treegroupscommand.cpp

   1 /*
   2  *  treegroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "treegroupscommand.h"
  11 #include "subsample.h"
  12 #include "consensus.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> TreeGroupCommand::setParameters(){
  16         try {
  17                 CommandParameter pshared("shared", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pshared);
  18                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pphylip);
  19                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
  20                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "ColumnName",false,false); parameters.push_back(pcolumn);
  21         CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  22         CommandParameter psubsample("subsample", "String", "", "", "", "", "",false,false); parameters.push_back(psubsample);
  23         CommandParameter pcutoff("cutoff", "Number", "", "10", "", "", "",false,false); parameters.push_back(pcutoff);
  24                 CommandParameter pprecision("precision", "Number", "", "100", "", "", "",false,false); parameters.push_back(pprecision);
  25                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  26                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  27                 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);
  28
  29         CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  30 //CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput);
  31                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  32                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  33
  34                 vector<string> myArray;
  35                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  36                 return myArray;
  37         }
  38         catch(exception& e) {
  39                 m->errorOut(e, "TreeGroupCommand", "setParameters");
  40                 exit(1);
  41         }
  42 }
  43 //**********************************************************************************************************************
  44 string TreeGroupCommand::getHelpString(){
  45         try {
  46                 string helpString = "";
  47                 ValidCalculators validCalculator;
  48                 helpString += "The tree.shared command creates a .tre to represent the similiarity between groups or sequences.\n";
  49                 helpString += "The tree.shared command parameters are shared, groups, calc, phylip, column, name, cutoff, precision, processors, subsample, iters and label.\n";
  50                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n";
  51                 helpString += "The group names are separated by dashes. The label allow you to select what distance levels you would like trees created for, and are also separated by dashes.\n";
  52                 helpString += "The phylip or column parameter are required if you do not provide a sharedfile, and only one may be used.  If you use a column file the name filename is required. \n";
  53                 helpString += "If you do not provide a cutoff value 10.00 is assumed. If you do not provide a precision value then 100 is assumed.\n";
  54                 helpString += "The tree.shared command should be in the following format: tree.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n";
  55         helpString += "The iters parameter allows you to choose the number of times you would like to run the subsample.\n";
  56         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. The subsample parameter may only be used with a shared file.\n";
  57                 helpString += "Example tree.shared(groups=A-B-C, calc=jabund-sorabund).\n";
  58                 helpString += "The default value for groups is all the groups in your groupfile.\n";
  59                 helpString += "The default value for calc is jclass-thetayc.\n";
  60                 helpString += "The tree.shared command outputs a .tre file for each calculator you specify at each distance you choose.\n";
  61                 helpString += validCalculator.printCalc("treegroup");
  62                 helpString += "Or the tree.shared command can be in the following format: tree.shared(phylip=yourPhylipFile).\n";
  63                 helpString += "Example tree.shared(phylip=abrecovery.dist).\n";
  64                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  65                 return helpString;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "TreeGroupCommand", "getHelpString");
  69                 exit(1);
  70         }
  71 }
  72 //**********************************************************************************************************************
  73 TreeGroupCommand::TreeGroupCommand(){
  74         try {
  75                 abort = true; calledHelp = true;
  76                 setParameters();
  77                 //initialize outputTypes
  78                 vector<string> tempOutNames;
  79                 outputTypes["tree"] = tempOutNames;
  80         }
  81         catch(exception& e) {
  82                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
  83                 exit(1);
  84         }
  85 }
  86 //**********************************************************************************************************************
  87
  88 TreeGroupCommand::TreeGroupCommand(string option)  {
  89         try {
  90                 abort = false; calledHelp = false;
  91                 allLines = 1;
  92
  93                 //allow user to run help
  94                 if(option == "help") { help(); abort = true; calledHelp = true; }
  95                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  96
  97                 else {
  98                         vector<string> myArray = setParameters();
  99
 100                         OptionParser parser(option);
 101                         map<string, string> parameters = parser. getParameters();
 102
 103                         ValidParameters validParameter;
 104                         map<string, string>::iterator it;
 105
 106                         //check to make sure all parameters are valid for command
 107                         for (it = parameters.begin(); it != parameters.end(); it++) {
 108                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 109                         }
 110
 111                         //initialize outputTypes
 112                         vector<string> tempOutNames;
 113                         outputTypes["tree"] = tempOutNames;
 114
 115                         //if the user changes the input directory command factory will send this info to us in the output parameter
 116                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 117                         if (inputDir == "not found"){   inputDir = "";          }
