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 "sharedsobscollectsummary.h"
  12 #include "sharedchao1.h"
  13 #include "sharedace.h"
  14 #include "sharednseqs.h"
  15 #include "sharedjabund.h"
  16 #include "sharedsorabund.h"
  17 #include "sharedjclass.h"
  18 #include "sharedsorclass.h"
  19 #include "sharedjest.h"
  20 #include "sharedsorest.h"
  21 #include "sharedthetayc.h"
  22 #include "sharedthetan.h"
  23 #include "sharedkstest.h"
  24 #include "whittaker.h"
  25 #include "sharedochiai.h"
  26 #include "sharedanderbergs.h"
  27 #include "sharedkulczynski.h"
  28 #include "sharedkulczynskicody.h"
  29 #include "sharedlennon.h"
  30 #include "sharedmorisitahorn.h"
  31 #include "sharedbraycurtis.h"
  32 #include "sharedjackknife.h"
  33 #include "whittaker.h"
  34 #include "odum.h"
  35 #include "canberra.h"
  36 #include "structeuclidean.h"
  37 #include "structchord.h"
  38 #include "hellinger.h"
  39 #include "manhattan.h"
  40 #include "structpearson.h"
  41 #include "soergel.h"
  42 #include "spearman.h"
  43 #include "structkulczynski.h"
  44 #include "structchi2.h"
  45 #include "speciesprofile.h"
  46 #include "hamming.h"
  47 #include "gower.h"
  48 #include "memchi2.h"
  49 #include "memchord.h"
  50 #include "memeuclidean.h"
  51 #include "mempearson.h"
  52
  53 //**********************************************************************************************************************
  54 vector<string> TreeGroupCommand::setParameters(){
  55         try {
  56                 CommandParameter pshared("shared", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pshared);
  57                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pphylip);
  58                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
  59                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "ColumnName",false,false); parameters.push_back(pcolumn);
  60                 CommandParameter pcutoff("cutoff", "Number", "", "10", "", "", "",false,false); parameters.push_back(pcutoff);
  61                 CommandParameter pprecision("precision", "Number", "", "100", "", "", "",false,false); parameters.push_back(pprecision);
  62                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  63                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  64                 CommandParameter pcalc("calc", "Multiple", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-whittaker-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-hamming-structchi2-gower-memchi2-memchord-memeuclidean-mempearson", "jclass-thetayc", "", "", "",true,false); parameters.push_back(pcalc);
  65                 CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput);
  66                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  67                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  68
  69                 vector<string> myArray;
  70                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  71                 return myArray;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "TreeGroupCommand", "setParameters");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 string TreeGroupCommand::getHelpString(){
  80         try {
  81                 string helpString = "";
  82                 ValidCalculators validCalculator;
  83                 helpString += "The tree.shared command creates a .tre to represent the similiarity between groups or sequences.\n";
  84                 helpString += "The tree.shared command parameters are shared, groups, calc, phylip, column, name, cutoff, precision and label.\n";
  85                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n";
  86                 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";
  87                 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";
  88                 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";
  89                 helpString += "The tree.shared command should be in the following format: tree.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n";
  90                 helpString += "Example tree.shared(groups=A-B-C, calc=jabund-sorabund).\n";
  91                 helpString += "The default value for groups is all the groups in your groupfile.\n";
  92                 helpString += "The default value for calc is jclass-thetayc.\n";
  93                 helpString += "The tree.shared command outputs a .tre file for each calculator you specify at each distance you choose.\n";
  94                 helpString += validCalculator.printCalc("treegroup");
  95                 helpString += "Or the tree.shared command can be in the following format: tree.shared(phylip=yourPhylipFile).\n";
  96                 helpString += "Example tree.shared(phylip=abrecovery.dist).\n";
  97                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  98                 return helpString;
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "TreeGroupCommand", "getHelpString");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106 TreeGroupCommand::TreeGroupCommand(){
 107         try {
 108                 abort = true; calledHelp = true;
 109                 setParameters();
 110                 //initialize outputTypes
 111                 vector<string> tempOutNames;
 112                 outputTypes["tree"] = tempOutNames;
 113         }
 114         catch(exception& e) {
 115                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 116                 exit(1);
 117         }
 118 }
 119 //**********************************************************************************************************************
 120
 121 TreeGroupCommand::TreeGroupCommand(string option)  {
 122         try {
 123                 abort = false; calledHelp = false;
 124                 allLines = 1;
 125
 126                 //allow user to run help
 127                 if(option == "help") { help(); abort = true; calledHelp = true; }
 128                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 129
 130                 else {
 131                         vector<string> myArray = setParameters();
 132
 133                         OptionParser parser(option);
 134                         map<string, string> parameters = parser. getParameters();
 135
 136                         ValidParameters validParameter;
