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";     }
 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";      }
 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";  }
 193
 194                         namefile = validParameter.validFile(parameters, "name", true);
 195                         if (namefile == "not open") { abort = true; }
 196                         else if (namefile == "not found") { namefile = ""; }
 197
 198                         if ((phylipfile == "") && (columnfile == "") && (sharedfile == "")) {
 199                                 //is there are current file available for either of these?
 200                                 //give priority to shared, then column, then phylip
 201                                 sharedfile = m->getSharedFile();
 202                                 if (sharedfile != "") {  inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 203                                 else {
 204                                         columnfile = m->getColumnFile();
 205                                         if (columnfile != "") { inputfile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 206                                         else {
 207                                                 phylipfile = m->getPhylipFile();
 208                                                 if (phylipfile != "") { inputfile = phylipfile;  format = "phylip";  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 209                                                 else {
 210                                                         m->mothurOut("No valid current files. You must provide a shared, phylip or column file."); m->mothurOutEndLine();
 211                                                         abort = true;
 212                                                 }
 213                                         }
 214                                 }
 215                         }
 216                         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; }
 217
 218                         if (columnfile != "") {
 219                                 if (namefile == "") {
 220                                         namefile = m->getNameFile();
 221                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 222                                         else {
 223                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 224                                                 abort = true;
 225                                         }
 226                                 }
 227                         }
 228
 229                         //check for optional parameter and set defaults
 230                         // ...at some point should added some additional type checking...
 231                         label = validParameter.validFile(parameters, "label", false);
 232                         if (label == "not found") { label = ""; }
 233                         else {
 234                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 235                                 else { allLines = 1;  }
 236                         }
 237
 238                         groups = validParameter.validFile(parameters, "groups", false);
 239                         if (groups == "not found") { groups = ""; }
 240                         else {
 241                                 m->splitAtDash(groups, Groups);
 242                                 m->Groups = Groups;
 243                         }
 244
 245                         calc = validParameter.validFile(parameters, "calc", false);
 246                         if (calc == "not found") { calc = "jclass-thetayc";  }
 247                         else {
 248                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 249                         }
 250                         m->splitAtDash(calc, Estimators);
 251                         if (m->inUsersGroups("citation", Estimators)) {
 252                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 253                                 //remove citation from list of calcs
 254                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 255                         }
 256
 257                         string temp;
 258                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 259                         convert(temp, precision);
 260
 261                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 262                         convert(temp, cutoff);
 263                         cutoff += (5 / (precision * 10.0));
 264
 265                         //if the user changes the output directory command factory will send this info to us in the output parameter
 266                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 267                                 outputDir = "";
 268                                 outputDir += m->hasPath(inputfile); //if user entered a file with a path then preserve it
 269                         }
 270                 }
 271
 272         }
 273         catch(exception& e) {
 274                 m->errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 275                 exit(1);
 276         }
 277 }
 278 //**********************************************************************************************************************
 279
 280 TreeGroupCommand::~TreeGroupCommand(){
 281         if (abort == false) {
 282                 if (format == "sharedfile") {  delete input; }
 283                 else { delete readMatrix;  delete matrix; delete list; }
 284                 delete tmap;
 285         }
 286
 287 }
 288
 289 //**********************************************************************************************************************
 290
 291 int TreeGroupCommand::execute(){
 292         try {
 293
 294                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 295
 296                 if (format == "sharedfile") {
 297
 298                         ValidCalculators validCalculator;
 299
 300                         for (int i=0; i<Estimators.size(); i++) {
 301                                 if (validCalculator.isValidCalculator("treegroup", Estimators[i]) == true) {
 302                                         if (Estimators[i] == "sharedsobs") {
 303                                                 treeCalculators.push_back(new SharedSobsCS());
 304                                         }else if (Estimators[i] == "sharedchao") {
 305                                                 treeCalculators.push_back(new SharedChao1());
 306                                         }else if (Estimators[i] == "sharedace") {
