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","phylip",false,true,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,true); 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,true); 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::getOutputPattern(string type) {
  65     try {
  66         string pattern = "";
  67
  68         if (type == "phylip") {  pattern = "[filename],[calc],[distance],[outputtag],dist-[filename],[calc],[distance],[outputtag],[tag2],dist"; }
  69         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  70
  71         return pattern;
  72     }
  73     catch(exception& e) {
  74         m->errorOut(e, "MatrixOutputCommand", "getOutputPattern");
  75         exit(1);
  76     }
  77 }
  78 //**********************************************************************************************************************
  79 MatrixOutputCommand::MatrixOutputCommand(){
  80         try {
  81                 abort = true; calledHelp = true;
  82                 setParameters();
  83                 vector<string> tempOutNames;
  84                 outputTypes["phylip"] = tempOutNames;
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
  88                 exit(1);
  89         }
  90 }
  91 //**********************************************************************************************************************
  92
  93 MatrixOutputCommand::MatrixOutputCommand(string option)  {
  94         try {
  95                 abort = false; calledHelp = false;
  96                 allLines = 1;
  97
  98                 //allow user to run help
  99                 if(option == "help") {  help(); abort = true; calledHelp = true; }
 100                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 101
 102                 else {
 103                         vector<string> myArray = setParameters();
 104
 105                         OptionParser parser(option);
 106                         map<string,string> parameters  = parser.getParameters();
 107                         map<string,string>::iterator it;
 108
 109                         ValidParameters validParameter;
 110
 111                         //check to make sure all parameters are valid for command
 112                         for (it = parameters.begin(); it != parameters.end(); it++) {
 113                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 114                         }
 115
 116                         //initialize outputTypes
 117                         vector<string> tempOutNames;
 118                         outputTypes["phylip"] = tempOutNames;
 119
 120                         //if the user changes the input directory command factory will send this info to us in the output parameter
 121                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 122                         if (inputDir == "not found"){   inputDir = "";          }
 123                         else {
 124                                 string path;
 125                                 it = parameters.find("shared");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 131                                 }
 132                         }
 133
 134                         sharedfile = validParameter.validFile(parameters, "shared", true);
 135                         if (sharedfile == "not found") {
 136                                 //if there is a current shared file, use it
 137                                 sharedfile = m->getSharedFile();
 138                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 139                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 140                         }else if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 141                         else { m->setSharedFile(sharedfile); }
 142
 143                         //if the user changes the output directory command factory will send this info to us in the output parameter
 144                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 145                                 outputDir = "";
 146                                 outputDir += m->hasPath(sharedfile); //if user entered a file with a path then preserve it
 147                         }
 148
 149                         //check for optional parameter and set defaults
 150                         // ...at some point should added some additional type checking...
 151                         label = validParameter.validFile(parameters, "label", false);
 152                         if (label == "not found") { label = ""; }
 153                         else {
 154                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 155                                 else { allLines = 1;  }
 156                         }
 157
 158                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
 159                         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"; }
 160
 161             mode = validParameter.validFile(parameters, "mode", false);         if(mode == "not found"){        mode = "average"; }
 162                         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"; }
 163
 164                         groups = validParameter.validFile(parameters, "groups", false);
 165                         if (groups == "not found") { groups = ""; }
 166                         else {
 167                                 m->splitAtDash(groups, Groups);
 168                                 m->setGroups(Groups);
 169                         }
 170
 171                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 172                         m->setProcessors(temp);
 173                         m->mothurConvert(temp, processors);
 174
 175                         calc = validParameter.validFile(parameters, "calc", false);
 176                         if (calc == "not found") { calc = "jclass-thetayc";  }
 177                         else {
 178                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 179                         }
 180                         m->splitAtDash(calc, Estimators);
 181                         if (m->inUsersGroups("citation", Estimators)) {