 137                         map<string, string>::iterator it;
 138
 139                         //check to make sure all parameters are valid for command
 140                         for (it = parameters.begin(); it != parameters.end(); it++) {
 141                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 142                         }
 143
 144                         //initialize outputTypes
 145                         vector<string> tempOutNames;
 146                         outputTypes["tree"] = tempOutNames;
 147
 148                         //if the user changes the input directory command factory will send this info to us in the output parameter
 149                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 150                         if (inputDir == "not found"){   inputDir = "";          }
 151                         else {
 152                                 string path;
 153                                 it = parameters.find("phylip");
 154                                 //user has given a template file
 155                                 if(it != parameters.end()){
 156                                         path = m->hasPath(it->second);
 157                                         //if the user has not given a path then, add inputdir. else leave path alone.
 158                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 159                                 }
 160
 161                                 it = parameters.find("column");
 162                                 //user has given a template file
 163                                 if(it != parameters.end()){
 164                                         path = m->hasPath(it->second);
 165                                         //if the user has not given a path then, add inputdir. else leave path alone.
 166                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 167                                 }
 168
 169                                 it = parameters.find("name");
 170                                 //user has given a template file
 171                                 if(it != parameters.end()){
 172                                         path = m->hasPath(it->second);
 173                                         //if the user has not given a path then, add inputdir. else leave path alone.
 174                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 175                                 }
 176                         }
 177
 178                         //check for required parameters
 179                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 180                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 181                         else if (phylipfile == "not found") { phylipfile = ""; }
 182                         else {  inputfile = phylipfile;  format = "phylip"; m->setPhylipFile(phylipfile);       }
 183
 184                         columnfile = validParameter.validFile(parameters, "column", true);
 185                         if (columnfile == "not open") { columnfile = ""; abort = true; }
 186                         else if (columnfile == "not found") { columnfile = ""; }
 187                         else {  inputfile = columnfile; format = "column";      m->setColumnFile(columnfile); }
 188
 189                         sharedfile = validParameter.validFile(parameters, "shared", true);
 190                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 191                         else if (sharedfile == "not found") { sharedfile = ""; }
 192                         else {  inputfile = sharedfile; format = "sharedfile";  m->setSharedFile(sharedfile); }
 193
 194                         namefile = validParameter.validFile(parameters, "name", true);
 195                         if (namefile == "not open") { abort = true; }
 196                         else if (namefile == "not found") { namefile = ""; }
 197                         else { m->setNameFile(namefile); }
 198
 199                         if ((phylipfile == "") && (columnfile == "") && (sharedfile == "")) {
 200                                 //is there are current file available for either of these?
 201                                 //give priority to shared, then column, then phylip
 202                                 sharedfile = m->getSharedFile();
 203                                 if (sharedfile != "") {  inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 204                                 else {
 205                                         columnfile = m->getColumnFile();
 206                                         if (columnfile != "") { inputfile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 207                                         else {
 208                                                 phylipfile = m->getPhylipFile();
 209                                                 if (phylipfile != "") { inputfile = phylipfile;  format = "phylip";  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 210                                                 else {
 211                                                         m->mothurOut("No valid current files. You must provide a shared, phylip or column file."); m->mothurOutEndLine();
 212                                                         abort = true;
 213                                                 }
 214                                         }
 215                                 }
 216                         }
 217                         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; }
 218
 219                         if (columnfile != "") {
 220                                 if (namefile == "") {
 221                                         namefile = m->getNameFile();
 222                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 223                                         else {
 224                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 225                                                 abort = true;
 226                                         }
 227                                 }
 228                         }
 229
 230                         //check for optional parameter and set defaults
 231                         // ...at some point should added some additional type checking...
