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","tree",false,false,true); parameters.push_back(pshared);
  18                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none","tree",false,false); parameters.push_back(pphylip);
  19                 CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "ColumnName","",false,false); parameters.push_back(pname);
  20                 CommandParameter pcount("count", "InputTypes", "", "", "NameCount", "none", "countcolumn","",false,false); parameters.push_back(pcount);
  21         CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "ColumnName-countcolumn","tree",false,false); parameters.push_back(pcolumn);
  22         CommandParameter piters("iters", "Number", "", "1000", "", "", "","",false,false); parameters.push_back(piters);
  23         CommandParameter psubsample("subsample", "String", "", "", "", "", "","",false,false); parameters.push_back(psubsample);
  24         CommandParameter pcutoff("cutoff", "Number", "", "10", "", "", "","",false,false); parameters.push_back(pcutoff);
  25                 CommandParameter pprecision("precision", "Number", "", "100", "", "", "","",false,false); parameters.push_back(pprecision);
  26                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
  27                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
  28                 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);
  29
  30         CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  31 //CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput);
  32                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  33                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  34
  35                 vector<string> myArray;
  36                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  37                 return myArray;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "TreeGroupCommand", "setParameters");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 string TreeGroupCommand::getHelpString(){
  46         try {
  47                 string helpString = "";
  48                 ValidCalculators validCalculator;
  49                 helpString += "The tree.shared command creates a .tre to represent the similiarity between groups or sequences.\n";
  50                 helpString += "The tree.shared command parameters are shared, groups, calc, phylip, column, name, cutoff, precision, processors, subsample, iters and label.\n";
  51                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n";
  52                 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";
  53                 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";
  54                 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";
  55                 helpString += "The tree.shared command should be in the following format: tree.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n";
  56         helpString += "The iters parameter allows you to choose the number of times you would like to run the subsample.\n";
  57         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";
  58                 helpString += "Example tree.shared(groups=A-B-C, calc=jabund-sorabund).\n";
  59                 helpString += "The default value for groups is all the groups in your groupfile.\n";
  60                 helpString += "The default value for calc is jclass-thetayc.\n";
  61                 helpString += "The tree.shared command outputs a .tre file for each calculator you specify at each distance you choose.\n";
  62                 helpString += validCalculator.printCalc("treegroup");
  63                 helpString += "Or the tree.shared command can be in the following format: tree.shared(phylip=yourPhylipFile).\n";
  64                 helpString += "Example tree.shared(phylip=abrecovery.dist).\n";
  65                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  66                 return helpString;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "TreeGroupCommand", "getHelpString");
  70                 exit(1);
  71         }
  72 }
  73 //**********************************************************************************************************************
  74 string TreeGroupCommand::getOutputPattern(string type) {
  75     try {
  76         string pattern = "";
  77
  78         if (type == "tree") {  pattern = "[filename],[calc],[distance],[tag],tre-[filename],tre"; }
  79         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  80
  81         return pattern;
  82     }
  83     catch(exception& e) {
  84         m->errorOut(e, "TreeGroupCommand", "getOutputPattern");
  85         exit(1);
  86     }
  87 }
  88 //**********************************************************************************************************************
  89 TreeGroupCommand::TreeGroupCommand(){
  90         try {
  91                 abort = true; calledHelp = true;
  92                 setParameters();
  93                 //initialize outputTypes
  94                 vector<string> tempOutNames;
  95                 outputTypes["tree"] = tempOutNames;
  96         }
  97         catch(exception& e) {
  98                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
  99                 exit(1);
 100         }
 101 }
 102 //**********************************************************************************************************************
 103
 104 TreeGroupCommand::TreeGroupCommand(string option)  {
 105         try {
 106                 abort = false; calledHelp = false;
 107                 allLines = 1;
 108
 109                 //allow user to run help
 110                 if(option == "help") { help(); abort = true; calledHelp = true; }
 111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 112
 113                 else {
 114                         vector<string> myArray = setParameters();
 115
 116                         OptionParser parser(option);
 117                         map<string, string> parameters = parser. getParameters();
 118
 119                         ValidParameters validParameter;
 120                         map<string, string>::iterator it;
 121
 122                         //check to make sure all parameters are valid for command
 123                         for (it = parameters.begin(); it != parameters.end(); it++) {
 124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 125                         }
 126
 127                         //initialize outputTypes
 128                         vector<string> tempOutNames;
 129                         outputTypes["tree"] = tempOutNames;
 130
 131                         //if the user changes the input directory command factory will send this info to us in the output parameter
 132                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 133                         if (inputDir == "not found"){   inputDir = "";          }
 134                         else {
 135                                 string path;
 136                                 it = parameters.find("phylip");
 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["phylip"] = inputDir + it->second;           }
 142                                 }