 118                         else {
 119                                 string path;
 120                                 it = parameters.find("phylip");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 126                                 }
 127
 128                                 it = parameters.find("column");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 134                                 }
 135
 136                                 it = parameters.find("name");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 142                                 }
 143                         }
 144
 145                         //check for required parameters
 146                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 147                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 148                         else if (phylipfile == "not found") { phylipfile = ""; }
 149                         else {  inputfile = phylipfile;  format = "phylip"; m->setPhylipFile(phylipfile);       }
 150
 151                         columnfile = validParameter.validFile(parameters, "column", true);
 152                         if (columnfile == "not open") { columnfile = ""; abort = true; }
 153                         else if (columnfile == "not found") { columnfile = ""; }
 154                         else {  inputfile = columnfile; format = "column";      m->setColumnFile(columnfile); }
 155
 156                         sharedfile = validParameter.validFile(parameters, "shared", true);
 157                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 158                         else if (sharedfile == "not found") { sharedfile = ""; }
 159                         else {  inputfile = sharedfile; format = "sharedfile";  m->setSharedFile(sharedfile); }
 160
 161                         namefile = validParameter.validFile(parameters, "name", true);
 162                         if (namefile == "not open") { abort = true; }
 163                         else if (namefile == "not found") { namefile = ""; }
 164                         else { m->setNameFile(namefile); }
 165
 166                         if ((phylipfile == "") && (columnfile == "") && (sharedfile == "")) {
 167                                 //is there are current file available for either of these?
 168                                 //give priority to shared, then column, then phylip
 169                                 sharedfile = m->getSharedFile();
 170                                 if (sharedfile != "") {  inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 171                                 else {
 172                                         columnfile = m->getColumnFile();
 173                                         if (columnfile != "") { inputfile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 174                                         else {
 175                                                 phylipfile = m->getPhylipFile();
 176                                                 if (phylipfile != "") { inputfile = phylipfile;  format = "phylip";  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 177                                                 else {
 178                                                         m->mothurOut("No valid current files. You must provide a shared, phylip or column file."); m->mothurOutEndLine();
 179                                                         abort = true;
 180                                                 }
 181                                         }
 182                                 }
 183                         }
 184                         else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When running the tree.shared command with a distance file you may not use both the column and the phylip parameters."); m->mothurOutEndLine(); abort = true; }
 185
 186                         if (columnfile != "") {
 187                                 if (namefile == "") {
 188                                         namefile = m->getNameFile();
 189                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 190                                         else {
 191                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 192                                                 abort = true;
 193                                         }
 194                                 }
 195                         }
 196
 197                         //check for optional parameter and set defaults
 198                         // ...at some point should added some additional type checking...
 199                         label = validParameter.validFile(parameters, "label", false);
 200                         if (label == "not found") { label = ""; }
 201                         else {
 202                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 203                                 else { allLines = 1;  }
 204                         }
 205
 206                         groups = validParameter.validFile(parameters, "groups", false);
 207                         if (groups == "not found") { groups = ""; }
 208                         else {
 209                                 m->splitAtDash(groups, Groups);
 210                                 m->setGroups(Groups);
 211                         }
 212
 213                         calc = validParameter.validFile(parameters, "calc", false);
 214                         if (calc == "not found") { calc = "jclass-thetayc";  }
 215                         else {
 216                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 217                         }
 218                         m->splitAtDash(calc, Estimators);
 219                         if (m->inUsersGroups("citation", Estimators)) {
 220                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 221                                 //remove citation from list of calcs
 222                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 223                         }
 224
 225                         string temp;
 226                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 227                         m->mothurConvert(temp, precision);
 228
 229                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 230                         m->mothurConvert(temp, cutoff);
 231                         cutoff += (5 / (precision * 10.0));
 232
 233             temp = validParameter.validFile(parameters, "processors", false);   if (temp == "not found"){       temp = m->getProcessors();      }
 234                         m->setProcessors(temp);
 235                         m->mothurConvert(temp, processors);
 236
 237             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 238                         m->mothurConvert(temp, iters);