 232                         label = validParameter.validFile(parameters, "label", false);
 233                         if (label == "not found") { label = ""; }
 234                         else {
 235                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 236                                 else { allLines = 1;  }
 237                         }
 238
 239                         groups = validParameter.validFile(parameters, "groups", false);
 240                         if (groups == "not found") { groups = ""; }
 241                         else {
 242                                 m->splitAtDash(groups, Groups);
 243                                 m->setGroups(Groups);
 244                         }
 245
 246                         calc = validParameter.validFile(parameters, "calc", false);
 247                         if (calc == "not found") { calc = "jclass-thetayc";  }
 248                         else {
 249                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 250                         }
 251                         m->splitAtDash(calc, Estimators);
 252                         if (m->inUsersGroups("citation", Estimators)) {
 253                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 254                                 //remove citation from list of calcs
 255                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 256                         }
 257
 258                         string temp;
 259                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 260                         m->mothurConvert(temp, precision);
 261
 262                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 263                         m->mothurConvert(temp, cutoff);
 264                         cutoff += (5 / (precision * 10.0));
 265
 266                         //if the user changes the output directory command factory will send this info to us in the output parameter
 267                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 268                                 outputDir = "";
 269                                 outputDir += m->hasPath(inputfile); //if user entered a file with a path then preserve it
 270                         }
 271                 }
 272
 273         }
 274         catch(exception& e) {
 275                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 276                 exit(1);
 277         }
 278 }
 279 //**********************************************************************************************************************
 280
 281 TreeGroupCommand::~TreeGroupCommand(){
 282         if (abort == false) {
 283                 if (format == "sharedfile") {  delete input; }
 284                 else { delete readMatrix;  delete matrix; delete list; }
 285                 delete tmap;
 286         }
 287
 288 }
 289
 290 //**********************************************************************************************************************
 291
 292 int TreeGroupCommand::execute(){
 293         try {
 294
 295                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 296
 297                 if (format == "sharedfile") {
 298
 299                         ValidCalculators validCalculator;
 300
 301                         for (int i=0; i<Estimators.size(); i++) {
 302                                 if (validCalculator.isValidCalculator("treegroup", Estimators[i]) == true) {
 303                                         if (Estimators[i] == "sharedsobs") {
 304                                                 treeCalculators.push_back(new SharedSobsCS());
 305                                         }else if (Estimators[i] == "sharedchao") {
 306                                                 treeCalculators.push_back(new SharedChao1());
 307                                         }else if (Estimators[i] == "sharedace") {
 308                                                 treeCalculators.push_back(new SharedAce());
 309                                         }else if (Estimators[i] == "jabund") {
 310                                                 treeCalculators.push_back(new JAbund());
 311                                         }else if (Estimators[i] == "sorabund") {
 312                                                 treeCalculators.push_back(new SorAbund());
 313                                         }else if (Estimators[i] == "jclass") {
 314                                                 treeCalculators.push_back(new Jclass());
 315                                         }else if (Estimators[i] == "sorclass") {
 316                                                 treeCalculators.push_back(new SorClass());
 317                                         }else if (Estimators[i] == "jest") {
 318                                                 treeCalculators.push_back(new Jest());
 319                                         }else if (Estimators[i] == "sorest") {
 320                                                 treeCalculators.push_back(new SorEst());
 321                                         }else if (Estimators[i] == "thetayc") {
 322                                                 treeCalculators.push_back(new ThetaYC());
 323                                         }else if (Estimators[i] == "thetan") {
 324                                                 treeCalculators.push_back(new ThetaN());
 325                                         }else if (Estimators[i] == "kstest") {
 326                                                 treeCalculators.push_back(new KSTest());
 327                                         }else if (Estimators[i] == "sharednseqs") {
 328                                                 treeCalculators.push_back(new SharedNSeqs());
