";         }
  74             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  75         }
  76         return outputFileName;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "MatrixOutputCommand", "getOutputFileNameTag");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84 MatrixOutputCommand::MatrixOutputCommand(){
  85         try {
  86                 abort = true; calledHelp = true;
  87                 setParameters();
  88                 vector<string> tempOutNames;
  89                 outputTypes["phylip"] = tempOutNames;
  90         }
  91         catch(exception& e) {
  92                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
  93                 exit(1);
  94         }
  95 }
  96 //**********************************************************************************************************************
  97
  98 MatrixOutputCommand::MatrixOutputCommand(string option)  {
  99         try {
 100                 abort = false; calledHelp = false;
 101                 allLines = 1;
 102
 103                 //allow user to run help
 104                 if(option == "help") {  help(); abort = true; calledHelp = true; }
 105                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 106
 107                 else {
 108                         vector<string> myArray = setParameters();
 109
 110                         OptionParser parser(option);
 111                         map<string,string> parameters  = parser.getParameters();
 112                         map<string,string>::iterator it;
 113
 114                         ValidParameters validParameter;
 115
 116                         //check to make sure all parameters are valid for command
 117                         for (it = parameters.begin(); it != parameters.end(); it++) {
 118                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 119                         }
 120
 121                         //initialize outputTypes
 122                         vector<string> tempOutNames;
 123                         outputTypes["phylip"] = tempOutNames;
 124
 125                         //if the user changes the input directory command factory will send this info to us in the output parameter
 126                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 127                         if (inputDir == "not found"){   inputDir = "";          }
 128                         else {
 129                                 string path;
 130                                 it = parameters.find("shared");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 136                                 }
 137                         }
 138
 139                         sharedfile = validParameter.validFile(parameters, "shared", true);
 140                         if (sharedfile == "not found") {
 141                                 //if there is a current shared file, use it
 142                                 sharedfile = m->getSharedFile();
 143                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 144                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 145                         }else if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 146                         else { m->setSharedFile(sharedfile); }
 147
 148                         //if the user changes the output directory command factory will send this info to us in the output parameter
 149                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 150                                 outputDir = "";
 151                                 outputDir += m->hasPath(sharedfile); //if user entered a file with a path then preserve it
 152                         }
 153
 154                         //check for optional parameter and set defaults
 155                         // ...at some point should added some additional type checking...
 156                         label = validParameter.validFile(parameters, "label", false);
 157                         if (label == "not found") { label = ""; }
 158                         else {
 159                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 160                                 else { allLines = 1;  }
 161                         }
 162
 163                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
 164                         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"; }
 165
 166             mode = validParameter.validFile(parameters, "mode", false);         if(mode == "not found"){        mode = "average"; }
 167                         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"; }
 168
 169                         groups = validParameter.validFile(parameters, "groups", false);
 170                         if (groups == "not found") { groups = ""; }
 171                         else {
 172                                 m->splitAtDash(groups, Groups);
 173                                 m->setGroups(Groups);
 174                         }
 175
 176                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 177                         m->setProcessors(temp);
 178                         m->mothurConvert(temp, processors);
 179
 180                         calc = validParameter.validFile(parameters, "calc", false);
 181                         if (calc == "not found") { calc = "jclass-thetayc";  }
 182                         else {
 183                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 184                         }
 185                         m->splitAtDash(calc, Estimators);
 186                         if (m->inUsersGroups("citation", Estimators)) {
 187                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 188                                 //remove citation from list of calcs
 189                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 190                         }
 191
 192             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 193                         m->mothurConvert(temp, iters);
 194