 307                                                 treeCalculators.push_back(new SharedAce());
 308                                         }else if (Estimators[i] == "jabund") {
 309                                                 treeCalculators.push_back(new JAbund());
 310                                         }else if (Estimators[i] == "sorabund") {
 311                                                 treeCalculators.push_back(new SorAbund());
 312                                         }else if (Estimators[i] == "jclass") {
 313                                                 treeCalculators.push_back(new Jclass());
 314                                         }else if (Estimators[i] == "sorclass") {
 315                                                 treeCalculators.push_back(new SorClass());
 316                                         }else if (Estimators[i] == "jest") {
 317                                                 treeCalculators.push_back(new Jest());
 318                                         }else if (Estimators[i] == "sorest") {
 319                                                 treeCalculators.push_back(new SorEst());
 320                                         }else if (Estimators[i] == "thetayc") {
 321                                                 treeCalculators.push_back(new ThetaYC());
 322                                         }else if (Estimators[i] == "thetan") {
 323                                                 treeCalculators.push_back(new ThetaN());
 324                                         }else if (Estimators[i] == "kstest") {
 325                                                 treeCalculators.push_back(new KSTest());
 326                                         }else if (Estimators[i] == "sharednseqs") {
 327                                                 treeCalculators.push_back(new SharedNSeqs());
 328                                         }else if (Estimators[i] == "ochiai") {
 329                                                 treeCalculators.push_back(new Ochiai());
 330                                         }else if (Estimators[i] == "anderberg") {
 331                                                 treeCalculators.push_back(new Anderberg());
 332                                         }else if (Estimators[i] == "kulczynski") {
 333                                                 treeCalculators.push_back(new Kulczynski());
 334                                         }else if (Estimators[i] == "kulczynskicody") {
 335                                                 treeCalculators.push_back(new KulczynskiCody());
 336                                         }else if (Estimators[i] == "lennon") {
 337                                                 treeCalculators.push_back(new Lennon());
 338                                         }else if (Estimators[i] == "morisitahorn") {
 339                                                 treeCalculators.push_back(new MorHorn());
 340                                         }else if (Estimators[i] == "braycurtis") {
 341                                                 treeCalculators.push_back(new BrayCurtis());
 342                                         }else if (Estimators[i] == "whittaker") {
 343                                                 treeCalculators.push_back(new Whittaker());
 344                                         }else if (Estimators[i] == "odum") {
 345                                                 treeCalculators.push_back(new Odum());
 346                                         }else if (Estimators[i] == "canberra") {
 347                                                 treeCalculators.push_back(new Canberra());
 348                                         }else if (Estimators[i] == "structeuclidean") {
 349                                                 treeCalculators.push_back(new StructEuclidean());
 350                                         }else if (Estimators[i] == "structchord") {
 351                                                 treeCalculators.push_back(new StructChord());
 352                                         }else if (Estimators[i] == "hellinger") {
 353                                                 treeCalculators.push_back(new Hellinger());
 354                                         }else if (Estimators[i] == "manhattan") {
 355                                                 treeCalculators.push_back(new Manhattan());
 356                                         }else if (Estimators[i] == "structpearson") {
 357                                                 treeCalculators.push_back(new StructPearson());
 358                                         }else if (Estimators[i] == "soergel") {
 359                                                 treeCalculators.push_back(new Soergel());
 360                                         }else if (Estimators[i] == "spearman") {
 361                                                 treeCalculators.push_back(new Spearman());
 362                                         }else if (Estimators[i] == "structkulczynski") {
 363                                                 treeCalculators.push_back(new StructKulczynski());
 364                                         }else if (Estimators[i] == "speciesprofile") {
 365                                                 treeCalculators.push_back(new SpeciesProfile());
 366                                         }else if (Estimators[i] == "hamming") {
 367                                                 treeCalculators.push_back(new Hamming());
 368                                         }else if (Estimators[i] == "structchi2") {
 369                                                 treeCalculators.push_back(new StructChi2());
 370                                         }else if (Estimators[i] == "gower") {
 371                                                 treeCalculators.push_back(new Gower());
 372                                         }else if (Estimators[i] == "memchi2") {
 373                                                 treeCalculators.push_back(new MemChi2());
 374                                         }else if (Estimators[i] == "memchord") {
 375                                                 treeCalculators.push_back(new MemChord());
 376                                         }else if (Estimators[i] == "memeuclidean") {
 377                                                 treeCalculators.push_back(new MemEuclidean());
 378                                         }else if (Estimators[i] == "mempearson") {
 379                                                 treeCalculators.push_back(new MemPearson());
 380                                         }
