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
 434                         if (m->control_pressed) { return 0; }
 435
 436                         //create tree file
 437                         makeSimsShared();
 438
 439                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 440                 }else{
 441                         //read in dist file
 442                         filename = inputfile;
 443
 444             ReadMatrix* readMatrix;
 445                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 446                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 447
 448                         readMatrix->setCutoff(cutoff);
 449
 450             ct = NULL;
 451             nameMap = NULL;
 452             if(namefile != ""){
 453                 nameMap = new NameAssignment(namefile);
 454                 nameMap->readMap();
 455                 readMatrix->read(nameMap);
 456             }else if (countfile != "") {
 457                 ct = new CountTable();
 458                 ct->readTable(countfile);
 459                 readMatrix->read(ct);
 460             }else {
 461                 readMatrix->read(nameMap);
 462             }
 463
 464                         list = readMatrix->getListVector();
 465                         SparseDistanceMatrix* dMatrix = readMatrix->getDMatrix();
 466
 467                         //make treemap
 468             if (ct != NULL) { delete ct; }
 469                         ct = new CountTable();
 470             set<string> nameMap;
 471             map<string, string> groupMap;
 472             set<string> gps;
 473             for (int i = 0; i < list->getNumBins(); i++) {
 474                 string bin = list->get(i);
 475                 nameMap.insert(bin);
 476                 gps.insert(bin);
 477                 groupMap[bin] = bin;
 478             }
 479             ct->createTable(nameMap, groupMap, gps);
 480
 481                         vector<string> namesGroups = ct->getNamesOfGroups();
 482                         m->setGroups(namesGroups);
 483
 484                         //clear globaldatas old tree names if any
 485                         m->Treenames.clear();
 486
 487                         //fills globaldatas tree names
 488                         m->Treenames = m->getGroups();
 489
 490                         //used in tree constructor
 491                         m->runParse = false;
 492
 493                         if (m->control_pressed) { return 0; }
 494
 495                         vector< vector<double> > matrix = makeSimsDist(dMatrix);
 496             delete readMatrix;
 497             delete dMatrix;
 498
 499                         if (m->control_pressed) { return 0; }
 500
 501                         //create a new filename
 502             map<string, string> variables;
 503             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 504                         string outputFile = getOutputFileName("tree",variables);
 505                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 506
 507                         Tree* newTree = createTree(matrix);
 508
 509             if (newTree != NULL) {  writeTree(outputFile, newTree); delete newTree; }
 510
 511                         if (m->control_pressed) { return 0; }
 512
 513                         m->mothurOut("Tree complete. "); m->mothurOutEndLine();
 514
 515                 }
 516
 517                 //reset groups parameter
 518                 m->clearGroups();
 519
 520                 //set tree file as new current treefile
 521                 string current = "";
 522                 itTypes = outputTypes.find("tree");
 523                 if (itTypes != outputTypes.end()) {
 524                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 525                 }
 526
 527                 m->mothurOutEndLine();
 528                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 529                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 530                 m->mothurOutEndLine();
 531
 532                 return 0;
 533         }
 534         catch(exception& e) {
 535                 m->errorOut(e, "TreeGroupCommand", "execute");
 536                 exit(1);
 537         }
 538 }
 539 //**********************************************************************************************************************
 540
 541 Tree* TreeGroupCommand::createTree(vector< vector<double> >& simMatrix){
 542         try {
 543                 //create tree
 544                 t = new Tree(ct, simMatrix);
 545
 546         if (m->control_pressed) { delete t; t = NULL; return t; }
 547
 548         //assemble tree
 549                 t->assembleTree();
 550
 551                 return t;
 552         }
 553         catch(exception& e) {
 554                 m->errorOut(e, "TreeGroupCommand", "createTree");
 555                 exit(1);
 556         }
 557 }
 558 /***********************************************************/
 559 int TreeGroupCommand::writeTree(string out, Tree* T) {
 560         try {
 561
 562         //print newick file
 563                 t->createNewickFile(out);
 564
 565         if (m->control_pressed) { m->mothurRemove(out); outputNames.pop_back(); return 1; }
 566
 567         return 0;
 568
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "TreeGroupCommand", "printSims");
 572                 exit(1);
 573         }
 574 }
 575
 576 /***********************************************************/
 577 void TreeGroupCommand::printSims(ostream& out, vector< vector<double> >& simMatrix) {
 578         try {
 579
 580         for (int m = 0; m < simMatrix.size(); m++)      {
 581                         //out << lookup[m]->getGroup() << '\t';
 582                         for (int n = 0; n < simMatrix.size(); n++)      {
 583                                 out << simMatrix[m][n] << '\t';
 584                         }
 585                         out << endl;
 586                 }
 587
 588         }
 589         catch(exception& e) {
 590                 m->errorOut(e, "TreeGroupCommand", "printSims");
 591                 exit(1);
 592         }
 593 }
 594 /***********************************************************/
 595 vector< vector<double> > TreeGroupCommand::makeSimsDist(SparseDistanceMatrix* matrix) {
 596         try {
 597                 numGroups = list->size();
 598
 599                 //initialize simMatrix