 195             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 196                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 197             else {
 198                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 199                 else { subsample = false; }
 200             }
 201
 202             if (subsample == false) { iters = 1; }
 203
 204                         if (abort == false) {
 205
 206                                 ValidCalculators validCalculator;
 207
 208                                 int i;
 209                                 for (i=0; i<Estimators.size(); i++) {
 210                                         if (validCalculator.isValidCalculator("matrix", Estimators[i]) == true) {
 211                                                 if (Estimators[i] == "sharedsobs") {
 212                                                         matrixCalculators.push_back(new SharedSobsCS());
 213                                                 }else if (Estimators[i] == "sharedchao") {
 214                                                         matrixCalculators.push_back(new SharedChao1());
 215                                                 }else if (Estimators[i] == "sharedace") {
 216                                                         matrixCalculators.push_back(new SharedAce());
 217                                                 }else if (Estimators[i] == "jabund") {
 218                                                         matrixCalculators.push_back(new JAbund());
 219                                                 }else if (Estimators[i] == "sorabund") {
 220                                                         matrixCalculators.push_back(new SorAbund());
 221                                                 }else if (Estimators[i] == "jclass") {
 222                                                         matrixCalculators.push_back(new Jclass());
 223                                                 }else if (Estimators[i] == "sorclass") {
 224                                                         matrixCalculators.push_back(new SorClass());
 225                                                 }else if (Estimators[i] == "jest") {
 226                                                         matrixCalculators.push_back(new Jest());
 227                                                 }else if (Estimators[i] == "sorest") {
 228                                                         matrixCalculators.push_back(new SorEst());
 229                                                 }else if (Estimators[i] == "thetayc") {
 230                                                         matrixCalculators.push_back(new ThetaYC());
 231                                                 }else if (Estimators[i] == "thetan") {
 232                                                         matrixCalculators.push_back(new ThetaN());
 233                                                 }else if (Estimators[i] == "kstest") {
 234                                                         matrixCalculators.push_back(new KSTest());
 235                                                 }else if (Estimators[i] == "sharednseqs") {
 236                                                         matrixCalculators.push_back(new SharedNSeqs());
 237                                                 }else if (Estimators[i] == "ochiai") {
 238                                                         matrixCalculators.push_back(new Ochiai());
 239                                                 }else if (Estimators[i] == "anderberg") {
 240                                                         matrixCalculators.push_back(new Anderberg());
 241                                                 }else if (Estimators[i] == "kulczynski") {
 242                                                         matrixCalculators.push_back(new Kulczynski());
 243                                                 }else if (Estimators[i] == "kulczynskicody") {
 244                                                         matrixCalculators.push_back(new KulczynskiCody());
 245                                                 }else if (Estimators[i] == "lennon") {
 246                                                         matrixCalculators.push_back(new Lennon());
 247                                                 }else if (Estimators[i] == "morisitahorn") {
 248                                                         matrixCalculators.push_back(new MorHorn());
 249                                                 }else if (Estimators[i] == "braycurtis") {
 250                                                         matrixCalculators.push_back(new BrayCurtis());
 251                                                 }else if (Estimators[i] == "whittaker") {
 252                                                         matrixCalculators.push_back(new Whittaker());
 253                                                 }else if (Estimators[i] == "odum") {
 254                                                         matrixCalculators.push_back(new Odum());
 255                                                 }else if (Estimators[i] == "canberra") {
 256                                                         matrixCalculators.push_back(new Canberra());
 257                                                 }else if (Estimators[i] == "structeuclidean") {
 258                                                         matrixCalculators.push_back(new StructEuclidean());
 259                                                 }else if (Estimators[i] == "structchord") {
 260                                                         matrixCalculators.push_back(new StructChord());
 261                                                 }else if (Estimators[i] == "hellinger") {
 262                                                         matrixCalculators.push_back(new Hellinger());
 263                                                 }else if (Estimators[i] == "manhattan") {
 264                                                         matrixCalculators.push_back(new Manhattan());