 329                                         }else if (Estimators[i] == "ochiai") {
 330                                                 treeCalculators.push_back(new Ochiai());
 331                                         }else if (Estimators[i] == "anderberg") {
 332                                                 treeCalculators.push_back(new Anderberg());
 333                                         }else if (Estimators[i] == "kulczynski") {
 334                                                 treeCalculators.push_back(new Kulczynski());
 335                                         }else if (Estimators[i] == "kulczynskicody") {
 336                                                 treeCalculators.push_back(new KulczynskiCody());
 337                                         }else if (Estimators[i] == "lennon") {
 338                                                 treeCalculators.push_back(new Lennon());
 339                                         }else if (Estimators[i] == "morisitahorn") {
 340                                                 treeCalculators.push_back(new MorHorn());
 341                                         }else if (Estimators[i] == "braycurtis") {
 342                                                 treeCalculators.push_back(new BrayCurtis());
 343                                         }else if (Estimators[i] == "whittaker") {
 344                                                 treeCalculators.push_back(new Whittaker());
 345                                         }else if (Estimators[i] == "odum") {
 346                                                 treeCalculators.push_back(new Odum());
 347                                         }else if (Estimators[i] == "canberra") {
 348                                                 treeCalculators.push_back(new Canberra());
 349                                         }else if (Estimators[i] == "structeuclidean") {
 350                                                 treeCalculators.push_back(new StructEuclidean());
 351                                         }else if (Estimators[i] == "structchord") {
 352                                                 treeCalculators.push_back(new StructChord());
 353                                         }else if (Estimators[i] == "hellinger") {
 354                                                 treeCalculators.push_back(new Hellinger());
 355                                         }else if (Estimators[i] == "manhattan") {
 356                                                 treeCalculators.push_back(new Manhattan());
 357                                         }else if (Estimators[i] == "structpearson") {
 358                                                 treeCalculators.push_back(new StructPearson());
 359                                         }else if (Estimators[i] == "soergel") {
 360                                                 treeCalculators.push_back(new Soergel());
 361                                         }else if (Estimators[i] == "spearman") {
 362                                                 treeCalculators.push_back(new Spearman());
 363                                         }else if (Estimators[i] == "structkulczynski") {
 364                                                 treeCalculators.push_back(new StructKulczynski());
 365                                         }else if (Estimators[i] == "speciesprofile") {
 366                                                 treeCalculators.push_back(new SpeciesProfile());
 367                                         }else if (Estimators[i] == "hamming") {
 368                                                 treeCalculators.push_back(new Hamming());
 369                                         }else if (Estimators[i] == "structchi2") {
 370                                                 treeCalculators.push_back(new StructChi2());
 371                                         }else if (Estimators[i] == "gower") {
 372                                                 treeCalculators.push_back(new Gower());
 373                                         }else if (Estimators[i] == "memchi2") {
 374                                                 treeCalculators.push_back(new MemChi2());
 375                                         }else if (Estimators[i] == "memchord") {
 376                                                 treeCalculators.push_back(new MemChord());
 377                                         }else if (Estimators[i] == "memeuclidean") {
 378                                                 treeCalculators.push_back(new MemEuclidean());
 379                                         }else if (Estimators[i] == "mempearson") {
 380                                                 treeCalculators.push_back(new MemPearson());
 381                                         }
 382                                 }
 383                         }
 384
 385                         //if the users entered no valid calculators don't execute command
 386                         if (treeCalculators.size() == 0) { m->mothurOut("You have given no valid calculators."); m->mothurOutEndLine(); return 0; }
 387
 388                         input = new InputData(sharedfile, "sharedfile");
 389                         lookup = input->getSharedRAbundVectors();
 390                         lastLabel = lookup[0]->getLabel();
 391
 392                         if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0; }
 393
 394                         //used in tree constructor
 395                         m->runParse = false;
 396
 397                         //create treemap class from groupmap for tree class to use
 398                         tmap = new TreeMap();
 399                         tmap->makeSim(m->getAllGroups());
 400