 600                 vector< vector<double> > simMatrix;
 601                 simMatrix.resize(numGroups);
 602                 for (int k = 0; k < simMatrix.size(); k++)      {
 603                         for (int j = 0; j < simMatrix.size(); j++)      {
 604                                 simMatrix[k].push_back(0.0);
 605                         }
 606                 }
 607
 608                 //go through sparse matrix and fill sims
 609                 //go through each cell in the sparsematrix
 610         for (int i = 0; i < matrix->seqVec.size(); i++) {
 611             for (int j = 0; j < matrix->seqVec[i].size(); j++) {
 612
 613                 //already checked everyone else in row
 614                 if (i < matrix->seqVec[i][j].index) {
 615                     simMatrix[i][matrix->seqVec[i][j].index] = -(matrix->seqVec[i][j].dist -1.0);
 616                     simMatrix[matrix->seqVec[i][j].index][i] = -(matrix->seqVec[i][j].dist -1.0);
 617
 618                     if (m->control_pressed) { return simMatrix; }
 619                 }
 620             }
 621                 }
 622
 623                 return simMatrix;
 624         }
 625         catch(exception& e) {
 626                 m->errorOut(e, "TreeGroupCommand", "makeSimsDist");
 627                 exit(1);
 628         }
 629 }
 630
 631 /***********************************************************/
 632 int TreeGroupCommand::makeSimsShared() {
 633         try {
 634
 635         if (subsample) {
 636             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 637                 subsampleSize = lookup[0]->getNumSeqs();
 638                 for (int i = 1; i < lookup.size(); i++) {
 639                     int thisSize = lookup[i]->getNumSeqs();
 640
 641                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 642                 }
 643             }else {
 644                 m->clearGroups();
 645                 Groups.clear();
 646                 vector<SharedRAbundVector*> temp;
 647                 for (int i = 0; i < lookup.size(); i++) {
 648                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 649                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 650                         delete lookup[i];
 651                     }else {
 652                         Groups.push_back(lookup[i]->getGroup());
 653                         temp.push_back(lookup[i]);
 654                     }
 655                 }
 656                 lookup = temp;
 657                 m->setGroups(Groups);
 658             }
 659
 660             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; }
 661         }
 662
 663         numGroups = lookup.size();
 664                 lines.resize(processors);
 665                 for (int i = 0; i < processors; i++) {
 666                         lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
 667                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups);
 668                 }
 669
 670                 set<string> processedLabels;
 671                 set<string> userLabels = labels;
 672
 673                 //as long as you are not at the end of the file or done wih the lines you want
 674                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 675                         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; }
 676
 677                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 678                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 679                                 process(lookup);
 680
 681                                 processedLabels.insert(lookup[0]->getLabel());
 682                                 userLabels.erase(lookup[0]->getLabel());
 683                         }
 684
 685                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 686                                 string saveLabel = lookup[0]->getLabel();
 687
 688                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 689                                 lookup = input->getSharedRAbundVectors(lastLabel);
 690
 691                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 692                                 process(lookup);
 693
 694                                 processedLabels.insert(lookup[0]->getLabel());
 695                                 userLabels.erase(lookup[0]->getLabel());
 696
 697                                 //restore real lastlabel to save below
 698                                 lookup[0]->setLabel(saveLabel);
 699                         }
 700
 701                         lastLabel = lookup[0]->getLabel();
 702
 703                         //get next line to process
 704                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 705                         lookup = input->getSharedRAbundVectors();
 706                 }
 707
 708                 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; }
 709
 710                 //output error messages about any remaining user labels
 711                 set<string>::iterator it;
 712                 bool needToRun = false;
 713                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 714                         m->mothurOut("Your file does not include the label " + *it);
 715                         if (processedLabels.count(lastLabel) != 1) {
 716                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 717                                 needToRun = true;
 718                         }else {
 719                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 720                         }
 721                 }
 722
 723                 //run last label if you need to
 724                 if (needToRun == true)  {
 725                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 726                         lookup = input->getSharedRAbundVectors(lastLabel);
 727
 728                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 729                         process(lookup);
 730                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 731                 }
 732
 733                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 734
 735                 return 0;
 736         }
 737         catch(exception& e) {
 738                 m->errorOut(e, "TreeGroupCommand", "makeSimsShared");
 739                 exit(1);
 740         }
 741 }
 742