 143
 144                                 it = parameters.find("column");
 145                                 //user has given a template file
 146                                 if(it != parameters.end()){
 147                                         path = m->hasPath(it->second);
 148                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 150                                 }
 151
 152                                 it = parameters.find("name");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 158                                 }
 159
 160                 it = parameters.find("count");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 166                                 }
 167                         }
 168
 169                         //check for required parameters
 170                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 171                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 172                         else if (phylipfile == "not found") { phylipfile = ""; }
 173                         else {  inputfile = phylipfile;  format = "phylip"; m->setPhylipFile(phylipfile);       }
 174
 175                         columnfile = validParameter.validFile(parameters, "column", true);
 176                         if (columnfile == "not open") { columnfile = ""; abort = true; }
 177                         else if (columnfile == "not found") { columnfile = ""; }
 178                         else {  inputfile = columnfile; format = "column";      m->setColumnFile(columnfile); }
 179
 180                         sharedfile = validParameter.validFile(parameters, "shared", true);
 181                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 182                         else if (sharedfile == "not found") { sharedfile = ""; }
 183                         else {  inputfile = sharedfile; format = "sharedfile";  m->setSharedFile(sharedfile); }
 184
 185                         namefile = validParameter.validFile(parameters, "name", true);
 186                         if (namefile == "not open") { abort = true; }
 187                         else if (namefile == "not found") { namefile = ""; }
 188                         else { m->setNameFile(namefile); }
 189
 190             countfile = validParameter.validFile(parameters, "count", true);
 191                         if (countfile == "not open") { abort = true; countfile = ""; }
 192                         else if (countfile == "not found") { countfile = ""; }
 193                         else { m->setCountTableFile(countfile); }
 194
 195                         if ((phylipfile == "") && (columnfile == "") && (sharedfile == "")) {
 196                                 //is there are current file available for either of these?
 197                                 //give priority to shared, then column, then phylip
 198                                 sharedfile = m->getSharedFile();
 199                                 if (sharedfile != "") {  inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 200                                 else {
 201                                         columnfile = m->getColumnFile();
 202                                         if (columnfile != "") { inputfile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 203                                         else {
 204                                                 phylipfile = m->getPhylipFile();
 205                                                 if (phylipfile != "") { inputfile = phylipfile;  format = "phylip";  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 206                                                 else {
 207                                                         m->mothurOut("No valid current files. You must provide a shared, phylip or column file."); m->mothurOutEndLine();
 208                                                         abort = true;
 209                                                 }
 210                                         }
 211                                 }
 212                         }
 213                         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; }
 214
 215                         if (columnfile != "") {
 216                                 if ((namefile == "") && (countfile == "")){
 217                                         namefile = m->getNameFile();
 218                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 219                                         else {
 220                                                 countfile = m->getCountTableFile();
 221                         if (countfile != "") {  m->mothurOut("Using " + countfile + " as input file for the count parameter."); m->mothurOutEndLine(); }
 222                         else {
 223                             m->mothurOut("You need to provide a namefile or countfile if you are going to use the column format."); m->mothurOutEndLine();
 224                             abort = true;
 225                         }
 226                                         }
 227                                 }
 228                         }
 229
 230
 231                         //check for optional parameter and set defaults
 232                         // ...at some point should added some additional type checking...
 233                         label = validParameter.validFile(parameters, "label", false);
 234                         if (label == "not found") { label = ""; }
 235                         else {
 236                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 237                                 else { allLines = 1;  }
 238                         }
 239
 240                         groups = validParameter.validFile(parameters, "groups", false);
 241                         if (groups == "not found") { groups = ""; }
 242                         else {
 243                                 m->splitAtDash(groups, Groups);
 244                                 m->setGroups(Groups);
 245                         }
 246
 247                         calc = validParameter.validFile(parameters, "calc", false);
 248                         if (calc == "not found") { calc = "jclass-thetayc";  }
 249                         else {
 250                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 251                         }
 252                         m->splitAtDash(calc, Estimators);
 253                         if (m->inUsersGroups("citation", Estimators)) {
 254                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 255                                 //remove citation from list of calcs
 256                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 257                         }
 258
 259                         string temp;
 260                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 261                         m->mothurConvert(temp, precision);
 262
 263                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 264                         m->mothurConvert(temp, cutoff);
 265                         cutoff += (5 / (precision * 10.0));
 266
 267             temp = validParameter.validFile(parameters, "processors", false);   if (temp == "not found"){       temp = m->getProcessors();      }