 239
 240             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 241                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 242             else {
 243                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 244                 else { subsample = false; }
 245             }
 246
 247             if (subsample == false) { iters = 1; }
 248
 249             if (subsample && (format != "sharedfile")) { m->mothurOut("[ERROR]: the subsample parameter can only be used with a shared file.\n"); abort=true; }
 250
 251                         //if the user changes the output directory command factory will send this info to us in the output parameter
 252                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 253                                 outputDir = "";
 254                                 outputDir += m->hasPath(inputfile); //if user entered a file with a path then preserve it
 255                         }
 256                 }
 257
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 261                 exit(1);
 262         }
 263 }
 264 //**********************************************************************************************************************
 265
 266 TreeGroupCommand::~TreeGroupCommand(){
 267         if (abort == false) {
 268                 if (format == "sharedfile") {  delete input; }
 269                 else { delete readMatrix;  delete matrix; delete list; }
 270                 delete tmap;
 271         }
 272
 273 }
 274
 275 //**********************************************************************************************************************
 276
 277 int TreeGroupCommand::execute(){
 278         try {
 279
 280                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 281
 282                 if (format == "sharedfile") {
 283
 284                         ValidCalculators validCalculator;
 285
 286                         for (int i=0; i<Estimators.size(); i++) {
 287                                 if (validCalculator.isValidCalculator("treegroup", Estimators[i]) == true) {
 288                                         if (Estimators[i] == "sharedsobs") {
 289                                                 treeCalculators.push_back(new SharedSobsCS());
 290                                         }else if (Estimators[i] == "sharedchao") {
 291                                                 treeCalculators.push_back(new SharedChao1());
 292                                         }else if (Estimators[i] == "sharedace") {
 293                                                 treeCalculators.push_back(new SharedAce());
 294                                         }else if (Estimators[i] == "jabund") {
 295                                                 treeCalculators.push_back(new JAbund());
 296                                         }else if (Estimators[i] == "sorabund") {
 297                                                 treeCalculators.push_back(new SorAbund());
 298                                         }else if (Estimators[i] == "jclass") {
 299                                                 treeCalculators.push_back(new Jclass());
 300                                         }else if (Estimators[i] == "sorclass") {
 301                                                 treeCalculators.push_back(new SorClass());
 302                                         }else if (Estimators[i] == "jest") {
 303                                                 treeCalculators.push_back(new Jest());
 304                                         }else if (Estimators[i] == "sorest") {
 305                                                 treeCalculators.push_back(new SorEst());
 306                                         }else if (Estimators[i] == "thetayc") {
 307                                                 treeCalculators.push_back(new ThetaYC());
 308                                         }else if (Estimators[i] == "thetan") {
 309                                                 treeCalculators.push_back(new ThetaN());
 310                                         }else if (Estimators[i] == "kstest") {
 311                                                 treeCalculators.push_back(new KSTest());
 312                                         }else if (Estimators[i] == "sharednseqs") {
 313                                                 treeCalculators.push_back(new SharedNSeqs());
 314                                         }else if (Estimators[i] == "ochiai") {
 315                                                 treeCalculators.push_back(new Ochiai());
 316                                         }else if (Estimators[i] == "anderberg") {
 317                                                 treeCalculators.push_back(new Anderberg());
 318                                         }else if (Estimators[i] == "kulczynski") {
 319                                                 treeCalculators.push_back(new Kulczynski());
 320                                         }else if (Estimators[i] == "kulczynskicody") {
 321                                                 treeCalculators.push_back(new KulczynskiCody());
 322                                         }else if (Estimators[i] == "lennon") {
 323                                                 treeCalculators.push_back(new Lennon());
 324                                         }else if (Estimators[i] == "morisitahorn") {
 325                                                 treeCalculators.push_back(new MorHorn());
 326                                         }else if (Estimators[i] == "braycurtis") {
 327                                                 treeCalculators.push_back(new BrayCurtis());
 328                                         }else if (Estimators[i] == "whittaker") {
 329                                                 treeCalculators.push_back(new Whittaker());
 330                                         }else if (Estimators[i] == "odum") {
 331                                                 treeCalculators.push_back(new Odum());
 332                                         }else if (Estimators[i] == "canberra") {
 333                                                 treeCalculators.push_back(new Canberra());
 334                                         }else if (Estimators[i] == "structeuclidean") {
 335                                                 treeCalculators.push_back(new StructEuclidean());
 336                                         }else if (Estimators[i] == "structchord") {
 337                                                 treeCalculators.push_back(new StructChord());
 338                                         }else if (Estimators[i] == "hellinger") {
 339                                                 treeCalculators.push_back(new Hellinger());