 381                                 }
 382                         }
 383
 384                         //if the users entered no valid calculators don't execute command
 385                         if (treeCalculators.size() == 0) { m->mothurOut("You have given no valid calculators."); m->mothurOutEndLine(); return 0; }
 386
 387                         input = new InputData(sharedfile, "sharedfile");
 388                         lookup = input->getSharedRAbundVectors();
 389                         lastLabel = lookup[0]->getLabel();
 390
 391                         if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0; }
 392
 393                         //used in tree constructor
 394                         m->runParse = false;
 395
 396                         //create treemap class from groupmap for tree class to use
 397                         tmap = new TreeMap();
 398                         tmap->makeSim(m->namesOfGroups);
 399
 400                         //clear globaldatas old tree names if any
 401                         m->Treenames.clear();
 402
 403                         //fills globaldatas tree names
 404                         m->Treenames = m->Groups;
 405
 406                         if (m->control_pressed) { return 0; }
 407
 408                         //create tree file
 409                         makeSimsShared();
 410
 411                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 412                 }else{
 413                         //read in dist file
 414                         filename = inputfile;
 415
 416                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 417                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 418
 419                         readMatrix->setCutoff(cutoff);
 420
 421                         if(namefile != ""){
 422                                 nameMap = new NameAssignment(namefile);
 423                                 nameMap->readMap();
 424                         }
 425                         else{
 426                                 nameMap = NULL;
 427                         }
 428
 429                         readMatrix->read(nameMap);
 430                         list = readMatrix->getListVector();
 431                         matrix = readMatrix->getMatrix();
 432
 433                         //make treemap
 434                         tmap = new TreeMap();
 435
 436                         if (m->control_pressed) { return 0; }
 437
 438                         tmap->makeSim(list);
 439
 440                         m->Groups = tmap->namesOfGroups;
 441
 442                         //clear globaldatas old tree names if any
 443                         m->Treenames.clear();
 444
 445                         //fills globaldatas tree names
 446                         m->Treenames = m->Groups;
 447
 448                         //used in tree constructor
 449                         m->runParse = false;
 450
 451                         if (m->control_pressed) { return 0; }
 452
 453                         makeSimsDist();
 454
 455                         if (m->control_pressed) { return 0; }
 456
 457                         //create a new filename
 458                         outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + "tre";
 459                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 460
 461                         createTree();
 462
 463                         if (m->control_pressed) { return 0; }
 464
 465                         m->mothurOut("Tree complete. "); m->mothurOutEndLine();
 466
 467                 }
 468
 469                 //reset groups parameter
 470                 m->Groups.clear();
 471
 472                 //set tree file as new current treefile
 473                 string current = "";
 474                 itTypes = outputTypes.find("tree");
 475                 if (itTypes != outputTypes.end()) {
 476                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 477                 }
 478
 479                 m->mothurOutEndLine();
 480                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 481                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 482                 m->mothurOutEndLine();
 483
 484                 return 0;
 485         }
 486         catch(exception& e) {
 487                 m->errorOut(e, "TreeGroupCommand", "execute");
 488                 exit(1);
 489         }
 490 }
 491 //**********************************************************************************************************************
 492
 493 int TreeGroupCommand::createTree(){
 494         try {
 495                 //create tree
 496                 t = new Tree(tmap);
 497
 498                 //do merges and create tree structure by setting parents and children
 499                 //there are numGroups - 1 merges to do
 500                 for (int i = 0; i < (numGroups - 1); i++) {
 501                         float largest = -1000.0;
 502
 503                         if (m->control_pressed) { delete t; return 1; }
 504
 505                         int row, column;
 506                         //find largest value in sims matrix by searching lower triangle
 507                         for (int j = 1; j < simMatrix.size(); j++) {
 508                                 for (int k = 0; k < j; k++) {
 509                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 510                                 }
 511                         }
 512
 513                         //set non-leaf node info and update leaves to know their parents
 514                         //non-leaf
 515                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 516
 517                         //parents
 518                         t->tree[index[row]].setParent(numGroups + i);
 519                         t->tree[index[column]].setParent(numGroups + i);
 520
 521                         //blength = distance / 2;
 522                         float blength = ((1.0 - largest) / 2);
 523
 524                         //branchlengths
 525                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 526                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 527
 528                         //set your length to leaves to your childs length plus branchlength
 529                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 530
 531
 532                         //update index
 533                         index[row] = numGroups+i;
 534                         index[column] = numGroups+i;
 535
 536                         //remove highest value that caused the merge.