 268                         m->setProcessors(temp);
 269                         m->mothurConvert(temp, processors);
 270
 271             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 272                         m->mothurConvert(temp, iters);
 273
 274             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 275                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 276             else {
 277                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 278                 else { subsample = false; }
 279             }
 280
 281             if (subsample == false) { iters = 1; }
 282
 283             if (subsample && (format != "sharedfile")) { m->mothurOut("[ERROR]: the subsample parameter can only be used with a shared file.\n"); abort=true; }
 284
 285                         //if the user changes the output directory command factory will send this info to us in the output parameter
 286                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 287                                 outputDir = "";
 288                                 outputDir += m->hasPath(inputfile); //if user entered a file with a path then preserve it
 289                         }
 290                 }
 291
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 295                 exit(1);
 296         }
 297 }
 298 //**********************************************************************************************************************
 299
 300 TreeGroupCommand::~TreeGroupCommand(){
 301         if (abort == false) {
 302                 if (format == "sharedfile") {  delete input; }
 303                 else { delete list; }
 304                 delete ct;
 305         }
 306
 307 }
 308
 309 //**********************************************************************************************************************
 310
 311 int TreeGroupCommand::execute(){
 312         try {
 313
 314                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 315
 316                 if (format == "sharedfile") {
 317
 318                         ValidCalculators validCalculator;
 319
 320                         for (int i=0; i<Estimators.size(); i++) {
 321                                 if (validCalculator.isValidCalculator("treegroup", Estimators[i]) == true) {
 322                                         if (Estimators[i] == "sharedsobs") {
 323                                                 treeCalculators.push_back(new SharedSobsCS());
 324                                         }else if (Estimators[i] == "sharedchao") {
 325                                                 treeCalculators.push_back(new SharedChao1());
 326                                         }else if (Estimators[i] == "sharedace") {
 327                                                 treeCalculators.push_back(new SharedAce());
 328                                         }else if (Estimators[i] == "jabund") {
 329                                                 treeCalculators.push_back(new JAbund());
 330                                         }else if (Estimators[i] == "sorabund") {
 331                                                 treeCalculators.push_back(new SorAbund());
 332                                         }else if (Estimators[i] == "jclass") {
 333                                                 treeCalculators.push_back(new Jclass());
 334                                         }else if (Estimators[i] == "sorclass") {
 335                                                 treeCalculators.push_back(new SorClass());
 336                                         }else if (Estimators[i] == "jest") {
 337                                                 treeCalculators.push_back(new Jest());
 338                                         }else if (Estimators[i] == "sorest") {
 339                                                 treeCalculators.push_back(new SorEst());
 340                                         }else if (Estimators[i] == "thetayc") {
 341                                                 treeCalculators.push_back(new ThetaYC());
 342                                         }else if (Estimators[i] == "thetan") {
 343                                                 treeCalculators.push_back(new ThetaN());
 344                                         }else if (Estimators[i] == "kstest") {
 345                                                 treeCalculators.push_back(new KSTest());
 346                                         }else if (Estimators[i] == "sharednseqs") {
 347                                                 treeCalculators.push_back(new SharedNSeqs());
 348                                         }else if (Estimators[i] == "ochiai") {
 349                                                 treeCalculators.push_back(new Ochiai());
 350                                         }else if (Estimators[i] == "anderberg") {
 351                                                 treeCalculators.push_back(new Anderberg());
 352                                         }else if (Estimators[i] == "kulczynski") {
 353                                                 treeCalculators.push_back(new Kulczynski());
 354                                         }else if (Estimators[i] == "kulczynskicody") {
 355                                                 treeCalculators.push_back(new KulczynskiCody());
 356                                         }else if (Estimators[i] == "lennon") {
 357                                                 treeCalculators.push_back(new Lennon());
 358                                         }else if (Estimators[i] == "morisitahorn") {
 359                                                 treeCalculators.push_back(new MorHorn());
 360                                         }else if (Estimators[i] == "braycurtis") {
 361                                                 treeCalculators.push_back(new BrayCurtis());
 362                                         }else if (Estimators[i] == "whittaker") {
 363                                                 treeCalculators.push_back(new Whittaker());
 364                                         }else if (Estimators[i] == "odum") {
 365                                                 treeCalculators.push_back(new Odum());
 366                                         }else if (Estimators[i] == "canberra") {
 367                                                 treeCalculators.push_back(new Canberra());
 368                                         }else if (Estimators[i] == "structeuclidean") {
 369                                                 treeCalculators.push_back(new StructEuclidean());
 370                                         }else if (Estimators[i] == "structchord") {
 371                                                 treeCalculators.push_back(new StructChord());
 372                                         }else if (Estimators[i] == "hellinger") {
 373                                                 treeCalculators.push_back(new Hellinger());
 374                                         }else if (Estimators[i] == "manhattan") {
 375                                                 treeCalculators.push_back(new Manhattan());