 182                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 183                                 //remove citation from list of calcs
 184                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 185                         }
 186
 187             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 188                         m->mothurConvert(temp, iters);
 189
 190             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 191                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 192             else {
 193                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 194                 else { subsample = false; }
 195             }
 196
 197             if (subsample == false) { iters = 0; }
 198
 199                         if (abort == false) {
 200
 201                                 ValidCalculators validCalculator;
 202
 203                                 int i;
 204                                 for (i=0; i<Estimators.size(); i++) {
 205                                         if (validCalculator.isValidCalculator("matrix", Estimators[i]) == true) {
 206                                                 if (Estimators[i] == "sharedsobs") {
 207                                                         matrixCalculators.push_back(new SharedSobsCS());
 208                                                 }else if (Estimators[i] == "sharedchao") {
 209                                                         matrixCalculators.push_back(new SharedChao1());
 210                                                 }else if (Estimators[i] == "sharedace") {
 211                                                         matrixCalculators.push_back(new SharedAce());
 212                                                 }else if (Estimators[i] == "jabund") {
 213                                                         matrixCalculators.push_back(new JAbund());
 214                                                 }else if (Estimators[i] == "sorabund") {
 215                                                         matrixCalculators.push_back(new SorAbund());
 216                                                 }else if (Estimators[i] == "jclass") {
 217                                                         matrixCalculators.push_back(new Jclass());
 218                                                 }else if (Estimators[i] == "sorclass") {
 219                                                         matrixCalculators.push_back(new SorClass());
 220                                                 }else if (Estimators[i] == "jest") {
 221                                                         matrixCalculators.push_back(new Jest());
 222                                                 }else if (Estimators[i] == "sorest") {
 223                                                         matrixCalculators.push_back(new SorEst());
 224                                                 }else if (Estimators[i] == "thetayc") {
 225                                                         matrixCalculators.push_back(new ThetaYC());
 226                                                 }else if (Estimators[i] == "thetan") {
 227                                                         matrixCalculators.push_back(new ThetaN());
 228                                                 }else if (Estimators[i] == "kstest") {
 229                                                         matrixCalculators.push_back(new KSTest());
 230                                                 }else if (Estimators[i] == "sharednseqs") {
 231                                                         matrixCalculators.push_back(new SharedNSeqs());
 232                                                 }else if (Estimators[i] == "ochiai") {
 233                                                         matrixCalculators.push_back(new Ochiai());
 234                                                 }else if (Estimators[i] == "anderberg") {
 235                                                         matrixCalculators.push_back(new Anderberg());
 236                                                 }else if (Estimators[i] == "kulczynski") {
 237                                                         matrixCalculators.push_back(new Kulczynski());
 238                                                 }else if (Estimators[i] == "kulczynskicody") {
 239                                                         matrixCalculators.push_back(new KulczynskiCody());
 240                                                 }else if (Estimators[i] == "lennon") {
 241                                                         matrixCalculators.push_back(new Lennon());
 242                                                 }else if (Estimators[i] == "morisitahorn") {
 243                                                         matrixCalculators.push_back(new MorHorn());
 244                                                 }else if (Estimators[i] == "braycurtis") {
 245                                                         matrixCalculators.push_back(new BrayCurtis());
 246                                                 }else if (Estimators[i] == "whittaker") {
 247                                                         matrixCalculators.push_back(new Whittaker());
 248                                                 }else if (Estimators[i] == "odum") {
 249                                                         matrixCalculators.push_back(new Odum());
 250                                                 }else if (Estimators[i] == "canberra") {
 251                                                         matrixCalculators.push_back(new Canberra());
 252                                                 }else if (Estimators[i] == "structeuclidean") {
 253                                                         matrixCalculators.push_back(new StructEuclidean());
 254                                                 }else if (Estimators[i] == "structchord") {
 255                                                         matrixCalculators.push_back(new StructChord());
 256                                                 }else if (Estimators[i] == "hellinger") {
 257                                                         matrixCalculators.push_back(new Hellinger());