 265                                                 }else if (Estimators[i] == "structpearson") {
 266                                                         matrixCalculators.push_back(new StructPearson());
 267                                                 }else if (Estimators[i] == "soergel") {
 268                                                         matrixCalculators.push_back(new Soergel());
 269                                                 }else if (Estimators[i] == "spearman") {
 270                                                         matrixCalculators.push_back(new Spearman());
 271                                                 }else if (Estimators[i] == "structkulczynski") {
 272                                                         matrixCalculators.push_back(new StructKulczynski());
 273                                                 }else if (Estimators[i] == "speciesprofile") {
 274                                                         matrixCalculators.push_back(new SpeciesProfile());
 275                                                 }else if (Estimators[i] == "hamming") {
 276                                                         matrixCalculators.push_back(new Hamming());
 277                                                 }else if (Estimators[i] == "structchi2") {
 278                                                         matrixCalculators.push_back(new StructChi2());
 279                                                 }else if (Estimators[i] == "gower") {
 280                                                         matrixCalculators.push_back(new Gower());
 281                                                 }else if (Estimators[i] == "memchi2") {
 282                                                         matrixCalculators.push_back(new MemChi2());
 283                                                 }else if (Estimators[i] == "memchord") {
 284                                                         matrixCalculators.push_back(new MemChord());
 285                                                 }else if (Estimators[i] == "memeuclidean") {
 286                                                         matrixCalculators.push_back(new MemEuclidean());
 287                                                 }else if (Estimators[i] == "mempearson") {
 288                                                         matrixCalculators.push_back(new MemPearson());
 289                                                 }
 290                                         }
 291                                 }
 292
 293                         }
 294                 }
 295
 296         }
 297         catch(exception& e) {
 298                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
 299                 exit(1);
 300         }
 301 }
 302
 303 //**********************************************************************************************************************
 304
 305 MatrixOutputCommand::~MatrixOutputCommand(){}
 306
 307 //**********************************************************************************************************************
 308
 309 int MatrixOutputCommand::execute(){
 310         try {
 311
 312                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 313
 314                 //if the users entered no valid calculators don't execute command
 315                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
 316
 317                 input = new InputData(sharedfile, "sharedfile");
 318                 lookup = input->getSharedRAbundVectors();
 319                 string lastLabel = lookup[0]->getLabel();
 320
 321                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 322                 set<string> processedLabels;
 323                 set<string> userLabels = labels;
 324
 325                 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;}
 326
 327         if (subsample) {
 328             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 329                 subsampleSize = lookup[0]->getNumSeqs();
 330                 for (int i = 1; i < lookup.size(); i++) {
 331                     int thisSize = lookup[i]->getNumSeqs();
 332
 333                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 334                 }
 335             }else {
 336                 m->clearGroups();
 337                 Groups.clear();
 338                 vector<SharedRAbundVector*> temp;
 339                 for (int i = 0; i < lookup.size(); i++) {
 340                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 341                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 342                         delete lookup[i];
 343                     }else {
 344                         Groups.push_back(lookup[i]->getGroup());
 345                         temp.push_back(lookup[i]);
 346                     }
 347                 }
 348                 lookup = temp;
 349                 m->setGroups(Groups);
 350             }
 351
 352             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; }
 353         }
 354
 355                 numGroups = lookup.size();
 356         lines.resize(processors);
 357                 for (int i = 0; i < processors; i++) {
 358                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 359                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 360                 }
 361
 362                 if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->clearGroups(); return 0;  }
 363
 364                 //as long as you are not at the end of the file or done wih the lines you want
 365                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 366
 367                         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;  }
 368
 369                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 370                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 371                                 process(lookup);
 372