 537                         simMatrix[row][column] = -1000.0;
 538                         simMatrix[column][row] = -1000.0;
 539
 540                         //merge values in simsMatrix
 541                         for (int n = 0; n < simMatrix.size(); n++)      {
 542                                 //row becomes merge of 2 groups
 543                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 544                                 simMatrix[n][row] = simMatrix[row][n];
 545                                 //delete column
 546                                 simMatrix[column][n] = -1000.0;
 547                                 simMatrix[n][column] = -1000.0;
 548                         }
 549                 }
 550
 551                 //adjust tree to make sure root to tip length is .5
 552                 int root = t->findRoot();
 553                 t->tree[root].setBranchLength((0.5 - t->tree[root].getLengthToLeaves()));
 554
 555                 //assemble tree
 556                 t->assembleTree();
 557
 558                 if (m->control_pressed) { delete t; return 1; }
 559
 560                 //print newick file
 561                 t->createNewickFile(outputFile);
 562
 563                 //delete tree
 564                 delete t;
 565
 566                 if (m->control_pressed) { remove(outputFile.c_str()); outputNames.pop_back(); return 1; }
 567
 568                 return 0;
 569
 570         }
 571         catch(exception& e) {
 572                 m->errorOut(e, "TreeGroupCommand", "createTree");
 573                 exit(1);
 574         }
 575 }
 576 /***********************************************************/
 577 void TreeGroupCommand::printSims(ostream& out) {
 578         try {
 579
 580                 //output column headers
 581                 //out << '\t';
 582                 //for (int i = 0; i < lookup.size(); i++) {     out << lookup[i]->getGroup() << '\t';           }
 583                 //out << endl;
 584
 585
 586                 for (int m = 0; m < simMatrix.size(); m++)      {
 587                         //out << lookup[m]->getGroup() << '\t';
 588                         for (int n = 0; n < simMatrix.size(); n++)      {
 589                                 out << simMatrix[m][n] << '\t';
 590                         }
 591                         out << endl;
 592                 }
 593
 594         }
 595         catch(exception& e) {
 596                 m->errorOut(e, "TreeGroupCommand", "printSims");
 597                 exit(1);
 598         }
 599 }
 600 /***********************************************************/
 601 int TreeGroupCommand::makeSimsDist() {
 602         try {
 603                 numGroups = list->size();
 604
 605                 //initialize index
 606                 index.clear();
 607                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 608
 609                 //initialize simMatrix
 610                 simMatrix.clear();
 611                 simMatrix.resize(numGroups);
 612                 for (int k = 0; k < simMatrix.size(); k++)      {
 613                         for (int j = 0; j < simMatrix.size(); j++)      {
 614                                 simMatrix[k].push_back(0.0);
 615                         }
 616                 }
 617
 618                 //go through sparse matrix and fill sims
 619                 //go through each cell in the sparsematrix
 620                 for(MatData currentCell = matrix->begin(); currentCell != matrix->end(); currentCell++){
 621                         //similairity = -(distance-1)
 622                         simMatrix[currentCell->row][currentCell->column] = -(currentCell->dist -1.0);
 623                         simMatrix[currentCell->column][currentCell->row] = -(currentCell->dist -1.0);
 624
 625                         if (m->control_pressed) { return 1; }
 626
 627                 }
 628
 629                 return 0;
 630         }
 631         catch(exception& e) {
 632                 m->errorOut(e, "TreeGroupCommand", "makeSimsDist");
 633                 exit(1);
 634         }
 635 }
 636
 637 /***********************************************************/
 638 int TreeGroupCommand::makeSimsShared() {
 639         try {
 640                 set<string> processedLabels;
 641                 set<string> userLabels = labels;
 642
 643                 //as long as you are not at the end of the file or done wih the lines you want
 644                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 645                         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; }
 646
 647                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 648                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 649                                 process(lookup);
 650
 651                                 processedLabels.insert(lookup[0]->getLabel());
 652                                 userLabels.erase(lookup[0]->getLabel());
 653                         }
 654
 655                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 656                                 string saveLabel = lookup[0]->getLabel();
 657
 658                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 659                                 lookup = input->getSharedRAbundVectors(lastLabel);
 660
 661                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 662                                 process(lookup);
 663
 664                                 processedLabels.insert(lookup[0]->getLabel());
 665                                 userLabels.erase(lookup[0]->getLabel());
 666
 667                                 //restore real lastlabel to save below
 668                                 lookup[0]->setLabel(saveLabel);
 669                         }
 670
 671                         lastLabel = lookup[0]->getLabel();
 672
 673                         //get next line to process
 674                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 675                         lookup = input->getSharedRAbundVectors();
 676                 }
 677
 678                 if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; } return 1; }
 679
 680                 //output error messages about any remaining user labels
 681                 set<string>::iterator it;
 682                 bool needToRun = false;
 683                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 684                         m->mothurOut("Your file does not include the label " + *it);
 685                         if (processedLabels.count(lastLabel) != 1) {