 258                                                 }else if (Estimators[i] == "manhattan") {
 259                                                         matrixCalculators.push_back(new Manhattan());
 260                                                 }else if (Estimators[i] == "structpearson") {
 261                                                         matrixCalculators.push_back(new StructPearson());
 262                                                 }else if (Estimators[i] == "soergel") {
 263                                                         matrixCalculators.push_back(new Soergel());
 264                                                 }else if (Estimators[i] == "spearman") {
 265                                                         matrixCalculators.push_back(new Spearman());
 266                                                 }else if (Estimators[i] == "structkulczynski") {
 267                                                         matrixCalculators.push_back(new StructKulczynski());
 268                                                 }else if (Estimators[i] == "speciesprofile") {
 269                                                         matrixCalculators.push_back(new SpeciesProfile());
 270                                                 }else if (Estimators[i] == "hamming") {
 271                                                         matrixCalculators.push_back(new Hamming());
 272                                                 }else if (Estimators[i] == "structchi2") {
 273                                                         matrixCalculators.push_back(new StructChi2());
 274                                                 }else if (Estimators[i] == "gower") {
 275                                                         matrixCalculators.push_back(new Gower());
 276                                                 }else if (Estimators[i] == "memchi2") {
 277                                                         matrixCalculators.push_back(new MemChi2());
 278                                                 }else if (Estimators[i] == "memchord") {
 279                                                         matrixCalculators.push_back(new MemChord());
 280                                                 }else if (Estimators[i] == "memeuclidean") {
 281                                                         matrixCalculators.push_back(new MemEuclidean());
 282                                                 }else if (Estimators[i] == "mempearson") {
 283                                                         matrixCalculators.push_back(new MemPearson());
 284                                                 }
 285                                         }
 286                                 }
 287
 288                         }
 289                 }
 290
 291         }
 292         catch(exception& e) {
 293                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
 294                 exit(1);
 295         }
 296 }
 297
 298 //**********************************************************************************************************************
 299
 300 MatrixOutputCommand::~MatrixOutputCommand(){}
 301
 302 //**********************************************************************************************************************
 303
 304 int MatrixOutputCommand::execute(){
 305         try {
 306
 307                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 308
 309                 //if the users entered no valid calculators don't execute command
 310                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
 311
 312                 input = new InputData(sharedfile, "sharedfile");
 313                 lookup = input->getSharedRAbundVectors();
 314                 string lastLabel = lookup[0]->getLabel();
 315
 316                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 317                 set<string> processedLabels;
 318                 set<string> userLabels = labels;
 319
 320                 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;}
 321
 322         if (subsample) {
 323             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 324                 subsampleSize = lookup[0]->getNumSeqs();
 325                 for (int i = 1; i < lookup.size(); i++) {
 326                     int thisSize = lookup[i]->getNumSeqs();
 327
 328                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 329                 }
 330             }else {
 331                 m->clearGroups();
 332                 Groups.clear();
 333                 vector<SharedRAbundVector*> temp;
 334                 for (int i = 0; i < lookup.size(); i++) {
 335                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 336                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 337                         delete lookup[i];
 338                     }else {
 339                         Groups.push_back(lookup[i]->getGroup());
 340                         temp.push_back(lookup[i]);
 341                     }
 342                 }
 343                 lookup = temp;
 344                 m->setGroups(Groups);
 345             }
 346
 347             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; }
 348         }
 349
 350                 numGroups = lookup.size();
 351         lines.resize(processors);
 352                 for (int i = 0; i < processors; i++) {
 353                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 354                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 355                 }
 356
 357                 if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->clearGroups(); return 0;  }
 358
 359                 //as long as you are not at the end of the file or done wih the lines you want
 360                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 361
 362                         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;  }
 363
 364                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 365                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 366                                 process(lookup);