 743 /***********************************************************/
 744 int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 745         try{
 746                 vector< vector< vector<seqDist> > > calcDistsTotals;  //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
 747         vector< vector<seqDist>  > calcDists; calcDists.resize(treeCalculators.size());
 748
 749         for (int thisIter = 0; thisIter < iters; thisIter++) {
 750
 751             vector<SharedRAbundVector*> thisItersLookup = thisLookup;
 752
 753             if (subsample) {
 754                 SubSample sample;
 755                 vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
 756
 757                 //make copy of lookup so we don't get access violations
 758                 vector<SharedRAbundVector*> newLookup;
 759                 for (int k = 0; k < thisItersLookup.size(); k++) {
 760                     SharedRAbundVector* temp = new SharedRAbundVector();
 761                     temp->setLabel(thisItersLookup[k]->getLabel());
 762                     temp->setGroup(thisItersLookup[k]->getGroup());
 763                     newLookup.push_back(temp);
 764                 }
 765
 766                 //for each bin
 767                 for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 768                     if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 769                     for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 770                 }
 771
 772                 tempLabels = sample.getSample(newLookup, subsampleSize);
 773                 thisItersLookup = newLookup;
 774             }
 775
 776             if(processors == 1){
 777                 driver(thisItersLookup, 0, numGroups, calcDists);
 778             }else{
 779                 int process = 1;
 780                 vector<int> processIDS;
 781
 782 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 783                 //loop through and create all the processes you want
 784                 while (process != processors) {
 785                     int pid = fork();
 786
 787                     if (pid > 0) {
 788                         processIDS.push_back(pid);
 789                         process++;
 790                     }else if (pid == 0){
 791
 792                         driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
 793
 794                         string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".dist";
 795                         ofstream outtemp;
 796                         m->openOutputFile(tempdistFileName, outtemp);
 797
 798                         for (int i = 0; i < calcDists.size(); i++) {
 799                             outtemp << calcDists[i].size() << endl;
 800
 801                             for (int j = 0; j < calcDists[i].size(); j++) {
 802                                 outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
 803                             }
 804                         }
 805                         outtemp.close();
 806
 807                         exit(0);
 808                     }else {
 809                         m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 810                         for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 811                         exit(0);
 812                     }
 813                 }
 814
 815                 //parent do your part
 816                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 817
 818                 //force parent to wait until all the processes are done
 819                 for (int i = 0; i < processIDS.size(); i++) {
 820                     int temp = processIDS[i];
 821                     wait(&temp);
 822                 }
 823
 824                 for (int i = 0; i < processIDS.size(); i++) {
 825                     string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) +  ".dist";
 826                     ifstream intemp;
 827                     m->openInputFile(tempdistFileName, intemp);
 828
 829                     for (int k = 0; k < calcDists.size(); k++) {
 830                         int size = 0;
 831                         intemp >> size; m->gobble(intemp);
 832
 833                         for (int j = 0; j < size; j++) {
 834                             int seq1 = 0;
 835                             int seq2 = 0;
 836                             float dist = 1.0;
 837
 838                             intemp >> seq1 >> seq2 >> dist;   m->gobble(intemp);
 839
 840                             seqDist tempDist(seq1, seq2, dist);
 841                             calcDists[k].push_back(tempDist);
 842                         }
 843                     }
 844                     intemp.close();
 845                     m->mothurRemove(tempdistFileName);
 846                 }
 847 #else
 848                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 849                 //Windows version shared memory, so be careful when passing variables through the treeSharedData struct.
 850                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 851                 //Taking advantage of shared memory to pass results vectors.
 852                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 853
 854                 vector<treeSharedData*> pDataArray;
 855                 DWORD   dwThreadIdArray[processors-1];
 856                 HANDLE  hThreadArray[processors-1];
 857
 858                 //Create processor worker threads.
 859                 for( int i=1; i<processors; i++ ){
 860
 861                     //make copy of lookup so we don't get access violations
 862                     vector<SharedRAbundVector*> newLookup;
 863                     for (int k = 0; k < thisItersLookup.size(); k++) {
 864                         SharedRAbundVector* temp = new SharedRAbundVector();
 865                         temp->setLabel(thisItersLookup[k]->getLabel());
 866                         temp->setGroup(thisItersLookup[k]->getGroup());
 867                         newLookup.push_back(temp);
 868                     }
 869
 870                     //for each bin
 871                     for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
 872                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 873                         for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
 874                     }
 875
 876                     // Allocate memory for thread data.