 686                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 687                                 needToRun = true;
 688                         }else {
 689                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 690                         }
 691                 }
 692
 693                 //run last label if you need to
 694                 if (needToRun == true)  {
 695                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 696                         lookup = input->getSharedRAbundVectors(lastLabel);
 697
 698                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 699                         process(lookup);
 700                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 701                 }
 702
 703                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 704
 705                 return 0;
 706         }
 707         catch(exception& e) {
 708                 m->errorOut(e, "TreeGroupCommand", "makeSimsShared");
 709                 exit(1);
 710         }
 711 }
 712
 713 /***********************************************************/
 714 int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 715         try{
 716                                 EstOutput data;
 717                                 vector<SharedRAbundVector*> subset;
 718                                 numGroups = thisLookup.size();
 719
 720                                 //for each calculator
 721                                 for(int i = 0 ; i < treeCalculators.size(); i++) {
 722                                         //initialize simMatrix
 723                                         simMatrix.clear();
 724                                         simMatrix.resize(numGroups);
 725                                         for (int k = 0; k < simMatrix.size(); k++)      {
 726                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 727                                                         simMatrix[k].push_back(0.0);
 728                                                 }
 729                                         }
 730
 731                                         //initialize index
 732                                         index.clear();
 733                                         for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 734
 735                                         //create a new filename
 736                                         outputFile = outputDir + m->getRootName(m->getSimpleName(inputfile)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".tre";
 737                                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 738
 739                                         for (int k = 0; k < thisLookup.size(); k++) {
 740                                                 for (int l = k; l < thisLookup.size(); l++) {
 741                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 742                                                                 //get estimated similarity between 2 groups
 743
 744                                                                 subset.clear(); //clear out old pair of sharedrabunds
 745                                                                 //add new pair of sharedrabunds
 746                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 747
 748                                                                 //if this calc needs all groups to calculate the pair load all groups
 749                                                                 if (treeCalculators[i]->getNeedsAll()) {
 750                                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 751                                                                         for (int w = 0; w < thisLookup.size(); w++) {
 752                                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 753                                                                         }
 754                                                                 }
 755
 756                                                                 data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 757                                                 //cout << thisLookup[k]->getGroup() << '\t' << thisLookup[l]->getGroup() << '\t' << (1.0 - data[0]) << endl;
 758                                                                 if (m->control_pressed) { return 1; }
 759
 760                                                                 //save values in similarity matrix
 761                                                                 simMatrix[k][l] = -(data[0]-1.0);
 762                                                                 simMatrix[l][k] = -(data[0]-1.0);
 763                                                         }
 764                                                 }
 765                                         }
 766
 767                                         //createdistance file from simMatrix
 768                                         /*string o = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".dist";
 769                                         ofstream outDist;
 770                                         m->openOutputFile(o, outDist);
 771                                         outDist << simMatrix.size() << endl;
 772                                         for (int k = 0; k < simMatrix.size(); k++) {
 773                                                 outDist << thisLookup[k]->getGroup() << '\t';
 774                                                 for (int l = 0; l < k; l++) {
 775                                                         outDist << (1.0-simMatrix[k][l]) << '\t';
 776                                                 }
 777                                                 outDist << endl;
 778                                         }
 779                                         outDist.close();*/
 780
 781
 782                                         if (m->control_pressed) { return 1; }
 783                                         //creates tree from similarity matrix and write out file
 784                                         createTree();
 785
 786                                         if (m->control_pressed) { return 1; }
 787                                 }
 788
 789                                 return 0;
 790
 791         }
 792         catch(exception& e) {
 793                 m->errorOut(e, "TreeGroupCommand", "process");
 794                 exit(1);
 795         }
 796 }
 797 /***********************************************************/
 798
 799
 800