 340                                         }else if (Estimators[i] == "manhattan") {
 341                                                 treeCalculators.push_back(new Manhattan());
 342                                         }else if (Estimators[i] == "structpearson") {
 343                                                 treeCalculators.push_back(new StructPearson());
 344                                         }else if (Estimators[i] == "soergel") {
 345                                                 treeCalculators.push_back(new Soergel());
 346                                         }else if (Estimators[i] == "spearman") {
 347                                                 treeCalculators.push_back(new Spearman());
 348                                         }else if (Estimators[i] == "structkulczynski") {
 349                                                 treeCalculators.push_back(new StructKulczynski());
 350                                         }else if (Estimators[i] == "speciesprofile") {
 351                                                 treeCalculators.push_back(new SpeciesProfile());
 352                                         }else if (Estimators[i] == "hamming") {
 353                                                 treeCalculators.push_back(new Hamming());
 354                                         }else if (Estimators[i] == "structchi2") {
 355                                                 treeCalculators.push_back(new StructChi2());
 356                                         }else if (Estimators[i] == "gower") {
 357                                                 treeCalculators.push_back(new Gower());
 358                                         }else if (Estimators[i] == "memchi2") {
 359                                                 treeCalculators.push_back(new MemChi2());
 360                                         }else if (Estimators[i] == "memchord") {
 361                                                 treeCalculators.push_back(new MemChord());
 362                                         }else if (Estimators[i] == "memeuclidean") {
 363                                                 treeCalculators.push_back(new MemEuclidean());
 364                                         }else if (Estimators[i] == "mempearson") {
 365                                                 treeCalculators.push_back(new MemPearson());
 366                                         }
 367                                 }
 368                         }
 369
 370                         //if the users entered no valid calculators don't execute command
 371                         if (treeCalculators.size() == 0) { m->mothurOut("You have given no valid calculators."); m->mothurOutEndLine(); return 0; }
 372
 373                         input = new InputData(sharedfile, "sharedfile");
 374                         lookup = input->getSharedRAbundVectors();
 375                         lastLabel = lookup[0]->getLabel();
 376
 377                         if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0; }
 378
 379                         //used in tree constructor
 380                         m->runParse = false;
 381
 382                         //create treemap class from groupmap for tree class to use
 383                         tmap = new TreeMap();
 384                         tmap->makeSim(m->getAllGroups());
 385
 386                         //clear globaldatas old tree names if any
 387                         m->Treenames.clear();
 388
 389                         //fills globaldatas tree names
 390                         m->Treenames = m->getGroups();
 391
 392                         if (m->control_pressed) { return 0; }
 393
 394                         //create tree file
 395                         makeSimsShared();
 396
 397                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 398                 }else{
 399                         //read in dist file
 400                         filename = inputfile;
 401
 402                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 403                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 404
 405                         readMatrix->setCutoff(cutoff);
 406
 407                         if(namefile != ""){
 408                                 nameMap = new NameAssignment(namefile);
 409                                 nameMap->readMap();
 410                         }
 411                         else{
 412                                 nameMap = NULL;
 413                         }
 414
 415                         readMatrix->read(nameMap);
 416                         list = readMatrix->getListVector();
 417                         matrix = readMatrix->getMatrix();
 418
 419                         //make treemap
 420                         tmap = new TreeMap();
 421
 422                         if (m->control_pressed) { return 0; }
 423
 424                         tmap->makeSim(list);
 425
 426                         vector<string> namesGroups = tmap->getNamesOfGroups();
 427                         m->setGroups(namesGroups);
 428
 429                         //clear globaldatas old tree names if any
 430                         m->Treenames.clear();
 431
 432                         //fills globaldatas tree names
 433                         m->Treenames = m->getGroups();
 434
 435                         //used in tree constructor
 436                         m->runParse = false;
 437
 438                         if (m->control_pressed) { return 0; }
 439
 440                         vector< vector<double> > matrix = makeSimsDist();
 441
 442                         if (m->control_pressed) { return 0; }
 443
 444                         //create a new filename
 445                         string outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + "tre";
 446                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 447
 448                         Tree* newTree = createTree(matrix);
 449
 450             if (newTree != NULL) {  writeTree(outputFile, newTree); delete newTree; }
 451
 452                         if (m->control_pressed) { return 0; }
 453
 454                         m->mothurOut("Tree complete. "); m->mothurOutEndLine();
 455
 456                 }
 457
 458                 //reset groups parameter
 459                 m->clearGroups();
 460
 461                 //set tree file as new current treefile
 462                 string current = "";
 463                 itTypes = outputTypes.find("tree");
 464                 if (itTypes != outputTypes.end()) {
 465                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 466                 }
 467
 468                 m->mothurOutEndLine();