 376                                         }else if (Estimators[i] == "structpearson") {
 377                                                 treeCalculators.push_back(new StructPearson());
 378                                         }else if (Estimators[i] == "soergel") {
 379                                                 treeCalculators.push_back(new Soergel());
 380                                         }else if (Estimators[i] == "spearman") {
 381                                                 treeCalculators.push_back(new Spearman());
 382                                         }else if (Estimators[i] == "structkulczynski") {
 383                                                 treeCalculators.push_back(new StructKulczynski());
 384                                         }else if (Estimators[i] == "speciesprofile") {
 385                                                 treeCalculators.push_back(new SpeciesProfile());
 386                                         }else if (Estimators[i] == "hamming") {
 387                                                 treeCalculators.push_back(new Hamming());
 388                                         }else if (Estimators[i] == "structchi2") {
 389                                                 treeCalculators.push_back(new StructChi2());
 390                                         }else if (Estimators[i] == "gower") {
 391                                                 treeCalculators.push_back(new Gower());
 392                                         }else if (Estimators[i] == "memchi2") {
 393                                                 treeCalculators.push_back(new MemChi2());
 394                                         }else if (Estimators[i] == "memchord") {
 395                                                 treeCalculators.push_back(new MemChord());
 396                                         }else if (Estimators[i] == "memeuclidean") {
 397                                                 treeCalculators.push_back(new MemEuclidean());
 398                                         }else if (Estimators[i] == "mempearson") {
 399                                                 treeCalculators.push_back(new MemPearson());
 400                                         }
 401                                 }
 402                         }
 403
 404                         //if the users entered no valid calculators don't execute command
 405                         if (treeCalculators.size() == 0) { m->mothurOut("You have given no valid calculators."); m->mothurOutEndLine(); return 0; }
 406
 407                         input = new InputData(sharedfile, "sharedfile");
 408                         lookup = input->getSharedRAbundVectors();
 409                         lastLabel = lookup[0]->getLabel();
 410
 411                         if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0; }
 412
 413                         //used in tree constructor
 414                         m->runParse = false;
 415
 416                         //create treemap class from groupmap for tree class to use
 417                         ct = new CountTable();
 418             set<string> nameMap;
 419             map<string, string> groupMap;
 420             set<string> gps;
 421             for (int i = 0; i < m->getAllGroups().size(); i++) {
 422                 nameMap.insert(m->getAllGroups()[i]);
 423                 gps.insert(m->getAllGroups()[i]);
 424                 groupMap[m->getAllGroups()[i]] = m->getAllGroups()[i];
 425             }
 426             ct->createTable(nameMap, groupMap, gps);
 427
 428                         //clear globaldatas old tree names if any
 429                         m->Treenames.clear();
 430
 431                         //fills globaldatas tree names
 432                         //m->Treenames = m->getGroups();
 433             for (int k = 0; k < lookup.size(); k++) {
 434                 m->Treenames.push_back(lookup[k]->getGroup());
 435             }
 436
 437                         if (m->control_pressed) { return 0; }
 438
 439                         //create tree file
 440                         makeSimsShared();
 441
 442                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 443                 }else{
 444                         //read in dist file
 445                         filename = inputfile;
 446
 447             ReadMatrix* readMatrix;
 448                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 449                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 450
 451                         readMatrix->setCutoff(cutoff);
 452
 453             ct = NULL;
 454             nameMap = NULL;
 455             if(namefile != ""){
 456                 nameMap = new NameAssignment(namefile);
 457                 nameMap->readMap();
 458                 readMatrix->read(nameMap);
 459             }else if (countfile != "") {
 460                 ct = new CountTable();
 461                 ct->readTable(countfile, true);
 462                 readMatrix->read(ct);
 463             }else {
 464                 readMatrix->read(nameMap);
 465             }
 466
 467                         list = readMatrix->getListVector();
 468                         SparseDistanceMatrix* dMatrix = readMatrix->getDMatrix();
 469
 470             //clear globaldatas old tree names if any
 471                         m->Treenames.clear();
 472
 473                         //make treemap
 474             if (ct != NULL) { delete ct; }
 475                         ct = new CountTable();
 476             set<string> nameMap;
 477             map<string, string> groupMap;
 478             set<string> gps;
 479             for (int i = 0; i < list->getNumBins(); i++) {
 480                 string bin = list->get(i);
 481                 nameMap.insert(bin);
 482                 gps.insert(bin);
 483                 groupMap[bin] = bin;
 484                 m->Treenames.push_back(bin);
 485             }
 486             ct->createTable(nameMap, groupMap, gps);
 487
 488                         vector<string> namesGroups = ct->getNamesOfGroups();
 489                         m->setGroups(namesGroups);
 490
 491                         //used in tree constructor
 492                         m->runParse = false;
 493
 494                         if (m->control_pressed) { return 0; }
 495
 496                         vector< vector<double> > matrix = makeSimsDist(dMatrix);
 497             delete readMatrix;
 498             delete dMatrix;
 499
 500                         if (m->control_pressed) { return 0; }
 501
 502                         //create a new filename
 503             map<string, string> variables;
 504             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 505                         string outputFile = getOutputFileName("tree",variables);
 506                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 507
 508                         Tree* newTree = createTree(matrix);
 509
 510             if (newTree != NULL) {  writeTree(outputFile, newTree); delete newTree; }
 511
 512                         if (m->control_pressed) { return 0; }
 513