 373                                 processedLabels.insert(lookup[0]->getLabel());
 374                                 userLabels.erase(lookup[0]->getLabel());
 375                         }
 376
 377                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 378                                 string saveLabel = lookup[0]->getLabel();
 379
 380                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 381                                 lookup = input->getSharedRAbundVectors(lastLabel);
 382
 383                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 384                                 process(lookup);
 385
 386                                 processedLabels.insert(lookup[0]->getLabel());
 387                                 userLabels.erase(lookup[0]->getLabel());
 388
 389                                 //restore real lastlabel to save below
 390                                 lookup[0]->setLabel(saveLabel);
 391                         }
 392
 393                         lastLabel = lookup[0]->getLabel();
 394
 395                         //get next line to process
 396                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 397                         lookup = input->getSharedRAbundVectors();
 398                 }
 399
 400                 if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 401
 402                 //output error messages about any remaining user labels
 403                 set<string>::iterator it;
 404                 bool needToRun = false;
 405                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 406                         m->mothurOut("Your file does not include the label " + *it);
 407                         if (processedLabels.count(lastLabel) != 1) {
 408                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 409                                 needToRun = true;
 410                         }else {
 411                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 412                         }
 413                 }
 414
 415                 if (m->control_pressed) { outputTypes.clear(); delete input;  for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 416
 417                 //run last label if you need to
 418                 if (needToRun == true)  {
 419                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  }
 420                         lookup = input->getSharedRAbundVectors(lastLabel);
 421
 422                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 423                         process(lookup);
 424                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 425                 }
 426
 427                 if (m->control_pressed) { outputTypes.clear();  delete input;  for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 428
 429                 //reset groups parameter
 430                 m->clearGroups();
 431
 432                 //set phylip file as new current phylipfile
 433                 string current = "";
 434                 itTypes = outputTypes.find("phylip");
 435                 if (itTypes != outputTypes.end()) {
 436                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 437                 }
 438
 439                 m->mothurOutEndLine();
 440                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 441                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 442                 m->mothurOutEndLine();
 443
 444
 445                 return 0;
 446         }
 447         catch(exception& e) {
 448                 m->errorOut(e, "MatrixOutputCommand", "execute");
 449                 exit(1);
 450         }
 451 }
 452 /***********************************************************/
 453 void MatrixOutputCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 454         try {
 455
 456                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 457
 458                 //output num seqs
 459                 out << simMatrix.size() << endl;
 460
 461                 if (output == "lt") {
 462                         for (int b = 0; b < simMatrix.size(); b++)      {
 463                                 out << lookup[b]->getGroup() << '\t';
 464                                 for (int n = 0; n < b; n++)     {
 465                                         out << simMatrix[b][n] << '\t';
 466                                 }
 467                                 out << endl;
 468                         }
 469                 }else{
 470                         for (int b = 0; b < simMatrix.size(); b++)      {
 471                                 out << lookup[b]->getGroup() << '\t';
 472                                 for (int n = 0; n < simMatrix[b].size(); n++)   {
 473                                         out << simMatrix[b][n] << '\t';
 474                                 }
 475                                 out << endl;
 476                         }
 477                 }
 478         }
 479         catch(exception& e) {
 480                 m->errorOut(e, "MatrixOutputCommand", "printSims");
 481                 exit(1);
 482         }
 483 }
 484 /***********************************************************/
 485 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
 486         try {
 487                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 488         vector< vector<seqDist>  > calcDists; calcDists.resize(matrixCalculators.size());
 489
 490         for (int thisIter = 0; thisIter < iters+1; thisIter++) {
 491
 492             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 493
 494             if (subsample && (thisIter != 0)) {
 495                 SubSample sample;
 496                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 497
 498                 //make copy of lookup so we don't get access violations
 499                 vector<SharedRAbundVector*> newLookup;
 500                 for (int k = 0; k < thisItersLookup.size(); k++) {
 501                     SharedRAbundVector* temp = new SharedRAbundVector();
