tree.cpp

   1 /*
   2  *  tree.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/22/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "tree.h"
  11
  12
  13 /*****************************************************************/
  14 Tree::Tree() {
  15         try {
  16
  17                 globaldata = GlobalData::getInstance();
  18                 numLeaves = globaldata->gTreemap->getNumSeqs();
  19                 numNodes = 2*numLeaves - 1;
  20
  21                 tree.resize(numNodes);
  22
  23                 //initialize tree with correct number of nodes, name and group info.
  24                 for (int i = 0; i < numNodes; i++) {
  25                         //initialize leaf nodes
  26                         if (i <= (numLeaves-1)) {
  27                                 tree[i].setName(globaldata->gTreemap->namesOfSeqs[i]);
  28                                 tree[i].setGroup(globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i]));
  29                                 //the node knows its index
  30                                 tree[i].setIndex(i);
  31                                 //set pcount and pGroup for groupname to 1.
  32                                 tree[i].pcount[globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i])] = 1;
  33                                 tree[i].pGroups[globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i])] = 1;
  34                                 //Treemap knows name, group and index to speed up search
  35                                 globaldata->gTreemap->setIndex(globaldata->gTreemap->namesOfSeqs[i], i);
  36
  37                         //intialize non leaf nodes
  38                         }else if (i > (numLeaves-1)) {
  39                                 tree[i].setName("");
  40                                 tree[i].setGroup("");
  41                                 //the node knows its index
  42                                 tree[i].setIndex(i);
  43                         }
  44                 }
  45         }
  46         catch(exception& e) {
  47                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function Tree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  48                 exit(1);
  49         }
  50         catch(...) {
  51                 cout << "An unknown error has occurred in the Tree class function Tree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  52                 exit(1);
  53         }
  54 }
  55
  56 /*****************************************************************/
  57
  58 int Tree::getIndex(string searchName) {
  59         try {
  60                 //Treemap knows name, group and index to speed up search
  61                 // getIndex function will return the vector index or -1 if seq is not found.
  62                 int index = globaldata->gTreemap->getIndex(searchName);
  63                 return index;
  64
  65         }
  66         catch(exception& e) {
  67                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function getIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  68                 exit(1);
  69         }
  70         catch(...) {
  71                 cout << "An unknown error has occurred in the Tree class function getIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  72                 exit(1);
  73         }
  74 }
  75 /*****************************************************************/
  76
  77 void Tree::setIndex(string searchName, int index) {
  78         try {
  79                 //set index in treemap
  80                 globaldata->gTreemap->setIndex(searchName, index);
  81         }
  82         catch(exception& e) {
  83                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function setIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  84                 exit(1);
  85         }
  86         catch(...) {
  87                 cout << "An unknown error has occurred in the Tree class function setIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  88                 exit(1);
  89         }
  90 }
  91 /*****************************************************************/
  92 void Tree::assembleTree() {
  93         try {
  94                 //build the pGroups in non leaf nodes to be used in the parsimony calcs.
  95                 for (int i = numLeaves; i < numNodes; i++) {
  96                         tree[i].pGroups = (mergeGroups(i));
  97                         tree[i].pcount = (mergeGcounts(i));
  98                 }
  99         }
 100         catch(exception& e) {
 101                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function assembleTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 102                 exit(1);
 103         }
 104         catch(...) {
 105                 cout << "An unknown error has occurred in the Tree class function assembleTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 106                 exit(1);
 107         }
 108 }
 109 /*****************************************************************/
 110 void Tree::getCopy(Tree* copy) {
 111         try {
 112
 113                 //for each node in the tree copy its info
 114                 for (int i = 0; i < numNodes; i++) {
 115                         //copy name
 116                         tree[i].setName(copy->tree[i].getName());
 117
 118                         //copy group
 119                         tree[i].setGroup(copy->tree[i].getGroup());
 120
 121                         //copy branch length
 122                         tree[i].setBranchLength(copy->tree[i].getBranchLength());
 123
 124                         //copy parent
 125                         tree[i].setParent(copy->tree[i].getParent());
 126
 127                         //copy children
 128                         tree[i].setChildren(copy->tree[i].getLChild(), copy->tree[i].getRChild());
 129
 130                         //copy index in node and tmap
 131                         tree[i].setIndex(copy->tree[i].getIndex());
 132                         setIndex(copy->tree[i].getName(), getIndex(copy->tree[i].getName()));
 133
 134                         //copy pGroups
 135                         tree[i].pGroups = copy->tree[i].pGroups;
 136
 137                         //copy pcount
 138                         tree[i].pcount = copy->tree[i].pcount;
 139                 }
 140         }
 141         catch(exception& e) {
 142                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function getCopy. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 143                 exit(1);
 144         }
 145         catch(...) {
 146                 cout << "An unknown error has occurred in the Tree class function getCopy. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 147                 exit(1);
 148         }
 149 }
 150 /*****************************************************************/
 151 //returns a map with a groupname and the number of times that group was seen in the children
 152 //for instance if your children are white and black then it would return a map with 2 entries
 153 // p[white] = 1 and p[black] = 1.  Now go up a level and merge that with a node who has p[white] = 1
 154 //and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
 155
 156 map<string, int> Tree::mergeGroups(int i) {
 157         try {
 158                 int lc = tree[i].getLChild();
 159                 int rc = tree[i].getRChild();
 160
 161                 //set parsimony groups to left child
 162                 map<string,int> parsimony = tree[lc].pGroups;
 163
 164                 int maxPars = 1;
 165
 166                 //look at right child groups and update maxPars if right child has something higher for that group.