 469                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 470                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 471                 m->mothurOutEndLine();
 472
 473                 return 0;
 474         }
 475         catch(exception& e) {
 476                 m->errorOut(e, "TreeGroupCommand", "execute");
 477                 exit(1);
 478         }
 479 }
 480 //**********************************************************************************************************************
 481
 482 Tree* TreeGroupCommand::createTree(vector< vector<double> >& simMatrix){
 483         try {
 484                 //create tree
 485                 t = new Tree(tmap, simMatrix);
 486
 487         if (m->control_pressed) { delete t; t = NULL; return t; }
 488
 489         //assemble tree
 490         map<string, string> empty;
 491                 t->assembleTree(empty);
 492
 493                 return t;
 494         }
 495         catch(exception& e) {
 496                 m->errorOut(e, "TreeGroupCommand", "createTree");
 497                 exit(1);
 498         }
 499 }
 500 /***********************************************************/
 501 int TreeGroupCommand::writeTree(string out, Tree* T) {
 502         try {
 503
 504         //print newick file
 505                 t->createNewickFile(out);
 506
 507         if (m->control_pressed) { m->mothurRemove(out); outputNames.pop_back(); return 1; }
 508
 509         return 0;
 510
 511         }
 512         catch(exception& e) {
 513                 m->errorOut(e, "TreeGroupCommand", "printSims");
 514                 exit(1);
 515         }
 516 }
 517
 518 /***********************************************************/
 519 void TreeGroupCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 520         try {
 521
 522         for (int m = 0; m < simMatrix.size(); m++)      {
 523                         //out << lookup[m]->getGroup() << '\t';
 524                         for (int n = 0; n < simMatrix.size(); n++)      {
 525                                 out << simMatrix[m][n] << '\t';
 526                         }
 527                         out << endl;
 528                 }
 529
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "TreeGroupCommand", "printSims");
 533                 exit(1);
 534         }
 535 }
 536 /***********************************************************/
 537 vector< vector<double> > TreeGroupCommand::makeSimsDist() {
 538         try {
 539                 numGroups = list->size();
 540
 541                 //initialize simMatrix
 542                 vector< vector<double> > simMatrix;
 543                 simMatrix.resize(numGroups);
 544                 for (int k = 0; k < simMatrix.size(); k++)      {
 545                         for (int j = 0; j < simMatrix.size(); j++)      {
 546                                 simMatrix[k].push_back(0.0);
 547                         }
 548                 }
 549
 550                 //go through sparse matrix and fill sims
 551                 //go through each cell in the sparsematrix
 552                 for(MatData currentCell = matrix->begin(); currentCell != matrix->end(); currentCell++){
 553                         //similairity = -(distance-1)
 554                         simMatrix[currentCell->row][currentCell->column] = -(currentCell->dist -1.0);
 555                         simMatrix[currentCell->column][currentCell->row] = -(currentCell->dist -1.0);
 556
 557                         if (m->control_pressed) { return simMatrix; }
 558
 559                 }
 560
 561                 return simMatrix;
 562         }
 563         catch(exception& e) {
 564                 m->errorOut(e, "TreeGroupCommand", "makeSimsDist");
 565                 exit(1);
 566         }
 567 }
 568
 569 /***********************************************************/
 570 int TreeGroupCommand::makeSimsShared() {
 571         try {
 572
 573         if (subsample) {
 574             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 575                 subsampleSize = lookup[0]->getNumSeqs();
 576                 for (int i = 1; i < lookup.size(); i++) {
 577                     int thisSize = lookup[i]->getNumSeqs();
 578
 579                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 580                 }
 581             }else {
 582                 m->clearGroups();
 583                 Groups.clear();
 584                 vector<SharedRAbundVector*> temp;
 585                 for (int i = 0; i < lookup.size(); i++) {
 586                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 587                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 588                         delete lookup[i];
 589                     }else {
 590                         Groups.push_back(lookup[i]->getGroup());
 591                         temp.push_back(lookup[i]);
 592                     }
 593                 }
 594                 lookup = temp;
 595                 m->setGroups(Groups);
 596             }
 597
 598             if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); m->control_pressed = true; return 0; }
 599         }
 600
 601         numGroups = lookup.size();
 602                 lines.resize(processors);
 603                 for (int i = 0; i < processors; i++) {
 604                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 605                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 606                 }
 607
 608                 set<string> processedLabels;
 609                 set<string> userLabels = labels;
 610
 611                 //as long as you are not at the end of the file or done wih the lines you want
 612                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 613                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; } return 1; }
 614
 615                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 616                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 617                                 process(lookup);
 618
 619                                 processedLabels.insert(lookup[0]->getLabel());
 620                                 userLabels.erase(lookup[0]->getLabel());
 621                         }
 622
 623                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 624                                 string saveLabel = lookup[0]->getLabel();
 625
 626                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 627                                 lookup = input->getSharedRAbundVectors(lastLabel);
 628
 629                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 630                                 process(lookup);
 631
 632                                 processedLabels.insert(lookup[0]->getLabel());
 633                                 userLabels.erase(lookup[0]->getLabel());
 634