 514                         m->mothurOut("Tree complete. "); m->mothurOutEndLine();
 515
 516                 }
 517
 518                 //reset groups parameter
 519                 m->clearGroups();
 520
 521                 //set tree file as new current treefile
 522                 string current = "";
 523                 itTypes = outputTypes.find("tree");
 524                 if (itTypes != outputTypes.end()) {
 525                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 526                 }
 527
 528                 m->mothurOutEndLine();
 529                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 530                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 531                 m->mothurOutEndLine();
 532
 533                 return 0;
 534         }
 535         catch(exception& e) {
 536                 m->errorOut(e, "TreeGroupCommand", "execute");
 537                 exit(1);
 538         }
 539 }
 540 //**********************************************************************************************************************
 541
 542 Tree* TreeGroupCommand::createTree(vector< vector<double> >& simMatrix){
 543         try {
 544                 //create tree
 545                 t = new Tree(ct, simMatrix);
 546
 547         if (m->control_pressed) { delete t; t = NULL; return t; }
 548
 549         //assemble tree
 550                 t->assembleTree();
 551
 552                 return t;
 553         }
 554         catch(exception& e) {
 555                 m->errorOut(e, "TreeGroupCommand", "createTree");
 556                 exit(1);
 557         }
 558 }
 559 /***********************************************************/
 560 int TreeGroupCommand::writeTree(string out, Tree* T) {
 561         try {
 562
 563         //print newick file
 564                 t->createNewickFile(out);
 565
 566         if (m->control_pressed) { m->mothurRemove(out); outputNames.pop_back(); return 1; }
 567
 568         return 0;
 569
 570         }
 571         catch(exception& e) {
 572                 m->errorOut(e, "TreeGroupCommand", "printSims");
 573                 exit(1);
 574         }
 575 }
 576
 577 /***********************************************************/
 578 void TreeGroupCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 579         try {
 580
 581         for (int m = 0; m < simMatrix.size(); m++)      {
 582                         //out << lookup[m]->getGroup() << '\t';
 583                         for (int n = 0; n < simMatrix.size(); n++)      {
 584                                 out << simMatrix[m][n] << '\t';
 585                         }
 586                         out << endl;
 587                 }
 588
 589         }
 590         catch(exception& e) {
 591                 m->errorOut(e, "TreeGroupCommand", "printSims");
 592                 exit(1);
 593         }
 594 }
 595 /***********************************************************/
 596 vector< vector<double> > TreeGroupCommand::makeSimsDist(SparseDistanceMatrix* matrix) {
 597         try {
 598                 numGroups = list->size();
 599
 600                 //initialize simMatrix
 601                 vector< vector<double> > simMatrix;
 602                 simMatrix.resize(numGroups);
 603                 for (int k = 0; k < simMatrix.size(); k++)      {
 604                         for (int j = 0; j < simMatrix.size(); j++)      {
 605                                 simMatrix[k].push_back(0.0);
 606                         }
 607                 }
 608
 609                 //go through sparse matrix and fill sims
 610                 //go through each cell in the sparsematrix
 611         for (int i = 0; i < matrix->seqVec.size(); i++) {
 612             for (int j = 0; j < matrix->seqVec[i].size(); j++) {
 613
 614                 //already checked everyone else in row
 615                 if (i < matrix->seqVec[i][j].index) {
 616                     simMatrix[i][matrix->seqVec[i][j].index] = -(matrix->seqVec[i][j].dist -1.0);
 617                     simMatrix[matrix->seqVec[i][j].index][i] = -(matrix->seqVec[i][j].dist -1.0);
 618
 619                     if (m->control_pressed) { return simMatrix; }
 620                 }
 621             }
 622                 }
 623
 624                 return simMatrix;
 625         }
 626         catch(exception& e) {
 627                 m->errorOut(e, "TreeGroupCommand", "makeSimsDist");
 628                 exit(1);
 629         }
 630 }
 631
 632 /***********************************************************/
 633 int TreeGroupCommand::makeSimsShared() {
 634         try {
 635
 636         if (subsample) {
 637             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 638                 subsampleSize = lookup[0]->getNumSeqs();
 639                 for (int i = 1; i < lookup.size(); i++) {
 640                     int thisSize = lookup[i]->getNumSeqs();
 641
 642                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 643                 }
 644             }else {
 645                 m->clearGroups();
 646                 Groups.clear();
 647                 m->Treenames.clear();
 648                 vector<SharedRAbundVector*> temp;
 649                 for (int i = 0; i < lookup.size(); i++) {
 650                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 651                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 652                         delete lookup[i];
 653                     }else {
 654                         Groups.push_back(lookup[i]->getGroup());
 655                         temp.push_back(lookup[i]);
 656                         m->Treenames.push_back(lookup[i]->getGroup());
 657                     }
 658                 }
 659                 lookup = temp;
 660                 m->setGroups(Groups);
 661             }
 662
 663             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; }
 664         }
 665
 666         numGroups = lookup.size();
 667                 lines.resize(processors);
 668                 for (int i = 0; i < processors; i++) {
 669                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 670                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 671                 }
 672
 673                 set<string> processedLabels;
 674                 set<string> userLabels = labels;
 675
 676                 //as long as you are not at the end of the file or done wih the lines you want
 677                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 678                         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; }
 679
 680                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 681                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 682                                 process(lookup);
 683
 684                                 processedLabels.insert(lookup[0]->getLabel());
 685                                 userLabels.erase(lookup[0]->getLabel());
 686                         }
 687
 688                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 689                                 string saveLabel = lookup[0]->getLabel();
 690
