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 "sharedjabund.h"
  12 #include "sharedsorabund.h"
  13 #include "sharedjclass.h"
  14 #include "sharedsorclass.h"
  15 #include "sharedjest.h"
  16 #include "sharedsorest.h"
  17 #include "sharedthetayc.h"
  18 #include "sharedthetan.h"
  19 #include "sharedmorisitahorn.h"
  20 #include "sharedbraycurtis.h"
  21
  22
  23 //**********************************************************************************************************************
  24
  25 TreeGroupCommand::TreeGroupCommand(string option){
  26         try {
  27                 globaldata = GlobalData::getInstance();
  28                 abort = false;
  29                 allLines = 1;
  30                 labels.clear();
  31                 Groups.clear();
  32                 Estimators.clear();
  33
  34                 //allow user to run help
  35                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  36
  37                 else {
  38                         //valid paramters for this command
  39                         string Array[] =  {"label","calc","groups", "phylip", "column", "name", "precision","cutoff","outputdir","inputdir"};
  40                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  41
  42                         OptionParser parser(option);
  43                         map<string, string> parameters = parser. getParameters();
  44
  45                         ValidParameters validParameter;
  46                         map<string, string>::iterator it;
  47
  48                         //check to make sure all parameters are valid for command
  49                         for (it = parameters.begin(); it != parameters.end(); it++) {
  50                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  51                         }
  52
  53                         //if the user changes the input directory command factory will send this info to us in the output parameter
  54                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  55                         if (inputDir == "not found"){   inputDir = "";          }
  56                         else {
  57                                 string path;
  58                                 it = parameters.find("phylip");
  59                                 //user has given a template file
  60                                 if(it != parameters.end()){
  61                                         path = hasPath(it->second);
  62                                         //if the user has not given a path then, add inputdir. else leave path alone.
  63                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  64                                 }
  65
  66                                 it = parameters.find("column");
  67                                 //user has given a template file
  68                                 if(it != parameters.end()){
  69                                         path = hasPath(it->second);
  70                                         //if the user has not given a path then, add inputdir. else leave path alone.
  71                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
  72                                 }
  73
  74                                 it = parameters.find("name");
  75                                 //user has given a template file
  76                                 if(it != parameters.end()){
  77                                         path = hasPath(it->second);
  78                                         //if the user has not given a path then, add inputdir. else leave path alone.
  79                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  80                                 }
  81                         }
  82
  83                         format = globaldata->getFormat();
  84
  85                         //required parameters
  86                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  87                         if (phylipfile == "not open") { abort = true; }
  88                         else if (phylipfile == "not found") { phylipfile = ""; }
  89                         else {  format = "phylip";  globaldata->setPhylipFile(phylipfile);      }
  90
  91                         columnfile = validParameter.validFile(parameters, "column", true);
  92                         if (columnfile == "not open") { abort = true; }
  93                         else if (columnfile == "not found") { columnfile = ""; }
  94                         else {  format = "column"; globaldata->setColumnFile(columnfile);       }
  95
  96                         namefile = validParameter.validFile(parameters, "name", true);
  97                         if (namefile == "not open") { abort = true; }
  98                         else if (namefile == "not found") { namefile = ""; }
  99                         else {  globaldata->setNameFile(namefile);      }
 100
 101                         //error checking on files
 102                         if ((globaldata->getSharedFile() == "") && ((phylipfile == "") && (columnfile == "")))  { mothurOut("You must run the read.otu command or provide a distance file before running the tree.shared command."); mothurOutEndLine(); abort = true; }
 103                         else if ((phylipfile != "") && (columnfile != "")) { mothurOut("When running the tree.shared command with a distance file you may not use both the column and the phylip parameters."); mothurOutEndLine(); abort = true; }
 104
 105                         if (columnfile != "") {
 106                                 if (namefile == "") {  mothurOut("You need to provide a namefile if you are going to use the column format."); mothurOutEndLine(); abort = true; }
 107                         }
 108
 109                         //check for optional parameter and set defaults
 110                         // ...at some point should added some additional type checking...