 877                     treeSharedData* tempSum = new treeSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
 878                     pDataArray.push_back(tempSum);
 879                     processIDS.push_back(i);
 880
 881                     hThreadArray[i-1] = CreateThread(NULL, 0, MyTreeSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 882                 }
 883
 884                 //parent do your part
 885                 driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
 886
 887                 //Wait until all threads have terminated.
 888                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 889
 890                 //Close all thread handles and free memory allocations.
 891                 for(int i=0; i < pDataArray.size(); i++){
 892                     for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
 893
 894                     for (int k = 0; k < calcDists.size(); k++) {
 895                         int size = pDataArray[i]->calcDists[k].size();
 896                         for (int j = 0; j < size; j++) {    calcDists[k].push_back(pDataArray[i]->calcDists[k][j]);    }
 897                     }
 898
 899                     CloseHandle(hThreadArray[i]);
 900                     delete pDataArray[i];
 901                 }
 902
 903 #endif
 904             }
 905
 906             calcDistsTotals.push_back(calcDists);
 907
 908             if (subsample) {
 909
 910                 //clean up memory
 911                 for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
 912                 thisItersLookup.clear();
 913                 for (int i = 0; i < calcDists.size(); i++) {  calcDists[i].clear(); }
 914             }
 915                 }
 916
 917         if (iters != 1) {
 918             //we need to find the average distance and standard deviation for each groups distance
 919
 920             vector< vector<seqDist>  > calcAverages; calcAverages.resize(treeCalculators.size());
 921             for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 922                 calcAverages[i].resize(calcDistsTotals[0][i].size());
 923
 924                 for (int j = 0; j < calcAverages[i].size(); j++) {
 925                     calcAverages[i][j].seq1 = calcDists[i][j].seq1;
 926                     calcAverages[i][j].seq2 = calcDists[i][j].seq2;
 927                     calcAverages[i][j].dist = 0.0;
 928                 }
 929             }
 930
 931             for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 932                 for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 933                     for (int j = 0; j < calcAverages[i].size(); j++) {
 934                         calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist;
 935                     }
 936                 }
 937             }
 938
 939             for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 940                 for (int j = 0; j < calcAverages[i].size(); j++) {
 941                     calcAverages[i][j].dist /= (float) iters;
 942                 }
 943             }
 944
 945             //create average tree for each calc
 946             for (int i = 0; i < calcDists.size(); i++) {
 947                 vector< vector<double> > matrix; //square matrix to represent the distance
 948                 matrix.resize(thisLookup.size());
 949                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
 950
 951                 for (int j = 0; j < calcAverages[i].size(); j++) {
 952                     int row = calcAverages[i][j].seq1;
 953                     int column = calcAverages[i][j].seq2;
 954                     float dist = calcAverages[i][j].dist;
 955
 956                     matrix[row][column] = dist;
 957                     matrix[column][row] = dist;
 958                 }
 959
 960                 //create a new filename
 961                 map<string, string> variables;
 962                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 963                 variables["[calc]"] = treeCalculators[i]->getName();
 964                 variables["[distance]"] = thisLookup[0]->getLabel();
 965                 variables["[tag]"] = "ave";
 966                 string outputFile = getOutputFileName("tree",variables);
 967                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 968
 969                 //creates tree from similarity matrix and write out file
 970                 Tree* newTree = createTree(matrix);
 971                 if (newTree != NULL) { writeTree(outputFile, newTree); }
 972             }
 973
 974             //create all trees for each calc and find their consensus tree
 975             for (int i = 0; i < calcDists.size(); i++) {
 976                 if (m->control_pressed) { break; }
 977
 978                 //create a new filename
 979                 //create a new filename
 980                 map<string, string> variables;
 981                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 982                 variables["[calc]"] = treeCalculators[i]->getName();
 983                 variables["[distance]"] = thisLookup[0]->getLabel();
 984                 variables["[tag]"] = "all";
 985                 string outputFile = getOutputFileName("tree",variables);
 986                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 987
 988                 ofstream outAll;
 989                 m->openOutputFile(outputFile, outAll);
 990
 991                 vector<Tree*> trees;
 992                 for (int myIter = 0; myIter < iters; myIter++) {
 993
 994                     if(m->control_pressed) { break; }
 995
 996                     //initialize matrix
 997                     vector< vector<double> > matrix; //square matrix to represent the distance
 998                     matrix.resize(thisLookup.size());
 999                     for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