 691                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 692                                 lookup = input->getSharedRAbundVectors(lastLabel);
 693
 694                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 695                                 process(lookup);
 696
 697                                 processedLabels.insert(lookup[0]->getLabel());
 698                                 userLabels.erase(lookup[0]->getLabel());
 699
 700                                 //restore real lastlabel to save below
 701                                 lookup[0]->setLabel(saveLabel);
 702                         }
 703
 704                         lastLabel = lookup[0]->getLabel();
 705
 706                         //get next line to process
 707                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 708                         lookup = input->getSharedRAbundVectors();
 709                 }
 710
 711                 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; }
 712
 713                 //output error messages about any remaining user labels
 714                 set<string>::iterator it;
 715                 bool needToRun = false;
 716                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 717                         m->mothurOut("Your file does not include the label " + *it);
 718                         if (processedLabels.count(lastLabel) != 1) {
 719                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 720                                 needToRun = true;
 721                         }else {
 722                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 723                         }
 724                 }
 725
 726                 //run last label if you need to
 727                 if (needToRun == true)  {
 728                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 729                         lookup = input->getSharedRAbundVectors(lastLabel);
 730
 731                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 732                         process(lookup);
 733                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 734                 }
 735
 736                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 737
 738                 return 0;
 739         }
 740         catch(exception& e) {
 741                 m->errorOut(e, "TreeGroupCommand", "makeSimsShared");
 742                 exit(1);
 743         }
 744 }
 745
 746 /***********************************************************/
 747 int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 748         try{
 749                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 750         vector< vector<seqDist>  > calcDists; calcDists.resize(treeCalculators.size());
 751
 752         for (int thisIter = 0; thisIter < iters; thisIter++) {
 753
 754             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 755
 756             if (subsample) {
 757                 SubSample sample;
 758                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 759
 760                 //make copy of lookup so we don't get access violations
 761                 vector<SharedRAbundVector*> newLookup;
 762                 for (int k = 0; k < thisItersLookup.size(); k++) {
 763                     SharedRAbundVector* temp = new SharedRAbundVector();
 764                     temp->setLabel(thisItersLookup[k]->getLabel());
 765                     temp->setGroup(thisItersLookup[k]->getGroup());
 766                     newLookup.push_back(temp);
 767                 }
 768
 769                 //for each bin
 770                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 771                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 772                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 773                 }
 774
 775                 tempLabels = sample.getSample(newLookup, subsampleSize);
 776                 thisItersLookup = newLookup;
 777             }
 778
 779             if(processors == 1){
 780                 driver(thisItersLookup, 0, numGroups, calcDists);
 781             }else{
 782                 int process = 1;
 783                 vector<int> processIDS;
 784
 785 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 786                 //loop through and create all the processes you want
 787                 while (process != processors) {
 788                     int pid = fork();
 789
 790                     if (pid > 0) {
 791                         processIDS.push_back(pid);
 792                         process++;
 793                     }else if (pid == 0){
 794
 795                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 796
 797                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 798                         ofstream outtemp;
 799                         m->openOutputFile(tempdistFileName, outtemp);
 800
 801                         for (int i = 0; i < calcDists.size(); i++) {
 802                             outtemp << calcDists[i].size() << endl;
 803
 804                             for (int j = 0; j < calcDists[i].size(); j++) {
 805                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 806                             }
 807                         }
 808                         outtemp.close();
 809
 810                         exit(0);
 811                     }else {
 812                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 813                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 814                         exit(0);
 815                     }
 816                 }
 817
 818                 //parent do your part
 819                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 820
 821                 //force parent to wait until all the processes are done
 822                 for (int i = 0; i < processIDS.size(); i++) {
 823                     int temp = processIDS[i];
 824                     wait(&temp);
 825                 }
 826
 827                 for (int i = 0; i < processIDS.size(); i++) {
 828                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 829                     ifstream intemp;
 830                     m->openInputFile(tempdistFileName, intemp);
 831
 832                     for (int k = 0; k < calcDists.size(); k++) {
 833                         int size = 0;
 834                         intemp >> size; m->gobble(intemp);
 835
 836                         for (int j = 0; j < size; j++) {
 837                             int seq1 = 0;
 838                             int seq2 = 0;
 839                             float dist = 1.0;
 840
 841                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 842
 843                             seqDist tempDist(seq1, seq2, dist);
 844                             calcDists[k].push_back(tempDist);
 845                         }
 846                     }
 847                     intemp.close();
 848                     m->mothurRemove(tempdistFileName);
 849                 }
 850 #else
 851                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 852                 //Windows version shared memory, so be careful when passing variables through the treeSharedData struct.