 111                         label = validParameter.validFile(parameters, "label", false);
 112                         if (label == "not found") { label = ""; }
 113                         else {
 114                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
 115                                 else { allLines = 1;  }
 116                         }
 117
 118                         //if the user has not specified any labels use the ones from read.otu
 119                         if(label == "") {
 120                                 allLines = globaldata->allLines;
 121                                 labels = globaldata->labels;
 122                         }
 123
 124                         groups = validParameter.validFile(parameters, "groups", false);
 125                         if (groups == "not found") { groups = ""; }
 126                         else {
 127                                 splitAtDash(groups, Groups);
 128                                 globaldata->Groups = Groups;
 129                         }
 130
 131                         calc = validParameter.validFile(parameters, "calc", false);
 132                         if (calc == "not found") { calc = "jclass-thetayc";  }
 133                         else {
 134                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 135                         }
 136                         splitAtDash(calc, Estimators);
 137
 138                         string temp;
 139                         temp = validParameter.validFile(parameters, "precision", false);                        if (temp == "not found") { temp = "100"; }
 140                         convert(temp, precision);
 141
 142                         temp = validParameter.validFile(parameters, "cutoff", false);                   if (temp == "not found") { temp = "10"; }
 143                         convert(temp, cutoff);
 144                         cutoff += (5 / (precision * 10.0));
 145
 146                         //if the user changes the output directory command factory will send this info to us in the output parameter
 147                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 148                                 outputDir = "";
 149                                 outputDir += hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 150                         }
 151
 152
 153                         if (abort == false) {
 154
 155                                 validCalculator = new ValidCalculators();
 156
 157                                 if (format == "sharedfile") {
 158                                         int i;
 159                                         for (i=0; i<Estimators.size(); i++) {
 160                                                 if (validCalculator->isValidCalculator("treegroup", Estimators[i]) == true) {
 161                                                         if (Estimators[i] == "jabund") {
 162                                                                 treeCalculators.push_back(new JAbund());
 163                                                         }else if (Estimators[i] == "sorabund") {
 164                                                                 treeCalculators.push_back(new SorAbund());
 165                                                         }else if (Estimators[i] == "jclass") {
 166                                                                 treeCalculators.push_back(new Jclass());
 167                                                         }else if (Estimators[i] == "sorclass") {
 168                                                                 treeCalculators.push_back(new SorClass());
 169                                                         }else if (Estimators[i] == "jest") {
 170                                                                 treeCalculators.push_back(new Jest());
 171                                                         }else if (Estimators[i] == "sorest") {
 172                                                                 treeCalculators.push_back(new SorEst());
 173                                                         }else if (Estimators[i] == "thetayc") {
 174                                                                 treeCalculators.push_back(new ThetaYC());
 175                                                         }else if (Estimators[i] == "thetan") {
 176                                                                 treeCalculators.push_back(new ThetaN());
 177                                                         }else if (Estimators[i] == "morisitahorn") {
 178                                                                 treeCalculators.push_back(new MorHorn());
 179                                                         }else if (Estimators[i] == "braycurtis") {
 180                                                                 treeCalculators.push_back(new BrayCurtis());
 181                                                         }
 182                                                 }
 183                                         }
 184                                 }
 185                         }
 186                 }
 187
 188         }
 189         catch(exception& e) {
 190                 errorOut(e, "TreeGroupCommand", "TreeGroupCommand");
 191                 exit(1);
 192         }
 193 }
 194
 195 //**********************************************************************************************************************
 196
 197 void TreeGroupCommand::help(){
 198         try {
 199                 mothurOut("The tree.shared command creates a .tre to represent the similiarity between groups or sequences.\n");
 200                 mothurOut("The tree.shared command can only be executed after a successful read.otu command or by providing a distance file.\n");
 201                 mothurOut("The tree.shared command parameters are groups, calc, phylip, column, name, cutoff, precision and label.\n");
 202                 mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n");
 203                 mothurOut("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");
 204                 mothurOut("The phylip or column parameter are required if you do not run the read.otu command first, and only one may be used.  If you use a column file the name filename is required. \n");
 205                 mothurOut("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");
 206                 mothurOut("The tree.shared command should be in the following format: tree.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n");
 207                 mothurOut("Example tree.shared(groups=A-B-C, calc=jabund-sorabund).\n");
 208                 mothurOut("The default value for groups is all the groups in your groupfile.\n");
 209                 mothurOut("The default value for calc is jclass-thetayc.\n");