 367
 368                                 processedLabels.insert(lookup[0]->getLabel());
 369                                 userLabels.erase(lookup[0]->getLabel());
 370                         }
 371
 372                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 373                                 string saveLabel = lookup[0]->getLabel();
 374
 375                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 376                                 lookup = input->getSharedRAbundVectors(lastLabel);
 377
 378                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 379                                 process(lookup);
 380
 381                                 processedLabels.insert(lookup[0]->getLabel());
 382                                 userLabels.erase(lookup[0]->getLabel());
 383
 384                                 //restore real lastlabel to save below
 385                                 lookup[0]->setLabel(saveLabel);
 386                         }
 387
 388                         lastLabel = lookup[0]->getLabel();
 389
 390                         //get next line to process
 391                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 392                         lookup = input->getSharedRAbundVectors();
 393                 }
 394
 395                 if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 396
 397                 //output error messages about any remaining user labels
 398                 set<string>::iterator it;
 399                 bool needToRun = false;
 400                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 401                         m->mothurOut("Your file does not include the label " + *it);
 402                         if (processedLabels.count(lastLabel) != 1) {
 403                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 404                                 needToRun = true;
 405                         }else {
 406                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 407                         }
 408                 }
 409
 410                 if (m->control_pressed) { outputTypes.clear(); delete input;  for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 411
 412                 //run last label if you need to
 413                 if (needToRun == true)  {
 414                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  }
 415                         lookup = input->getSharedRAbundVectors(lastLabel);
 416
 417                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 418                         process(lookup);
 419                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 420                 }
 421
 422                 if (m->control_pressed) { outputTypes.clear();  delete input;  for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0;  }
 423
 424                 //reset groups parameter
 425                 m->clearGroups();
 426
 427                 //set phylip file as new current phylipfile
 428                 string current = "";
 429                 itTypes = outputTypes.find("phylip");
 430                 if (itTypes != outputTypes.end()) {
 431                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 432                 }
 433
 434                 m->mothurOutEndLine();
 435                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 436                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 437                 m->mothurOutEndLine();
 438
 439
 440                 return 0;
 441         }
 442         catch(exception& e) {
 443                 m->errorOut(e, "MatrixOutputCommand", "execute");
 444                 exit(1);
 445         }
 446 }
 447 /***********************************************************/
 448 void MatrixOutputCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 449         try {
 450
 451                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 452
 453                 //output num seqs
 454                 out << simMatrix.size() << endl;
 455
 456                 if (output == "lt") {
 457                         for (int b = 0; b < simMatrix.size(); b++)      {
 458                                 out << lookup[b]->getGroup() << '\t';
 459                                 for (int n = 0; n < b; n++)     {
 460                                         out << simMatrix[b][n] << '\t';
 461                                 }
 462                                 out << endl;
 463                         }
 464                 }else{
 465                         for (int b = 0; b < simMatrix.size(); b++)      {
 466                                 out << lookup[b]->getGroup() << '\t';
 467                                 for (int n = 0; n < simMatrix[b].size(); n++)   {
 468                                         out << simMatrix[b][n] << '\t';
 469                                 }
 470                                 out << endl;
 471                         }
 472                 }
 473         }
 474         catch(exception& e) {
 475                 m->errorOut(e, "MatrixOutputCommand", "printSims");
 476                 exit(1);
 477         }
 478 }
 479 /***********************************************************/
 480 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
 481         try {
 482                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 483         vector< vector<seqDist>  > calcDists; calcDists.resize(matrixCalculators.size());
 484
 485         for (int thisIter = 0; thisIter < iters+1; thisIter++) {
 486             map<string, string> variables;
 487             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 488             variables["[distance]"] = thisLookup[0]->getLabel();
 489             variables["[tag2]"] = "";
 490
 491             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 492
 493             if (subsample && (thisIter != 0)) {
 494                 SubSample sample;
 495                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 496
 497                 //make copy of lookup so we don't get access violations
 498                 vector<SharedRAbundVector*> newLookup;