 635                                 //restore real lastlabel to save below
 636                                 lookup[0]->setLabel(saveLabel);
 637                         }
 638
 639                         lastLabel = lookup[0]->getLabel();
 640
 641                         //get next line to process
 642                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 643                         lookup = input->getSharedRAbundVectors();
 644                 }
 645
 646                 if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; } return 1; }
 647
 648                 //output error messages about any remaining user labels
 649                 set<string>::iterator it;
 650                 bool needToRun = false;
 651                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 652                         m->mothurOut("Your file does not include the label " + *it);
 653                         if (processedLabels.count(lastLabel) != 1) {
 654                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 655                                 needToRun = true;
 656                         }else {
 657                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 658                         }
 659                 }
 660
 661                 //run last label if you need to
 662                 if (needToRun == true)  {
 663                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 664                         lookup = input->getSharedRAbundVectors(lastLabel);
 665
 666                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 667                         process(lookup);
 668                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 669                 }
 670
 671                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 672
 673                 return 0;
 674         }
 675         catch(exception& e) {
 676                 m->errorOut(e, "TreeGroupCommand", "makeSimsShared");
 677                 exit(1);
 678         }
 679 }
 680
 681 /***********************************************************/
 682 int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 683         try{
 684                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 685         vector< vector<seqDist>  > calcDists; calcDists.resize(treeCalculators.size());
 686
 687         for (int thisIter = 0; thisIter < iters; thisIter++) {
 688
 689             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 690
 691             if (subsample) {
 692                 SubSample sample;
 693                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 694
 695                 //make copy of lookup so we don't get access violations
 696                 vector<SharedRAbundVector*> newLookup;
 697                 for (int k = 0; k < thisItersLookup.size(); k++) {
 698                     SharedRAbundVector* temp = new SharedRAbundVector();
 699                     temp->setLabel(thisItersLookup[k]->getLabel());
 700                     temp->setGroup(thisItersLookup[k]->getGroup());
 701                     newLookup.push_back(temp);
 702                 }
 703
 704                 //for each bin
 705                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 706                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 707                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 708                 }
 709
 710                 tempLabels = sample.getSample(newLookup, subsampleSize);
 711                 thisItersLookup = newLookup;
 712             }
 713
 714             if(processors == 1){
 715                 driver(thisItersLookup, 0, numGroups, calcDists);
 716             }else{
 717                 int process = 1;
 718                 vector<int> processIDS;
 719
 720 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 721                 //loop through and create all the processes you want
 722                 while (process != processors) {
 723                     int pid = fork();
 724
 725                     if (pid > 0) {
 726                         processIDS.push_back(pid);
 727                         process++;
 728                     }else if (pid == 0){
 729
 730                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 731
 732                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 733                         ofstream outtemp;
 734                         m->openOutputFile(tempdistFileName, outtemp);
 735
 736                         for (int i = 0; i < calcDists.size(); i++) {
 737                             outtemp << calcDists[i].size() << endl;
 738
 739                             for (int j = 0; j < calcDists[i].size(); j++) {
 740                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 741                             }
 742                         }
 743                         outtemp.close();
 744
 745                         exit(0);
 746                     }else {
 747                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 748                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 749                         exit(0);
 750                     }
 751                 }
 752
 753                 //parent do your part
 754                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 755
 756                 //force parent to wait until all the processes are done
 757                 for (int i = 0; i < processIDS.size(); i++) {
 758                     int temp = processIDS[i];
 759                     wait(&temp);
 760                 }
 761
 762                 for (int i = 0; i < processIDS.size(); i++) {
 763                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 764                     ifstream intemp;
 765                     m->openInputFile(tempdistFileName, intemp);
 766
 767                     for (int k = 0; k < calcDists.size(); k++) {
 768                         int size = 0;
 769                         intemp >> size; m->gobble(intemp);
 770
 771                         for (int j = 0; j < size; j++) {
 772                             int seq1 = 0;
 773                             int seq2 = 0;
 774                             float dist = 1.0;
 775
 776                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 777
 778                             seqDist tempDist(seq1, seq2, dist);
 779                             calcDists[k].push_back(tempDist);
 780                         }
 781                     }
 782                     intemp.close();
 783                     m->mothurRemove(tempdistFileName);
 784                 }
 785 #else
 786                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 787                 //Windows version shared memory, so be careful when passing variables through the treeSharedData struct.