 210                 mothurOut("The tree.shared command outputs a .tre file for each calculator you specify at each distance you choose.\n");
 211                 validCalculator->printCalc("treegroup", cout);
 212                 mothurOut("Or the tree.shared command can be in the following format: tree.shared(phylip=yourPhylipFile).\n");
 213                 mothurOut("Example tree.shared(phylip=abrecovery.dist).\n");
 214                 mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 215         }
 216         catch(exception& e) {
 217                 errorOut(e, "TreeGroupCommand", "help");
 218                 exit(1);
 219         }
 220 }
 221
 222
 223 //**********************************************************************************************************************
 224
 225 TreeGroupCommand::~TreeGroupCommand(){
 226         if (abort == false) {
 227
 228                 if (format == "sharedfile") { delete read;  delete input; globaldata->ginput = NULL; }
 229                 else { delete readMatrix;  delete matrix; delete list; }
 230                 delete tmap;  globaldata->gTreemap = NULL;
 231                 delete validCalculator;
 232         }
 233
 234 }
 235
 236 //**********************************************************************************************************************
 237
 238 int TreeGroupCommand::execute(){
 239         try {
 240
 241                 if (abort == true) { return 0; }
 242
 243                 if (format == "sharedfile") {
 244                         //if the users entered no valid calculators don't execute command
 245                         if (treeCalculators.size() == 0) { mothurOut("You have given no valid calculators."); mothurOutEndLine(); return 0; }
 246
 247                         //you have groups
 248                         read = new ReadOTUFile(globaldata->inputFileName);
 249                         read->read(&*globaldata);
 250
 251                         input = globaldata->ginput;
 252                         lookup = input->getSharedRAbundVectors();
 253                         lastLabel = lookup[0]->getLabel();
 254
 255                         if (lookup.size() < 2) { mothurOut("You have not provided enough valid groups.  I cannot run the command."); mothurOutEndLine(); return 0; }
 256
 257                         //used in tree constructor
 258                         globaldata->runParse = false;
 259
 260                         //create tree file
 261                         makeSimsShared();
 262                 }else{
 263                         //read in dist file
 264                         filename = globaldata->inputFileName;
 265
 266                         if (format == "column") { readMatrix = new ReadColumnMatrix(filename); }
 267                         else if (format == "phylip") { readMatrix = new ReadPhylipMatrix(filename); }
 268
 269                         readMatrix->setCutoff(cutoff);
 270
 271                         if(namefile != ""){
 272                                 nameMap = new NameAssignment(namefile);
 273                                 nameMap->readMap();
 274                         }
 275                         else{
 276                                 nameMap = NULL;
 277                         }
 278
 279                         readMatrix->read(nameMap);
 280                         list = readMatrix->getListVector();
 281                         matrix = readMatrix->getMatrix();
 282
 283                         //make treemap
 284                         tmap = new TreeMap();
 285                         tmap->makeSim(list);
 286                         globaldata->gTreemap = tmap;
 287
 288                         globaldata->Groups = tmap->namesOfGroups;
 289
 290                         //clear globaldatas old tree names if any
 291                         globaldata->Treenames.clear();
 292
 293                         //fills globaldatas tree names
 294                         globaldata->Treenames = globaldata->Groups;
 295
 296                         //used in tree constructor
 297                         globaldata->runParse = false;
 298
 299                         makeSimsDist();
 300
 301                         //create a new filename
 302                         outputFile = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + "tre";
 303
 304                         createTree();
 305                         mothurOut("Tree complete. "); mothurOutEndLine();
 306                 }
 307
 308                 //reset groups parameter
 309                 globaldata->Groups.clear();
 310
 311                 return 0;
 312         }
 313         catch(exception& e) {
 314                 errorOut(e, "TreeGroupCommand", "execute");
 315                 exit(1);
 316         }
 317 }
 318 //**********************************************************************************************************************
 319
 320 void TreeGroupCommand::createTree(){
 321         try {
 322                 //create tree
 323                 t = new Tree();
 324
 325                 //do merges and create tree structure by setting parents and children
 326                 //there are numGroups - 1 merges to do
 327                 for (int i = 0; i < (numGroups - 1); i++) {
 328                         float largest = -1000.0;
 329
 330                         int row, column;
 331                         //find largest value in sims matrix by searching lower triangle
 332                         for (int j = 1; j < simMatrix.size(); j++) {
 333                                 for (int k = 0; k < j; k++) {
 334                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 335                                 }
 336                         }
 337
 338                         //set non-leaf node info and update leaves to know their parents
 339                         //non-leaf
 340                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 341
 342                         //parents
 343                         t->tree[index[row]].setParent(numGroups + i);
 344                         t->tree[index[column]].setParent(numGroups + i);
 345
 346                         //blength = distance / 2;
 347                         float blength = ((1.0 - largest) / 2);
 348
 349                         //branchlengths
 350                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 351                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 352
 353                         //set your length to leaves to your childs length plus branchlength
 354                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 355
 356
 357                         //update index
 358                         index[row] = numGroups+i;
 359                         index[column] = numGroups+i;
 360
 361                         //remove highest value that caused the merge.