 499                 for (int k = 0; k < thisItersLookup.size(); k++) {
 500                     SharedRAbundVector* temp = new SharedRAbundVector();
 501                     temp->setLabel(thisItersLookup[k]->getLabel());
 502                     temp->setGroup(thisItersLookup[k]->getGroup());
 503                     newLookup.push_back(temp);
 504                 }
 505
 506                 //for each bin
 507                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 508                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 509                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 510                 }
 511
 512                 tempLabels = sample.getSample(newLookup, subsampleSize);
 513                 thisItersLookup = newLookup;
 514             }
 515
 516             if(processors == 1){
 517                 driver(thisItersLookup, 0, numGroups, calcDists);
 518             }else{
 519                 int process = 1;
 520                 vector<int> processIDS;
 521
 522                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 523                 //loop through and create all the processes you want
 524                 while (process != processors) {
 525                     int pid = fork();
 526
 527                     if (pid > 0) {
 528                         processIDS.push_back(pid);
 529                         process++;
 530                     }else if (pid == 0){
 531
 532                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 533
 534                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 535                         ofstream outtemp;
 536                         m->openOutputFile(tempdistFileName, outtemp);
 537
 538                         for (int i = 0; i < calcDists.size(); i++) {
 539                             outtemp << calcDists[i].size() << endl;
 540
 541                             for (int j = 0; j < calcDists[i].size(); j++) {
 542                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 543                             }
 544                         }
 545                         outtemp.close();
 546
 547                         exit(0);
 548                     }else {
 549                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 550                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 551                         exit(0);
 552                     }
 553                 }
 554
 555                 //parent do your part
 556                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 557
 558                 //force parent to wait until all the processes are done
 559                 for (int i = 0; i < processIDS.size(); i++) {
 560                     int temp = processIDS[i];
 561                     wait(&temp);
 562                 }
 563
 564                 for (int i = 0; i < processIDS.size(); i++) {
 565                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 566                     ifstream intemp;
 567                     m->openInputFile(tempdistFileName, intemp);
 568
 569                     for (int k = 0; k < calcDists.size(); k++) {
 570                         int size = 0;
 571                         intemp >> size; m->gobble(intemp);
 572
 573                         for (int j = 0; j < size; j++) {
 574                             int seq1 = 0;
 575                             int seq2 = 0;
 576                             float dist = 1.0;
 577
 578                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 579
 580                             seqDist tempDist(seq1, seq2, dist);
 581                             calcDists[k].push_back(tempDist);
 582                         }
 583                     }
 584                     intemp.close();
 585                     m->mothurRemove(tempdistFileName);
 586                 }
 587                 #else
 588                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 589                 //Windows version shared memory, so be careful when passing variables through the distSharedData struct.
 590                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 591                 //Taking advantage of shared memory to pass results vectors.
 592                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 593
 594                 vector<distSharedData*> pDataArray;
 595                 DWORD   dwThreadIdArray[processors-1];
 596                 HANDLE  hThreadArray[processors-1];
 597
 598                 //Create processor worker threads.
 599                 for( int i=1; i<processors; i++ ){
 600
 601                     //make copy of lookup so we don't get access violations
 602                     vector<SharedRAbundVector*> newLookup;
 603                     for (int k = 0; k < thisItersLookup.size(); k++) {
 604                         SharedRAbundVector* temp = new SharedRAbundVector();
 605                         temp->setLabel(thisItersLookup[k]->getLabel());
 606                         temp->setGroup(thisItersLookup[k]->getGroup());
 607                         newLookup.push_back(temp);
 608                     }
 609
 610                     //for each bin
 611                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 612                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 613                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 614                     }
 615
 616                     // Allocate memory for thread data.
 617                     distSharedData* tempSum = new distSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 618                     pDataArray.push_back(tempSum);
 619                     processIDS.push_back(i);
 620
 621                     hThreadArray[i-1] = CreateThread(NULL, 0, MyDistSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 622                 }
 623
 624                 //parent do your part
 625                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 626
 627                 //Wait until all threads have terminated.
 628                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 629
 630                 //Close all thread handles and free memory allocations.