 788                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 789                 //Taking advantage of shared memory to pass results vectors.
 790                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 791
 792                 vector<treeSharedData*> pDataArray;
 793                 DWORD   dwThreadIdArray[processors-1];
 794                 HANDLE  hThreadArray[processors-1];
 795
 796                 //Create processor worker threads.
 797                 for( int i=1; i<processors; i++ ){
 798
 799                     //make copy of lookup so we don't get access violations
 800                     vector<SharedRAbundVector*> newLookup;
 801                     for (int k = 0; k < thisItersLookup.size(); k++) {
 802                         SharedRAbundVector* temp = new SharedRAbundVector();
 803                         temp->setLabel(thisItersLookup[k]->getLabel());
 804                         temp->setGroup(thisItersLookup[k]->getGroup());
 805                         newLookup.push_back(temp);
 806                     }
 807
 808                     //for each bin
 809                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 810                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 811                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 812                     }
 813
 814                     // Allocate memory for thread data.
 815                     treeSharedData* tempSum = new treeSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 816                     pDataArray.push_back(tempSum);
 817                     processIDS.push_back(i);
 818
 819                     hThreadArray[i-1] = CreateThread(NULL, 0, MyTreeSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 820                 }
 821
 822                 //parent do your part
 823                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 824
 825                 //Wait until all threads have terminated.
 826                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 827
 828                 //Close all thread handles and free memory allocations.
 829                 for(int i=0; i < pDataArray.size(); i++){
 830                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 831
 832                     for (int k = 0; k < calcDists.size(); k++) {
 833                         int size = pDataArray[i]->calcDists[k].size();
 834                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 835                     }
 836
 837                     CloseHandle(hThreadArray[i]);
 838                     delete pDataArray[i];
 839                 }
 840
 841 #endif
 842             }
 843
 844             calcDistsTotals.push_back(calcDists);
 845
 846             if (subsample) {
 847
 848                 //clean up memory
 849                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 850                 thisItersLookup.clear();
 851                 for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 852             }
 853                 }
 854
 855         if (iters != 1) {
 856             //we need to find the average distance and standard deviation for each groups distance
 857
 858             vector< vector<seqDist>  > calcAverages; calcAverages.resize(treeCalculators.size());
 859             for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 860                 calcAverages[i].resize(calcDistsTotals[0][i].size());
 861
 862                 for (int j = 0; j < calcAverages[i].size(); j++) {
 863                     calcAverages[i][j].seq1 = calcDists[i][j].seq1;
 864                     calcAverages[i][j].seq2 = calcDists[i][j].seq2;
 865                     calcAverages[i][j].dist = 0.0;
 866                 }
 867             }
 868
 869             for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 870                 for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 871                     for (int j = 0; j < calcAverages[i].size(); j++) {
 872                         calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
 873                     }
 874                 }
 875             }
 876
 877             for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 878                 for (int j = 0; j < calcAverages[i].size(); j++) {
 879                     calcAverages[i][j].dist /= (float) iters;
 880                 }
 881             }
 882
 883             //create average tree for each calc
 884             for (int i = 0; i < calcDists.size(); i++) {
 885                 vector< vector<double> > matrix; //square matrix to represent the distance
 886                 matrix.resize(thisLookup.size());
 887                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 888
 889                 for (int j = 0; j < calcAverages[i].size(); j++) {
 890                     int row = calcAverages[i][j].seq1;
 891                     int column = calcAverages[i][j].seq2;
 892                     float dist = calcAverages[i][j].dist;
 893
 894                     matrix[row][column] = dist;
 895                     matrix[column][row] = dist;
 896                 }
 897
 898                 //create a new filename
 899                 string outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".ave.tre";
 900                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 901
 902                 //creates tree from similarity matrix and write out file
 903                 Tree* newTree = createTree(matrix);
 904                 if (newTree != NULL) { writeTree(outputFile, newTree); }
 905             }
 906
 907             //create all trees for each calc and find their consensus tree
 908             for (int i = 0; i < calcDists.size(); i++) {
 909                 if (m->control_pressed) { break; }
 910
 911                 //create a new filename
 912                 string outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".all.tre";
 913                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 914