 853                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 854                 //Taking advantage of shared memory to pass results vectors.
 855                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 856
 857                 vector<treeSharedData*> pDataArray;
 858                 DWORD   dwThreadIdArray[processors-1];
 859                 HANDLE  hThreadArray[processors-1];
 860
 861                 //Create processor worker threads.
 862                 for( int i=1; i<processors; i++ ){
 863
 864                     //make copy of lookup so we don't get access violations
 865                     vector<SharedRAbundVector*> newLookup;
 866                     for (int k = 0; k < thisItersLookup.size(); k++) {
 867                         SharedRAbundVector* temp = new SharedRAbundVector();
 868                         temp->setLabel(thisItersLookup[k]->getLabel());
 869                         temp->setGroup(thisItersLookup[k]->getGroup());
 870                         newLookup.push_back(temp);
 871                     }
 872
 873                     //for each bin
 874                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 875                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 876                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 877                     }
 878
 879                     // Allocate memory for thread data.
 880                     treeSharedData* tempSum = new treeSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 881                     pDataArray.push_back(tempSum);
 882                     processIDS.push_back(i);
 883
 884                     hThreadArray[i-1] = CreateThread(NULL, 0, MyTreeSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 885                 }
 886
 887                 //parent do your part
 888                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 889
 890                 //Wait until all threads have terminated.
 891                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 892
 893                 //Close all thread handles and free memory allocations.
 894                 for(int i=0; i < pDataArray.size(); i++){
 895                     if (pDataArray[i]->count != (pDataArray[i]->end-pDataArray[i]->start)) {
 896                         m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end-pDataArray[i]->start) + " groups assigned to it, quitting. \n"); m->control_pressed = true;
 897                     }
 898                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 899
 900                     for (int k = 0; k < calcDists.size(); k++) {
 901                         int size = pDataArray[i]->calcDists[k].size();
 902                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 903                     }
 904
 905                     CloseHandle(hThreadArray[i]);
 906                     delete pDataArray[i];
 907                 }
 908
 909 #endif
 910             }
 911
 912             calcDistsTotals.push_back(calcDists);
 913
 914             if (subsample) {
 915
 916                 //clean up memory
 917                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 918                 thisItersLookup.clear();
 919                 for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 920             }
 921                 }
 922
 923                 if (m->debug) {  m->mothurOut("[DEBUG]: done with iters.\n"); }
 924
 925         if (iters != 1) {
 926             //we need to find the average distance and standard deviation for each groups distance
 927             vector< vector<seqDist>  > calcAverages = m->getAverages(calcDistsTotals);
 928
 929             if (m->debug) {  m->mothurOut("[DEBUG]: found averages.\n"); }
 930
 931             //create average tree for each calc
 932             for (int i = 0; i < calcDists.size(); i++) {
 933                 vector< vector<double> > matrix; //square matrix to represent the distance
 934                 matrix.resize(thisLookup.size());
 935                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 936
 937                 for (int j = 0; j < calcAverages[i].size(); j++) {
 938                     int row = calcAverages[i][j].seq1;
 939                     int column = calcAverages[i][j].seq2;
 940                     float dist = calcAverages[i][j].dist;
 941
 942                     matrix[row][column] = dist;
 943                     matrix[column][row] = dist;
 944                 }
 945
 946                 //create a new filename
 947                 map<string, string> variables;
 948                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 949                 variables["[calc]"] = treeCalculators[i]->getName();
 950                 variables["[distance]"] = thisLookup[0]->getLabel();
 951                 variables["[tag]"] = "ave";
 952                 string outputFile = getOutputFileName("tree",variables);
 953                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 954
 955                 //creates tree from similarity matrix and write out file
 956                 Tree* newTree = createTree(matrix);
 957                 if (newTree != NULL) { writeTree(outputFile, newTree); }
 958             }
 959
 960             if (m->debug) {  m->mothurOut("[DEBUG]: done averages trees.\n"); }
 961
 962             //create all trees for each calc and find their consensus tree
 963             for (int i = 0; i < calcDists.size(); i++) {
 964                 if (m->control_pressed) { break; }
 965
 966                 //create a new filename
 967                 //create a new filename
 968                 map<string, string> variables;
 969                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 970                 variables["[calc]"] = treeCalculators[i]->getName();
 971                 variables["[distance]"] = thisLookup[0]->getLabel();
 972                 variables["[tag]"] = "all";
 973                 string outputFile = getOutputFileName("tree",variables);
 974                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 975
 976                 ofstream outAll;
 977                 m->openOutputFile(outputFile, outAll);
 978
 979                 vector<Tree*> trees;
 980                 for (int myIter = 0; myIter < iters; myIter++) {