 401                         //clear globaldatas old tree names if any
 402                         m->Treenames.clear();
 403
 404                         //fills globaldatas tree names
 405                         m->Treenames = m->getGroups();
 406
 407                         if (m->control_pressed) { return 0; }
 408
 409                         //create tree file
 410                         makeSimsShared();
 411
 412                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 413                 }else{
 414                         //read in dist file
 415                         filename = inputfile;
 416
 417                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 418                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 419
 420                         readMatrix->setCutoff(cutoff);
 421
 422                         if(namefile != ""){
 423                                 nameMap = new NameAssignment(namefile);
 424                                 nameMap->readMap();
 425                         }
 426                         else{
 427                                 nameMap = NULL;
 428                         }
 429
 430                         readMatrix->read(nameMap);
 431                         list = readMatrix->getListVector();
 432                         matrix = readMatrix->getMatrix();
 433
 434                         //make treemap
 435                         tmap = new TreeMap();
 436
 437                         if (m->control_pressed) { return 0; }
 438
 439                         tmap->makeSim(list);
 440
 441                         vector<string> namesGroups = tmap->getNamesOfGroups();
 442                         m->setGroups(namesGroups);
 443
 444                         //clear globaldatas old tree names if any
 445                         m->Treenames.clear();
 446
 447                         //fills globaldatas tree names
 448                         m->Treenames = m->getGroups();
 449
 450                         //used in tree constructor
 451                         m->runParse = false;
 452
 453                         if (m->control_pressed) { return 0; }
 454
 455                         makeSimsDist();
 456
 457                         if (m->control_pressed) { return 0; }
 458
 459                         //create a new filename
 460                         outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + "tre";
 461                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 462
 463                         createTree();
 464
 465                         if (m->control_pressed) { return 0; }
 466
 467                         m->mothurOut("Tree complete. "); m->mothurOutEndLine();
 468
 469                 }
 470
 471                 //reset groups parameter
 472                 m->clearGroups();
 473
 474                 //set tree file as new current treefile
 475                 string current = "";
 476                 itTypes = outputTypes.find("tree");
 477                 if (itTypes != outputTypes.end()) {
 478                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 479                 }
 480
 481                 m->mothurOutEndLine();
 482                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 483                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 484                 m->mothurOutEndLine();
 485
 486                 return 0;
 487         }
 488         catch(exception& e) {
 489                 m->errorOut(e, "TreeGroupCommand", "execute");
 490                 exit(1);
 491         }
 492 }
 493 //**********************************************************************************************************************
 494
 495 int TreeGroupCommand::createTree(){
 496         try {
 497                 //create tree
 498                 t = new Tree(tmap);
 499
 500                 //do merges and create tree structure by setting parents and children
 501                 //there are numGroups - 1 merges to do
 502                 for (int i = 0; i < (numGroups - 1); i++) {
 503                         float largest = -1000.0;
 504
 505                         if (m->control_pressed) { delete t; return 1; }
 506
 507                         int row, column;
 508                         //find largest value in sims matrix by searching lower triangle
 509                         for (int j = 1; j < simMatrix.size(); j++) {
 510                                 for (int k = 0; k < j; k++) {
 511                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 512                                 }
 513                         }
 514
 515                         //set non-leaf node info and update leaves to know their parents
 516                         //non-leaf
 517                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 518
 519                         //parents
 520                         t->tree[index[row]].setParent(numGroups + i);
 521                         t->tree[index[column]].setParent(numGroups + i);
 522
 523                         //blength = distance / 2;
 524                         float blength = ((1.0 - largest) / 2);
 525
 526                         //branchlengths
 527                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 528                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 529
 530                         //set your length to leaves to your childs length plus branchlength
 531                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 532
 533
 534                         //update index
 535                         index[row] = numGroups+i;
 536                         index[column] = numGroups+i;
 537
 538                         //remove highest value that caused the merge.