 915                 ofstream outAll;
 916                 m->openOutputFile(outputFile, outAll);
 917
 918                 vector<Tree*> trees;
 919                 for (int myIter = 0; myIter < iters; myIter++) {
 920
 921                     if(m->control_pressed) { break; }
 922
 923                     //initialize matrix
 924                     vector< vector<double> > matrix; //square matrix to represent the distance
 925                     matrix.resize(thisLookup.size());
 926                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 927
 928                     for (int j = 0; j < calcDistsTotals[myIter][i].size(); j++) {
 929                         int row = calcDistsTotals[myIter][i][j].seq1;
 930                         int column = calcDistsTotals[myIter][i][j].seq2;
 931                         double dist = calcDistsTotals[myIter][i][j].dist;
 932
 933                         matrix[row][column] = dist;
 934                         matrix[column][row] = dist;
 935                     }
 936
 937                     //creates tree from similarity matrix and write out file
 938                     Tree* newTree = createTree(matrix);
 939                     if (newTree != NULL) {
 940                         newTree->print(outAll);
 941                         trees.push_back(newTree);
 942                     }
 943                 }
 944                 outAll.close();
 945                 if (m->control_pressed) { for (int k = 0; k < trees.size(); k++) { delete trees[k]; } }
 946
 947                 Consensus consensus;
 948                 //clear old tree names if any
 949                 m->Treenames.clear(); m->Treenames = m->getGroups(); //may have changed if subsample eliminated groups
 950                 Tree* conTree = consensus.getTree(trees);
 951
 952                 //create a new filename
 953                 string conFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".cons.tre";
 954                 outputNames.push_back(conFile); outputTypes["tree"].push_back(conFile);
 955                 ofstream outTree;
 956                 m->openOutputFile(conFile, outTree);
 957
 958                 if (conTree != NULL) { conTree->print(outTree, "boot"); delete conTree; }
 959             }
 960
 961         }else {
 962
 963             for (int i = 0; i < calcDists.size(); i++) {
 964                 if (m->control_pressed) { break; }
 965
 966                 //initialize matrix
 967                 vector< vector<double> > matrix; //square matrix to represent the distance
 968                 matrix.resize(thisLookup.size());
 969                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 970
 971                 for (int j = 0; j < calcDists[i].size(); j++) {
 972                     int row = calcDists[i][j].seq1;
 973                     int column = calcDists[i][j].seq2;
 974                     double dist = calcDists[i][j].dist;
 975
 976                     matrix[row][column] = dist;
 977                     matrix[column][row] = dist;
 978                 }
 979
 980                 //create a new filename
 981                 string outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".tre";
 982                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 983
 984                 //creates tree from similarity matrix and write out file
 985                 Tree* newTree = createTree(matrix);
 986                 if (newTree != NULL) { writeTree(outputFile, newTree); delete newTree; }
 987             }
 988         }
 989
 990                 return 0;
 991         }
 992         catch(exception& e) {
 993                 m->errorOut(e, "TreeGroupCommand", "process");
 994                 exit(1);
 995         }
 996 }
 997 /**************************************************************************************************/
 998 int TreeGroupCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
 999         try {
1000                 vector<SharedRAbundVector*> subset;
1001                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
1002
1003                         for (int l = 0; l < k; l++) {
1004
1005                                 if (k != l) { //we dont need to similiarity of a groups to itself
1006                                         subset.clear(); //clear out old pair of sharedrabunds
1007                                         //add new pair of sharedrabunds
1008                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
1009
1010                                         for(int i=0;i<treeCalculators.size();i++) {
1011
1012                                                 //if this calc needs all groups to calculate the pair load all groups
1013                                                 if (treeCalculators[i]->getNeedsAll()) {
1014                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
1015                                                         for (int w = 0; w < thisLookup.size(); w++) {
1016                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
1017                                                         }
1018                                                 }
1019
1020                                                 vector<double> tempdata = treeCalculators[i]->getValues(subset); //saves the calculator outputs
1021
1022                                                 if (m->control_pressed) { return 1; }
1023
1024                                                 seqDist temp(l, k, -(tempdata[0]-1.0));
1025                                                 calcDists[i].push_back(temp);
1026                                         }
1027                                 }
1028                         }
1029                 }
1030
1031                 return 0;
1032         }
1033         catch(exception& e) {
1034                 m->errorOut(e, "TreeGroupCommand", "driver");
1035                 exit(1);
1036         }
1037 }
1038 /***********************************************************/
1039
1040
1041