 981
 982                     if(m->control_pressed) { break; }
 983
 984                     //initialize matrix
 985                     vector< vector<double> > matrix; //square matrix to represent the distance
 986                     matrix.resize(thisLookup.size());
 987                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 988
 989                     for (int j = 0; j < calcDistsTotals[myIter][i].size(); j++) {
 990                         int row = calcDistsTotals[myIter][i][j].seq1;
 991                         int column = calcDistsTotals[myIter][i][j].seq2;
 992                         double dist = calcDistsTotals[myIter][i][j].dist;
 993
 994                         matrix[row][column] = dist;
 995                         matrix[column][row] = dist;
 996                     }
 997
 998                     //creates tree from similarity matrix and write out file
 999                     Tree* newTree = createTree(matrix);
1000                     if (newTree != NULL) {
1001                         newTree->print(outAll);
1002                         trees.push_back(newTree);
1003                     }
1004                 }
1005                 outAll.close();
1006                 if (m->control_pressed) { for (int k = 0; k < trees.size(); k++) { delete trees[k]; } }
1007
1008                 if (m->debug) {  m->mothurOut("[DEBUG]: done all trees.\n"); }
1009
1010                 Consensus consensus;
1011                 //clear old tree names if any
1012                 m->Treenames.clear(); m->Treenames = m->getGroups(); //may have changed if subsample eliminated groups
1013                 Tree* conTree = consensus.getTree(trees);
1014
1015                 if (m->debug) {  m->mothurOut("[DEBUG]: done cons tree.\n"); }
1016
1017                 //create a new filename
1018                 variables["[tag]"] = "cons";
1019                 string conFile = getOutputFileName("tree",variables);
1020                 outputNames.push_back(conFile); outputTypes["tree"].push_back(conFile);
1021                 ofstream outTree;
1022                 m->openOutputFile(conFile, outTree);
1023
1024                 if (conTree != NULL) { conTree->print(outTree, "boot"); delete conTree; }
1025             }
1026
1027         }else {
1028
1029             for (int i = 0; i < calcDists.size(); i++) {
1030                 if (m->control_pressed) { break; }
1031
1032                 //initialize matrix
1033                 vector< vector<double> > matrix; //square matrix to represent the distance
1034                 matrix.resize(thisLookup.size());
1035                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
1036
1037                 for (int j = 0; j < calcDists[i].size(); j++) {
1038                     int row = calcDists[i][j].seq1;
1039                     int column = calcDists[i][j].seq2;
1040                     double dist = calcDists[i][j].dist;
1041
1042                     matrix[row][column] = dist;
1043                     matrix[column][row] = dist;
1044                 }
1045
1046                 //create a new filename
1047                 map<string, string> variables;
1048                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
1049                 variables["[calc]"] = treeCalculators[i]->getName();
1050                 variables["[distance]"] = thisLookup[0]->getLabel();
1051                 variables["[tag]"] = "";
1052                 string outputFile = getOutputFileName("tree",variables);
1053                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
1054
1055                 //creates tree from similarity matrix and write out file
1056                 Tree* newTree = createTree(matrix);
1057                 if (newTree != NULL) { writeTree(outputFile, newTree); delete newTree; }
1058             }
1059         }
1060
1061                 return 0;
1062         }
1063         catch(exception& e) {
1064                 m->errorOut(e, "TreeGroupCommand", "process");
1065                 exit(1);
1066         }
1067 }
1068 /**************************************************************************************************/
1069 int TreeGroupCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
1070         try {
1071                 vector<SharedRAbundVector*> subset;
1072                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
1073
1074                         for (int l = 0; l < k; l++) {
1075
1076                                 if (k != l) { //we dont need to similiarity of a groups to itself
1077                                         subset.clear(); //clear out old pair of sharedrabunds
1078                                         //add new pair of sharedrabunds
1079                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
1080
1081                                         for(int i=0;i<treeCalculators.size();i++) {
1082
1083                                                 //if this calc needs all groups to calculate the pair load all groups
1084                                                 if (treeCalculators[i]->getNeedsAll()) {
1085                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
1086                                                         for (int w = 0; w < thisLookup.size(); w++) {
1087                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
1088                                                         }
1089                                                 }
1090
1091                                                 vector<double> tempdata = treeCalculators[i]->getValues(subset); //saves the calculator outputs
1092
1093                                                 if (m->control_pressed) { return 1; }
1094
1095                                                 seqDist temp(l, k, -(tempdata[0]-1.0));
1096                                                 calcDists[i].push_back(temp);
1097                                         }
1098                                 }
1099                         }
1100                 }
1101
1102                 return 0;
1103         }
1104         catch(exception& e) {
1105                 m->errorOut(e, "TreeGroupCommand", "driver");
1106                 exit(1);
1107         }
1108 }
1109 /***********************************************************/
1110
1111
1112