 631                 for(int i=0; i < pDataArray.size(); i++){
 632                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 633
 634                     for (int k = 0; k < calcDists.size(); k++) {
 635                         int size = pDataArray[i]->calcDists[k].size();
 636                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 637                     }
 638
 639                     CloseHandle(hThreadArray[i]);
 640                     delete pDataArray[i];
 641                 }
 642
 643                 #endif
 644             }
 645
 646             if (subsample && (thisIter != 0)) {
 647                 if((thisIter) % 100 == 0){      m->mothurOut(toString(thisIter)); m->mothurOutEndLine();                }
 648                 calcDistsTotals.push_back(calcDists);
 649                 for (int i = 0; i < calcDists.size(); i++) {
 650                     for (int j = 0; j < calcDists[i].size(); j++) {
 651                         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");  }
 652                     }
 653                 }
 654                 //clean up memory
 655                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 656                 thisItersLookup.clear();
 657             }else { //print results for whole dataset
 658                 for (int i = 0; i < calcDists.size(); i++) {
 659                     if (m->control_pressed) { break; }
 660
 661                     //initialize matrix
 662                     vector< vector<double> > matrix; //square matrix to represent the distance
 663                     matrix.resize(thisLookup.size());
 664                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 665
 666                     for (int j = 0; j < calcDists[i].size(); j++) {
 667                         int row = calcDists[i][j].seq1;
 668                         int column = calcDists[i][j].seq2;
 669                         double dist = calcDists[i][j].dist;
 670
 671                         matrix[row][column] = dist;
 672                         matrix[column][row] = dist;
 673                     }
 674
 675                     variables["[outputtag]"] = output;
 676                     variables["[calc]"] = matrixCalculators[i]->getName();
 677                     string distFileName = getOutputFileName("phylip",variables);
 678                     outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 679
 680                     ofstream outDist;
 681                     m->openOutputFile(distFileName, outDist);
 682                     outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
 683
 684                     printSims(outDist, matrix);
 685
 686                     outDist.close();
 687                 }
 688             }
 689             for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 690                 }
 691
 692         if (iters != 0) {
 693             //we need to find the average distance and standard deviation for each groups distance
 694
 695             vector< vector<seqDist>  > calcAverages; calcAverages.resize(matrixCalculators.size());
 696             for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 697                 calcAverages[i].resize(calcDistsTotals[0][i].size());
 698
 699                 for (int j = 0; j < calcAverages[i].size(); j++) {
 700                     calcAverages[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 701                     calcAverages[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 702                     calcAverages[i][j].dist = 0.0;
 703                 }
 704             }
 705             if (mode == "average") {
 706                 for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 707                     for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 708                         for (int j = 0; j < calcAverages[i].size(); j++) {
 709                             calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
 710                             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");  }
 711                         }
 712                     }
 713                 }
 714
 715                 for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 716                     for (int j = 0; j < calcAverages[i].size(); j++) {
 717                         calcAverages[i][j].dist /= (float) iters;
 718                     }
 719                 }
 720             }else { //find median
 721                 for (int i = 0; i < calcAverages.size(); i++) { //for each calc
 722                     for (int j = 0; j < calcAverages[i].size(); j++) {  //for each comparison
 723                         vector<double> dists;
 724                         for (int thisIter = 0; thisIter < iters; thisIter++) { //for each subsample
 725                             dists.push_back(calcDistsTotals[thisIter][i][j].dist);
 726                         }
 727                         sort(dists.begin(), dists.end());
 728                         calcAverages[i][j].dist = dists[(iters/2)];
 729                     }
 730                 }
 731             }
 732             //find standard deviation
 733             vector< vector<seqDist>  > stdDev; stdDev.resize(matrixCalculators.size());
 734             for (int i = 0; i < stdDev.size(); i++) {  //initialize sums to zero.
 735                 stdDev[i].resize(calcDistsTotals[0][i].size());
 736
 737                 for (int j = 0; j < stdDev[i].size(); j++) {
 738                     stdDev[i][j].seq1 = calcDistsTotals[0][i][j].seq1;
 739                     stdDev[i][j].seq2 = calcDistsTotals[0][i][j].seq2;
 740                     stdDev[i][j].dist = 0.0;
 741                 }
 742             }
 743
 744             for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
 745                 for (int i = 0; i < stdDev.size(); i++) {
 746                     for (int j = 0; j < stdDev[i].size(); j++) {
 747                         stdDev[i][j].dist += ((calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist) * (calcDistsTotals[thisIter][i][j].dist - calcAverages[i][j].dist));
 748                     }
 749                 }
 750             }
 751
 752             for (int i = 0; i < stdDev.size(); i++) {  //finds average.