 539                         simMatrix[row][column] = -1000.0;
 540                         simMatrix[column][row] = -1000.0;
 541
 542                         //merge values in simsMatrix
 543                         for (int n = 0; n < simMatrix.size(); n++)      {
 544                                 //row becomes merge of 2 groups
 545                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 546                                 simMatrix[n][row] = simMatrix[row][n];
 547                                 //delete column
 548                                 simMatrix[column][n] = -1000.0;
 549                                 simMatrix[n][column] = -1000.0;
 550                         }
 551                 }
 552
 553                 //adjust tree to make sure root to tip length is .5
 554                 int root = t->findRoot();
 555                 t->tree[root].setBranchLength((0.5 - t->tree[root].getLengthToLeaves()));
 556
 557                 //assemble tree
 558                 t->assembleTree();
 559
 560                 if (m->control_pressed) { delete t; return 1; }
 561
 562                 //print newick file
 563                 t->createNewickFile(outputFile);
 564
 565                 //delete tree
 566                 delete t;
 567
 568                 if (m->control_pressed) { m->mothurRemove(outputFile); outputNames.pop_back(); return 1; }
 569
 570                 return 0;
 571
 572         }
 573         catch(exception& e) {
 574                 m->errorOut(e, "TreeGroupCommand", "createTree");
 575                 exit(1);
 576         }
 577 }
 578 /***********************************************************/
 579 void TreeGroupCommand::printSims(ostream& out) {
 580         try {
 581
 582                 //output column headers
 583                 //out << '\t';
 584                 //for (int i = 0; i < lookup.size(); i++) {     out << lookup[i]->getGroup() << '\t';           }
 585                 //out << endl;
 586
 587
 588                 for (int m = 0; m < simMatrix.size(); m++)      {
 589                         //out << lookup[m]->getGroup() << '\t';
 590                         for (int n = 0; n < simMatrix.size(); n++)      {
 591                                 out << simMatrix[m][n] << '\t';
 592                         }
 593                         out << endl;
 594                 }
 595
 596         }
 597         catch(exception& e) {
 598                 m->errorOut(e, "TreeGroupCommand", "printSims");
 599                 exit(1);
 600         }
 601 }
 602 /***********************************************************/
 603 int TreeGroupCommand::makeSimsDist() {
 604         try {
 605                 numGroups = list->size();
 606
 607                 //initialize index
 608                 index.clear();
 609                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 610
 611                 //initialize simMatrix
 612                 simMatrix.clear();
 613                 simMatrix.resize(numGroups);
 614                 for (int k = 0; k < simMatrix.size(); k++)      {
 615                         for (int j = 0; j < simMatrix.size(); j++)      {
 616                                 simMatrix[k].push_back(0.0);
 617                         }
 618                 }
 619
 620                 //go through sparse matrix and fill sims
 621                 //go through each cell in the sparsematrix
 622                 for(MatData currentCell = matrix->begin(); currentCell != matrix->end(); currentCell++){
 623                         //similairity = -(distance-1)
 624                         simMatrix[currentCell->row][currentCell->column] = -(currentCell->dist -1.0);
 625                         simMatrix[currentCell->column][currentCell->row] = -(currentCell->dist -1.0);
 626
 627                         if (m->control_pressed) { return 1; }
 628
 629                 }
 630
 631                 return 0;
 632         }
 633         catch(exception& e) {
 634                 m->errorOut(e, "TreeGroupCommand", "makeSimsDist");
 635                 exit(1);
 636         }
 637 }
 638
 639 /***********************************************************/
 640 int TreeGroupCommand::makeSimsShared() {
 641         try {
 642                 set<string> processedLabels;
 643                 set<string> userLabels = labels;
 644
 645                 //as long as you are not at the end of the file or done wih the lines you want
 646                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 647                         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; }
 648
 649                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 650                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 651                                 process(lookup);
 652
 653                                 processedLabels.insert(lookup[0]->getLabel());
 654                                 userLabels.erase(lookup[0]->getLabel());
 655                         }
 656
 657                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 658                                 string saveLabel = lookup[0]->getLabel();
 659
 660                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 661                                 lookup = input->getSharedRAbundVectors(lastLabel);
 662
 663                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 664                                 process(lookup);
 665
 666                                 processedLabels.insert(lookup[0]->getLabel());
 667                                 userLabels.erase(lookup[0]->getLabel());
 668
 669                                 //restore real lastlabel to save below
 670                                 lookup[0]->setLabel(saveLabel);
 671                         }
 672
 673                         lastLabel = lookup[0]->getLabel();
 674
 675                         //get next line to process
 676                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 677                         lookup = input->getSharedRAbundVectors();
 678                 }
 679
 680                 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; }
 681
 682                 //output error messages about any remaining user labels
 683                 set<string>::iterator it;
 684                 bool needToRun = false;
 685                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 686                         m->mothurOut("Your file does not include the label " + *it);
 687                         if (processedLabels.count(lastLabel) != 1) {