 167                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
 168                         it2 = parsimony.find(it->first);
 169                         if (it2 != parsimony.end()) {
 170                                 parsimony[it->first]++;
 171                         }else {
 172                                 parsimony[it->first] = 1;
 173                         }
 174
 175                         if(parsimony[it->first] > maxPars){
 176                                 maxPars = parsimony[it->first];
 177                         }
 178                 }
 179
 180                 // this is true if right child had a greater parsimony for a certain group
 181                 if(maxPars > 1){
 182                         //erase all the groups that are only 1 because you found something with 2.
 183                         for(it=parsimony.begin();it!=parsimony.end();it++){
 184                                 if(it->second == 1){
 185                                         parsimony.erase(it->first);
 186                                         it--;
 187                                 }
 188                         }
 189                         //set one remaining groups to 1
 190                         //so with our above example p[white] = 2 would be left and it would become p[white] = 1
 191                         for(it=parsimony.begin();it!=parsimony.end();it++){
 192                                 parsimony[it->first] = 1;
 193                         }
 194
 195                 }
 196
 197                 return parsimony;
 198         }
 199         catch(exception& e) {
 200                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 201                 exit(1);
 202         }
 203         catch(...) {
 204                 cout << "An unknown error has occurred in the Tree class function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 205                 exit(1);
 206         }
 207 }
 208 /*****************************************************************/
 209 //returns a map with a groupname and the number of times that group was seen in the children
 210 //for instance if your children are white and black then it would return a map with 2 entries
 211 // p[white] = 1 and p[black] = 1.  Now go up a level and merge that with a node who has p[white] = 1
 212 //and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
 213
 214 map<string, int> Tree::mergeUserGroups(int i) {
 215         try {
 216
 217                 int lc = tree[i].getLChild();
 218                 int rc = tree[i].getRChild();
 219
 220                 //loop through nodes groups removing the ones the user doesn't want
 221                 for (it = tree[lc].pGroups.begin(); it != tree[lc].pGroups.end(); it++) {
 222                         if (inUsersGroups(it->first, globaldata->Groups) != true) { tree[lc].pGroups.erase(it->first); }
 223                 }
 224
 225                 //loop through nodes groups removing the ones the user doesn't want
 226                 for (it = tree[rc].pGroups.begin(); it != tree[rc].pGroups.end(); it++) {
 227                         if (inUsersGroups(it->first, globaldata->Groups) != true) { tree[rc].pGroups.erase(it->first); }
 228                 }
 229
 230                 //set parsimony groups to left child
 231                 map<string,int> parsimony = tree[lc].pGroups;
 232
 233                 int maxPars = 1;
 234
 235                 //look at right child groups and update maxPars if right child has something higher for that group.
 236                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
 237                         it2 = parsimony.find(it->first);
 238                         if (it2 != parsimony.end()) {
 239                                 parsimony[it->first]++;
 240                         }else {
 241                                 parsimony[it->first] = 1;
 242                         }
 243
 244                         if(parsimony[it->first] > maxPars){
 245                                 maxPars = parsimony[it->first];
 246                         }
 247                 }
 248
 249                 // this is true if right child had a greater parsimony for a certain group
 250                 if(maxPars > 1){
 251                         //erase all the groups that are only 1 because you found something with 2.
 252                         for(it=parsimony.begin();it!=parsimony.end();it++){
 253                                 if(it->second == 1){
 254                                         parsimony.erase(it->first);
 255                                         it--;
 256                                 }
 257                         }
 258                         //set one remaining groups to 1
 259                         //so with our above example p[white] = 2 would be left and it would become p[white] = 1
 260                         for(it=parsimony.begin();it!=parsimony.end();it++){
 261                                 parsimony[it->first] = 1;
 262                         }
 263
 264                 }
 265
 266                 return parsimony;
 267         }
 268         catch(exception& e) {
 269                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 270                 exit(1);
 271         }
 272         catch(...) {
 273                 cout << "An unknown error has occurred in the Tree class function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 274                 exit(1);
 275         }
 276 }
 277
 278
 279 /**************************************************************************************************/
 280
 281 map<string,int> Tree::mergeGcounts(int position) {
 282         try{
 283                 map<string,int>::iterator pos;
 284
 285                 int lc = tree[position].getLChild();
 286                 int rc = tree[position].getRChild();
 287
 288                 map<string,int> sum = tree[lc].pcount;
 289
 290                 for(it=tree[rc].pcount.begin();it!=tree[rc].pcount.end();it++){
 291                         sum[it->first] += it->second;
 292                 }
 293                 return sum;
 294         }
 295         catch(exception& e) {
 296                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGcounts. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 297                 exit(1);
 298         }
 299         catch(...) {
 300                 cout << "An unknown error has occurred in the Tree class function mergeGcounts. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 301                 exit(1);
 302         }
 303 }
 304 /**************************************************************************************************/
 305
 306 void Tree::randomLabels() {
 307         try {
 308
 309                 //set up the groups the user wants to include
 310                 setGroups();
 311
 312                 for(int i = 0; i < numLeaves; i++){
 313                         int z;
 314                         //get random index to switch with
 315                         z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 316
 317                         //you only want to randomize the nodes that are from a group the user wants analyzed, so
 318                         //if either of the leaf nodes you are about to switch are not in the users groups then you don't want to switch them.