 502                     temp->setLabel(thisItersLookup[k]->getLabel());
 503                     temp->setGroup(thisItersLookup[k]->getGroup());
 504                     newLookup.push_back(temp);
 505                 }
 506
 507                 //for each bin
 508                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 509                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 510                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 511                 }
 512
 513                 tempLabels = sample.getSample(newLookup, subsampleSize);
 514                 thisItersLookup = newLookup;
 515             }
 516
 517             if(processors == 1){
 518                 driver(thisItersLookup, 0, numGroups, calcDists);
 519             }else{
 520                 int process = 1;
 521                 vector<int> processIDS;
 522
 523                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 524                 //loop through and create all the processes you want
 525                 while (process != processors) {
 526                     int pid = fork();
 527
 528                     if (pid > 0) {
 529                         processIDS.push_back(pid);
 530                         process++;
 531                     }else if (pid == 0){
 532
 533                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 534
 535                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 536                         ofstream outtemp;
 537                         m->openOutputFile(tempdistFileName, outtemp);
 538
 539                         for (int i = 0; i < calcDists.size(); i++) {
 540                             outtemp << calcDists[i].size() << endl;
 541
 542                             for (int j = 0; j < calcDists[i].size(); j++) {
 543                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 544                             }
 545                         }
 546                         outtemp.close();
 547
 548                         exit(0);
 549                     }else {
 550                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 551                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 552                         exit(0);
 553                     }
 554                 }
 555
 556                 //parent do your part
 557                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 558
 559                 //force parent to wait until all the processes are done
 560                 for (int i = 0; i < processIDS.size(); i++) {
 561                     int temp = processIDS[i];
 562                     wait(&temp);
 563                 }
 564
 565                 for (int i = 0; i < processIDS.size(); i++) {
 566                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 567                     ifstream intemp;
 568                     m->openInputFile(tempdistFileName, intemp);
 569
 570                     for (int k = 0; k < calcDists.size(); k++) {
 571                         int size = 0;
 572                         intemp >> size; m->gobble(intemp);
 573
 574                         for (int j = 0; j < size; j++) {
 575                             int seq1 = 0;
 576                             int seq2 = 0;
 577                             float dist = 1.0;
 578
 579                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 580
 581                             seqDist tempDist(seq1, seq2, dist);
 582                             calcDists[k].push_back(tempDist);
 583                         }
 584                     }
 585                     intemp.close();
 586                     m->mothurRemove(tempdistFileName);
 587                 }
 588                 #else
 589                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 590                 //Windows version shared memory, so be careful when passing variables through the distSharedData struct.
 591                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 592                 //Taking advantage of shared memory to pass results vectors.
 593                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 594
 595                 vector<distSharedData*> pDataArray;
 596                 DWORD   dwThreadIdArray[processors-1];
 597                 HANDLE  hThreadArray[processors-1];
 598
 599                 //Create processor worker threads.
 600                 for( int i=1; i<processors; i++ ){
 601
 602                     //make copy of lookup so we don't get access violations
 603                     vector<SharedRAbundVector*> newLookup;
 604                     for (int k = 0; k < thisItersLookup.size(); k++) {
 605                         SharedRAbundVector* temp = new SharedRAbundVector();
 606                         temp->setLabel(thisItersLookup[k]->getLabel());
 607                         temp->setGroup(thisItersLookup[k]->getGroup());
 608                         newLookup.push_back(temp);
 609                     }
 610
 611                     //for each bin
 612                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 613                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 614                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 615                     }
 616
 617                     // Allocate memory for thread data.
 618                     distSharedData* tempSum = new distSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 619                     pDataArray.push_back(tempSum);
 620                     processIDS.push_back(i);
 621
 622                     hThreadArray[i-1] = CreateThread(NULL, 0, MyDistSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 623                 }
 624
 625                 //parent do your part
 626                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 627
 628                 //Wait until all threads have terminated.
 629                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 630
 631                 //Close all thread handles and free memory allocations.