 362                         simMatrix[row][column] = -1000.0;
 363                         simMatrix[column][row] = -1000.0;
 364
 365                         //merge values in simsMatrix
 366                         for (int n = 0; n < simMatrix.size(); n++)      {
 367                                 //row becomes merge of 2 groups
 368                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 369                                 simMatrix[n][row] = simMatrix[row][n];
 370                                 //delete column
 371                                 simMatrix[column][n] = -1000.0;
 372                                 simMatrix[n][column] = -1000.0;
 373                         }
 374                 }
 375
 376                 //adjust tree to make sure root to tip length is .5
 377                 int root = t->findRoot();
 378                 t->tree[root].setBranchLength((0.5 - t->tree[root].getLengthToLeaves()));
 379
 380                 //assemble tree
 381                 t->assembleTree();
 382
 383                 //print newick file
 384                 t->createNewickFile(outputFile);
 385
 386                 //delete tree
 387                 delete t;
 388
 389         }
 390         catch(exception& e) {
 391                 errorOut(e, "TreeGroupCommand", "createTree");
 392                 exit(1);
 393         }
 394 }
 395 /***********************************************************/
 396 void TreeGroupCommand::printSims(ostream& out) {
 397         try {
 398
 399                 //output column headers
 400                 //out << '\t';
 401                 //for (int i = 0; i < lookup.size(); i++) {     out << lookup[i]->getGroup() << '\t';           }
 402                 //out << endl;
 403
 404
 405                 for (int m = 0; m < simMatrix.size(); m++)      {
 406                         //out << lookup[m]->getGroup() << '\t';
 407                         for (int n = 0; n < simMatrix.size(); n++)      {
 408                                 out << simMatrix[m][n] << '\t';
 409                         }
 410                         out << endl;
 411                 }
 412
 413         }
 414         catch(exception& e) {
 415                 errorOut(e, "TreeGroupCommand", "printSims");
 416                 exit(1);
 417         }
 418 }
 419 /***********************************************************/
 420 void TreeGroupCommand::makeSimsDist() {
 421         try {
 422                 numGroups = list->size();
 423
 424                 //initialize index
 425                 index.clear();
 426                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 427
 428                 //initialize simMatrix
 429                 simMatrix.clear();
 430                 simMatrix.resize(numGroups);
 431                 for (int m = 0; m < simMatrix.size(); m++)      {
 432                         for (int j = 0; j < simMatrix.size(); j++)      {
 433                                 simMatrix[m].push_back(0.0);
 434                         }
 435                 }
 436
 437                 //go through sparse matrix and fill sims
 438                 //go through each cell in the sparsematrix
 439                 for(MatData currentCell = matrix->begin(); currentCell != matrix->end(); currentCell++){
 440                         //similairity = -(distance-1)
 441                         simMatrix[currentCell->row][currentCell->column] = -(currentCell->dist -1.0);
 442                         simMatrix[currentCell->column][currentCell->row] = -(currentCell->dist -1.0);
 443                 }
 444
 445
 446         }
 447         catch(exception& e) {
 448                 errorOut(e, "TreeGroupCommand", "makeSimsDist");
 449                 exit(1);
 450         }
 451 }
 452
 453 /***********************************************************/
 454 void TreeGroupCommand::makeSimsShared() {
 455         try {
 456
 457                 //clear globaldatas old tree names if any
 458                 globaldata->Treenames.clear();
 459
 460                 //fills globaldatas tree names
 461                 globaldata->Treenames = globaldata->Groups;
 462
 463                 //create treemap class from groupmap for tree class to use
 464                 tmap = new TreeMap();
 465                 tmap->makeSim(globaldata->gGroupmap);
 466                 globaldata->gTreemap = tmap;
 467
 468                 set<string> processedLabels;
 469                 set<string> userLabels = labels;
 470
 471                 //as long as you are not at the end of the file or done wih the lines you want
 472                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 473
 474                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 475                                 mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 476                                 process(lookup);
 477
 478                                 processedLabels.insert(lookup[0]->getLabel());
 479                                 userLabels.erase(lookup[0]->getLabel());
 480                         }
 481
 482                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 483                                 string saveLabel = lookup[0]->getLabel();
 484
 485                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 486                                 lookup = input->getSharedRAbundVectors(lastLabel);
 487