1000
1001                     for (int j = 0; j < calcDistsTotals[myIter][i].size(); j++) {
1002                         int row = calcDistsTotals[myIter][i][j].seq1;
1003                         int column = calcDistsTotals[myIter][i][j].seq2;
1004                         double dist = calcDistsTotals[myIter][i][j].dist;
1005
1006                         matrix[row][column] = dist;
1007                         matrix[column][row] = dist;
1008                     }
1009
1010                     //creates tree from similarity matrix and write out file
1011                     Tree* newTree = createTree(matrix);
1012                     if (newTree != NULL) {
1013                         newTree->print(outAll);
1014                         trees.push_back(newTree);
1015                     }
1016                 }
1017                 outAll.close();
1018                 if (m->control_pressed) { for (int k = 0; k < trees.size(); k++) { delete trees[k]; } }
1019
1020                 Consensus consensus;
1021                 //clear old tree names if any
1022                 m->Treenames.clear(); m->Treenames = m->getGroups(); //may have changed if subsample eliminated groups
1023                 Tree* conTree = consensus.getTree(trees);
1024
1025                 //create a new filename
1026                 variables["[tag]"] = "cons";
1027                 string conFile = getOutputFileName("tree",variables);
1028                 outputNames.push_back(conFile); outputTypes["tree"].push_back(conFile);
1029                 ofstream outTree;
1030                 m->openOutputFile(conFile, outTree);
1031
1032                 if (conTree != NULL) { conTree->print(outTree, "boot"); delete conTree; }
1033             }
1034
1035         }else {
1036
1037             for (int i = 0; i < calcDists.size(); i++) {
1038                 if (m->control_pressed) { break; }
1039
1040                 //initialize matrix
1041                 vector< vector<double> > matrix; //square matrix to represent the distance
1042                 matrix.resize(thisLookup.size());
1043                 for (int k = 0; k < thisLookup.size(); k++) {  matrix[k].resize(thisLookup.size(), 0.0); }
1044
1045                 for (int j = 0; j < calcDists[i].size(); j++) {
1046                     int row = calcDists[i][j].seq1;
1047                     int column = calcDists[i][j].seq2;
1048                     double dist = calcDists[i][j].dist;
1049
1050                     matrix[row][column] = dist;
1051                     matrix[column][row] = dist;
1052                 }
1053
1054                 //create a new filename
1055                 map<string, string> variables;
1056                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
1057                 variables["[calc]"] = treeCalculators[i]->getName();
1058                 variables["[distance]"] = thisLookup[0]->getLabel();
1059                 variables["[tag]"] = "";
1060                 string outputFile = getOutputFileName("tree",variables);
1061                 outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
1062
1063                 //creates tree from similarity matrix and write out file
1064                 Tree* newTree = createTree(matrix);
1065                 if (newTree != NULL) { writeTree(outputFile, newTree); delete newTree; }
1066             }
1067         }
1068
1069                 return 0;
1070         }
1071         catch(exception& e) {
1072                 m->errorOut(e, "TreeGroupCommand", "process");
1073                 exit(1);
1074         }
1075 }
1076 /**************************************************************************************************/
1077 int TreeGroupCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, vector< vector<seqDist> >& calcDists) {
1078         try {
1079                 vector<SharedRAbundVector*> subset;
1080                 for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
1081
1082                         for (int l = 0; l < k; l++) {
1083
1084                                 if (k != l) { //we dont need to similiarity of a groups to itself
1085                                         subset.clear(); //clear out old pair of sharedrabunds
1086                                         //add new pair of sharedrabunds
1087                                         subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
1088
1089                                         for(int i=0;i<treeCalculators.size();i++) {
1090
1091                                                 //if this calc needs all groups to calculate the pair load all groups
1092                                                 if (treeCalculators[i]->getNeedsAll()) {
1093                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
1094                                                         for (int w = 0; w < thisLookup.size(); w++) {
1095                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
1096                                                         }
1097                                                 }
1098
1099                                                 vector<double> tempdata = treeCalculators[i]->getValues(subset); //saves the calculator outputs
1100
1101                                                 if (m->control_pressed) { return 1; }
1102
1103                                                 seqDist temp(l, k, -(tempdata[0]-1.0));
1104                                                 calcDists[i].push_back(temp);
1105                                         }
1106                                 }
1107                         }
1108                 }
1109
1110                 return 0;
1111         }
1112         catch(exception& e) {
1113                 m->errorOut(e, "TreeGroupCommand", "driver");
1114                 exit(1);
1115         }
1116 }
1117 /***********************************************************/
1118
1119
1120