 632                 for(int i=0; i < pDataArray.size(); i++){
 633                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 634
 635                     for (int k = 0; k < calcDists.size(); k++) {
 636                         int size = pDataArray[i]->calcDists[k].size();
 637                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 638                     }
 639
 640                     CloseHandle(hThreadArray[i]);
 641                     delete pDataArray[i];
 642                 }
 643
 644                 #endif
 645             }
 646
 647             if (subsample && (thisIter != 0)) {
 648                 if((thisIter) % 100 == 0){      m->mothurOut(toString(thisIter)); m->mothurOutEndLine();                }
 649                 calcDistsTotals.push_back(calcDists);
 650                 for (int i = 0; i < calcDists.size(); i++) {
 651                     for (int j = 0; j < calcDists[i].size(); j++) {
 652                         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");  }
 653                     }
 654                 }
 655                 //clean up memory
 656                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 657                 thisItersLookup.clear();
 658             }else { //print results for whole dataset
 659                 for (int i = 0; i < calcDists.size(); i++) {
 660                     if (m->control_pressed) { break; }
 661
 662                     //initialize matrix
 663                     vector< vector<double> > matrix; //square matrix to represent the distance
 664                     matrix.resize(thisLookup.size());
 665                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 666
 667                     for (int j = 0; j < calcDists[i].size(); j++) {
 668                         int row = calcDists[i][j].seq1;
 669                         int column = calcDists[i][j].seq2;
 670                         double dist = calcDists[i][j].dist;
 671
 672                         matrix[row][column] = dist;
 673                         matrix[column][row] = dist;
 674                     }
 675
 676                     string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + "." + getOutputFileNameTag("phylip");
 677                     outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 678                     ofstream outDist;
 679                     m->openOutputFile(distFileName, outDist);
 680                     outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
 681
 682                     printSims(outDist, matrix);
 683
 684                     outDist.close();
 685                 }
 686             }
 687             for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 688                 }
 689
 690         if (iters != 1) {
 691             //we need to find the average distance and standard deviation for each groups distance
 692
 693             vector< vector<seqDist>  > calcAverages; calcAverages.resize(matrixCalculators.size());
 694             for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 695                 calcAverages[i].resize(calcDistsTotals[0][i].size());
 696
 697                 for (int j = 0; j < calcAverages[i].size(); j++) {
 698                     calcAverages[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 699                     calcAverages[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 700                     calcAverages[i][j].dist = 0.0;
 701                 }
 702             }
 703             if (mode == "average") {
 704                 for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 705                     for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 706                         for (int j = 0; j < calcAverages[i].size(); j++) {
 707                             calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
 708                             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");  }
 709                         }
 710                     }
 711                 }
 712
 713                 for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 714                     for (int j = 0; j < calcAverages[i].size(); j++) {
 715                         calcAverages[i][j].dist /= (float) iters;
 716                     }
 717                 }
 718             }else { //find median
 719                 for (int i = 0; i < calcAverages.size(); i++) { //for each calc
 720                     for (int j = 0; j < calcAverages[i].size(); j++) {  //for each comparison
 721                         vector<double> dists;
 722                         for (int thisIter = 0; thisIter < iters; thisIter++) { //for each subsample
 723                             dists.push_back(calcDistsTotals[thisIter][i][j].dist);
 724                         }
 725                         sort(dists.begin(), dists.end());
 726                         calcAverages[i][j].dist = dists[(iters/2)];
 727                     }
 728                 }
 729             }
 730             //find standard deviation
 731             vector< vector<seqDist>  > stdDev; stdDev.resize(matrixCalculators.size());
 732             for (int i = 0; i < stdDev.size(); i++) {  //initialize sums to zero.
 733                 stdDev[i].resize(calcDistsTotals[0][i].size());
 734
 735                 for (int j = 0; j < stdDev[i].size(); j++) {
 736                     stdDev[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 737                     stdDev[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 738                     stdDev[i][j].dist = 0.0;
 739                 }
 740             }
 741
 742             for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
 743                 for (int i = 0; i < stdDev.size(); i++) {
 744                     for (int j = 0; j < stdDev[i].size(); j++) {
 745                         stdDev[i][j].dist += ((calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist) * (calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist));
 746                     }
 747                 }
 748             }
 749
 750             for (int i = 0; i < stdDev.size(); i++) {  //finds average.