 753                 for (int j = 0; j < stdDev[i].size(); j++) {
 754                     stdDev[i][j].dist /= (float) iters;
 755                     stdDev[i][j].dist = sqrt(stdDev[i][j].dist);
 756                 }
 757             }
 758
 759             //print results
 760             for (int i = 0; i < calcDists.size(); i++) {
 761                 vector< vector<double> > matrix; //square matrix to represent the distance
 762                 matrix.resize(thisLookup.size());
 763                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 764
 765                 vector< vector<double> > stdmatrix; //square matrix to represent the stdDev
 766                 stdmatrix.resize(thisLookup.size());
 767                 for (int k = 0; k < thisLookup.size(); k++) {  stdmatrix[k].resize(thisLookup.size(), 0.0); }
 768
 769
 770                 for (int j = 0; j < calcAverages[i].size(); j++) {
 771                     int row = calcAverages[i][j].seq1;
 772                     int column = calcAverages[i][j].seq2;
 773                     float dist = calcAverages[i][j].dist;
 774                     float stdDist = stdDev[i][j].dist;
 775
 776                     matrix[row][column] = dist;
 777                     matrix[column][row] = dist;
 778                     stdmatrix[row][column] = stdDist;
 779                     stdmatrix[column][row] = stdDist;
 780                 }
 781
 782                 map<string, string> variables;
 783                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 784                 variables["[distance]"] = thisLookup[0]->getLabel();
 785                 variables["[outputtag]"] = output;
 786                 variables["[tag2]"] = "ave";
 787                 variables["[calc]"] = matrixCalculators[i]->getName();
 788                 string distFileName = getOutputFileName("phylip",variables);
 789                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 790                 ofstream outAve;
 791                 m->openOutputFile(distFileName, outAve);
 792                 outAve.setf(ios::fixed, ios::floatfield); outAve.setf(ios::showpoint);
 793
 794                 printSims(outAve, matrix);
 795
 796                 outAve.close();
 797
 798                 variables["[tag2]"] = "std";
 799                 distFileName = getOutputFileName("phylip",variables);
 800                 outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
 801                 ofstream outSTD;
 802                 m->openOutputFile(distFileName, outSTD);
 803                 outSTD.setf(ios::fixed, ios::floatfield); outSTD.setf(ios::showpoint);
 804
 805                 printSims(outSTD, stdmatrix);
 806
 807                 outSTD.close();
 808
 809             }
 810         }
 811
 812                 return 0;
 813         }
 814         catch(exception& e) {
 815                 m->errorOut(e, "MatrixOutputCommand", "process");
 816                 exit(1);
 817         }
 818 }
 819 /**************************************************************************************************/
 820 int MatrixOutputCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
 821         try {
 822                 vector<SharedRAbundVector*> subset;
 823
 824                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
 825
 826                         for (int l = 0; l < k; l++) {
 827
 828                                 if (k != l) { //we dont need to similiarity of a groups to itself
 829                                         subset.clear(); //clear out old pair of sharedrabunds
 830                                         //add new pair of sharedrabunds
 831                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 832
 833                                         for(int i=0;i<matrixCalculators.size();i++) {
 834
 835                                                 //if this calc needs all groups to calculate the pair load all groups
 836                                                 if (matrixCalculators[i]->getNeedsAll()) {
 837                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 838                                                         for (int w = 0; w < thisLookup.size(); w++) {
 839                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 840                                                         }
 841                                                 }
 842
 843                                                 vector<double> tempdata = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
 844
 845                                                 if (m->control_pressed) { return 1; }
 846
 847                                                 seqDist temp(l, k, tempdata[0]);
 848                                                 calcDists[i].push_back(temp);
 849                                         }
 850                                 }
 851                         }
 852                 }
 853
 854                 return 0;
 855         }
 856         catch(exception& e) {
 857                 m->errorOut(e, "MatrixOutputCommand", "driver");
 858                 exit(1);
 859         }
 860 }
 861 /***********************************************************/
 862
 863
 864