 688                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 689                                 needToRun = true;
 690                         }else {
 691                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 692                         }
 693                 }
 694
 695                 //run last label if you need to
 696                 if (needToRun == true)  {
 697                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 698                         lookup = input->getSharedRAbundVectors(lastLabel);
 699
 700                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 701                         process(lookup);
 702                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 703                 }
 704
 705                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 706
 707                 return 0;
 708         }
 709         catch(exception& e) {
 710                 m->errorOut(e, "TreeGroupCommand", "makeSimsShared");
 711                 exit(1);
 712         }
 713 }
 714
 715 /***********************************************************/
 716 int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 717         try{
 718                                 EstOutput data;
 719                                 vector<SharedRAbundVector*> subset;
 720                                 numGroups = thisLookup.size();
 721
 722                                 //for each calculator
 723                                 for(int i = 0 ; i < treeCalculators.size(); i++) {
 724                                         //initialize simMatrix
 725                                         simMatrix.clear();
 726                                         simMatrix.resize(numGroups);
 727                                         for (int k = 0; k < simMatrix.size(); k++)      {
 728                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 729                                                         simMatrix[k].push_back(0.0);
 730                                                 }
 731                                         }
 732
 733                                         //initialize index
 734                                         index.clear();
 735                                         for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 736
 737                                         //create a new filename
 738                                         outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".tre";
 739                                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 740
 741                                         for (int k = 0; k < thisLookup.size(); k++) {
 742                                                 for (int l = k; l < thisLookup.size(); l++) {
 743                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 744                                                                 //get estimated similarity between 2 groups
 745
 746                                                                 subset.clear(); //clear out old pair of sharedrabunds
 747                                                                 //add new pair of sharedrabunds
 748                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 749
 750                                                                 //if this calc needs all groups to calculate the pair load all groups
 751                                                                 if (treeCalculators[i]->getNeedsAll()) {
 752                                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 753                                                                         for (int w = 0; w < thisLookup.size(); w++) {
 754                                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 755                                                                         }
 756                                                                 }
 757
 758                                                                 data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 759                                                 //cout << thisLookup[k]->getGroup() << '\t' << thisLookup[l]->getGroup() << '\t' << (1.0 - data[0]) << endl;
 760                                                                 if (m->control_pressed) { return 1; }
 761
 762                                                                 //save values in similarity matrix
 763                                                                 simMatrix[k][l] = -(data[0]-1.0);
 764                                                                 simMatrix[l][k] = -(data[0]-1.0);
 765                                                         }
 766                                                 }
 767                                         }
 768
 769                                         //createdistance file from simMatrix
 770                                         /*string o = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".dist";
 771                                         ofstream outDist;
 772                                         m->openOutputFile(o, outDist);
 773                                         outDist << simMatrix.size() << endl;
 774                                         for (int k = 0; k < simMatrix.size(); k++) {
 775                                                 outDist << thisLookup[k]->getGroup() << '\t';
 776                                                 for (int l = 0; l < k; l++) {
 777                                                         outDist << (1.0-simMatrix[k][l]) << '\t';
 778                                                 }
 779                                                 outDist << endl;
 780                                         }
 781                                         outDist.close();*/
 782
 783
 784                                         if (m->control_pressed) { return 1; }
 785                                         //creates tree from similarity matrix and write out file
 786                                         createTree();
 787
 788                                         if (m->control_pressed) { return 1; }
 789                                 }
 790
 791                                 return 0;
 792
 793         }
 794         catch(exception& e) {
 795                 m->errorOut(e, "TreeGroupCommand", "process");
 796                 exit(1);
 797         }
 798 }
 799 /***********************************************************/
 800
 801
 802