 751                 for (int j = 0; j < stdDev[i].size(); j++) {
 752                     stdDev[i][j].dist /= (float) iters;
 753                     stdDev[i][j].dist = sqrt(stdDev[i][j].dist);
 754                 }
 755             }
 756
 757             //print results
 758             for (int i = 0; i < calcDists.size(); i++) {
 759                 vector< vector<double> > matrix; //square matrix to represent the distance
 760                 matrix.resize(thisLookup.size());
 761                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 762
 763                 vector< vector<double> > stdmatrix; //square matrix to represent the stdDev
 764                 stdmatrix.resize(thisLookup.size());
 765                 for (int k = 0; k < thisLookup.size(); k++) {  stdmatrix[k].resize(thisLookup.size(), 0.0); }
 766
 767
 768                 for (int j = 0; j < calcAverages[i].size(); j++) {
 769                     int row = calcAverages[i][j].seq1;
 770                     int column = calcAverages[i][j].seq2;
 771                     float dist = calcAverages[i][j].dist;
 772                     float stdDist = stdDev[i][j].dist;
 773
 774                     matrix[row][column] = dist;
 775                     matrix[column][row] = dist;
 776                     stdmatrix[row][column] = stdDist;
 777                     stdmatrix[column][row] = stdDist;
 778                 }
 779
 780                 string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + ".ave." + getOutputFileNameTag("phylip");
 781                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 782                 ofstream outAve;
 783                 m->openOutputFile(distFileName, outAve);
 784                 outAve.setf(ios::fixed, ios::floatfield); outAve.setf(ios::showpoint);
 785
 786                 printSims(outAve, matrix);
 787
 788                 outAve.close();
 789
 790                 distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel()  + "." + output + ".std." + getOutputFileNameTag("phylip");
 791                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 792                 ofstream outSTD;
 793                 m->openOutputFile(distFileName, outSTD);
 794                 outSTD.setf(ios::fixed, ios::floatfield); outSTD.setf(ios::showpoint);
 795
 796                 printSims(outSTD, stdmatrix);
 797
 798                 outSTD.close();
 799
 800             }
 801         }
 802
 803                 return 0;
 804         }
 805         catch(exception& e) {
 806                 m->errorOut(e, "MatrixOutputCommand", "process");
 807                 exit(1);
 808         }
 809 }
 810 /**************************************************************************************************/
 811 int MatrixOutputCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
 812         try {
 813                 vector<SharedRAbundVector*> subset;
 814
 815                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
 816
 817                         for (int l = 0; l < k; l++) {
 818
 819                                 if (k != l) { //we dont need to similiarity of a groups to itself
 820                                         subset.clear(); //clear out old pair of sharedrabunds
 821                                         //add new pair of sharedrabunds
 822                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 823
 824                                         for(int i=0;i<matrixCalculators.size();i++) {
 825
 826                                                 //if this calc needs all groups to calculate the pair load all groups
 827                                                 if (matrixCalculators[i]->getNeedsAll()) {
 828                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 829                                                         for (int w = 0; w < thisLookup.size(); w++) {
 830                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 831                                                         }
 832                                                 }
 833
 834                                                 vector<double> tempdata = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
 835
 836                                                 if (m->control_pressed) { return 1; }
 837
 838                                                 seqDist temp(l, k, tempdata[0]);
 839                                                 calcDists[i].push_back(temp);
 840                                         }
 841                                 }
 842                         }
 843                 }
 844
 845                 return 0;
 846         }
 847         catch(exception& e) {
 848                 m->errorOut(e, "MatrixOutputCommand", "driver");
 849                 exit(1);
 850         }
 851 }
 852 /***********************************************************/
 853
 854
 855