 488                                 mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 489                                 process(lookup);
 490
 491                                 processedLabels.insert(lookup[0]->getLabel());
 492                                 userLabels.erase(lookup[0]->getLabel());
 493
 494                                 //restore real lastlabel to save below
 495                                 lookup[0]->setLabel(saveLabel);
 496                         }
 497
 498                         lastLabel = lookup[0]->getLabel();
 499
 500                         //get next line to process
 501                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 502                         lookup = input->getSharedRAbundVectors();
 503                 }
 504
 505                 //output error messages about any remaining user labels
 506                 set<string>::iterator it;
 507                 bool needToRun = false;
 508                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 509                         mothurOut("Your file does not include the label " + *it);
 510                         if (processedLabels.count(lastLabel) != 1) {
 511                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 512                                 needToRun = true;
 513                         }else {
 514                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 515                         }
 516                 }
 517
 518                 //run last label if you need to
 519                 if (needToRun == true)  {
 520                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {             delete lookup[i]; }             }
 521                         lookup = input->getSharedRAbundVectors(lastLabel);
 522
 523                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 524                         process(lookup);
 525                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 526                 }
 527
 528                 for(int i = 0 ; i < treeCalculators.size(); i++) {  delete treeCalculators[i]; }
 529         }
 530         catch(exception& e) {
 531                 errorOut(e, "TreeGroupCommand", "makeSimsShared");
 532                 exit(1);
 533         }
 534 }
 535
 536 /***********************************************************/
 537 void TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
 538         try{
 539                                 EstOutput data;
 540                                 vector<SharedRAbundVector*> subset;
 541                                 numGroups = thisLookup.size();
 542
 543                                 //for each calculator
 544                                 for(int i = 0 ; i < treeCalculators.size(); i++) {
 545                                         //initialize simMatrix
 546                                         simMatrix.clear();
 547                                         simMatrix.resize(numGroups);
 548                                         for (int m = 0; m < simMatrix.size(); m++)      {
 549                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 550                                                         simMatrix[m].push_back(0.0);
 551                                                 }
 552                                         }
 553
 554                                         //initialize index
 555                                         index.clear();
 556                                         for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 557
 558                                         //create a new filename
 559                                         outputFile = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + treeCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".tre";
 560
 561                                         for (int k = 0; k < thisLookup.size(); k++) {
 562                                                 for (int l = k; l < thisLookup.size(); l++) {
 563                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 564                                                                 //get estimated similarity between 2 groups
 565
 566                                                                 subset.clear(); //clear out old pair of sharedrabunds
 567                                                                 //add new pair of sharedrabunds
 568                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 569
 570                                                                 data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 571                                                                 //save values in similarity matrix
 572                                                                 simMatrix[k][l] = data[0];
 573                                                                 simMatrix[l][k] = data[0];
 574                                                         }
 575                                                 }
 576                                         }
 577
 578                                         //creates tree from similarity matrix and write out file
 579                                         createTree();
 580                                 }
 581
 582         }
 583         catch(exception& e) {
 584                 errorOut(e, "TreeGroupCommand", "process");
 585                 exit(1);
 586         }
 587 }
 588 /***********************************************************/
 589
 590
 591