 319                         bool treez, treei;
 320
 321                         treez = inUsersGroups(tree[z].getGroup(), globaldata->Groups);
 322                         treei = inUsersGroups(tree[i].getGroup(), globaldata->Groups);
 323
 324                         if ((treez == true) && (treei == true)) {
 325                                 //switches node i and node z's info.
 326                                 map<string,int> lib_hold = tree[z].pGroups;
 327                                 tree[z].pGroups = (tree[i].pGroups);
 328                                 tree[i].pGroups = (lib_hold);
 329
 330                                 string zgroup = tree[z].getGroup();
 331                                 tree[z].setGroup(tree[i].getGroup());
 332                                 tree[i].setGroup(zgroup);
 333
 334                                 string zname = tree[z].getName();
 335                                 tree[z].setName(tree[i].getName());
 336                                 tree[i].setName(zname);
 337
 338                                 map<string,int> gcount_hold = tree[z].pcount;
 339                                 tree[z].pcount = (tree[i].pcount);
 340                                 tree[i].pcount = (gcount_hold);
 341                         }
 342                 }
 343         }
 344         catch(exception& e) {
 345                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 346                 exit(1);
 347         }
 348         catch(...) {
 349                 cout << "An unknown error has occurred in the Tree class function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 350                 exit(1);
 351         }
 352 }
 353 /**************************************************************************************************/
 354
 355 void Tree::randomLabels(string groupA, string groupB) {
 356         try {
 357                 int numSeqsA = globaldata->gTreemap->seqsPerGroup[groupA];
 358                 int numSeqsB = globaldata->gTreemap->seqsPerGroup[groupB];
 359
 360                 vector<string> randomGroups(numSeqsA+numSeqsB, groupA);
 361                 for(int i=numSeqsA;i<randomGroups.size();i++){
 362                         randomGroups[i] = groupB;
 363                 }
 364                 random_shuffle(randomGroups.begin(), randomGroups.end());
 365
 366                 int randomCounter = 0;
 367                 for(int i=0;i<numLeaves;i++){
 368                         if(tree[i].getGroup() == groupA || tree[i].getGroup() == groupB){
 369                                 tree[i].setGroup(randomGroups[randomCounter]);
 370                                 tree[i].pcount.clear();
 371                                 tree[i].pcount[randomGroups[randomCounter]] = 1;
 372                                 tree[i].pGroups.clear();
 373                                 tree[i].pGroups[randomGroups[randomCounter]] = 1;
 374                                 randomCounter++;
 375                         }
 376                 }
 377         }
 378         catch(exception& e) {
 379                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 380                 exit(1);
 381         }
 382         catch(...) {
 383                 cout << "An unknown error has occurred in the Tree class function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 384                 exit(1);
 385         }
 386 }
 387 /**************************************************************************************************/
 388 void Tree::randomBlengths()  {
 389         try {
 390                 for(int i=numNodes-1;i>=0;i--){
 391                         int z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 392
 393                         float bl_hold = tree[z].getBranchLength();
 394                         tree[z].setBranchLength(tree[i].getBranchLength());
 395                         tree[i].setBranchLength(bl_hold);
 396                 }
 397         }
 398         catch(exception& e) {
 399                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 400                 exit(1);
 401         }
 402         catch(...) {
 403                 cout << "An unknown error has occurred in the Tree class function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 404                 exit(1);
 405         }
 406 }
 407 /*************************************************************************************************/
 408 void Tree::assembleRandomUnifracTree() {
 409         randomLabels();
 410         assembleTree();
 411 }
 412 /*************************************************************************************************/
 413 void Tree::assembleRandomUnifracTree(string groupA, string groupB) {
 414         randomLabels(groupA, groupB);
 415         assembleTree();
 416 }
 417
 418 /*************************************************************************************************/
 419 //for now it's just random topology but may become random labels as well later that why this is such a simple function now...
 420 void Tree::assembleRandomTree() {
 421         randomTopology();
 422         assembleTree();
 423 }
 424 /**************************************************************************************************/
 425
 426 void Tree::randomTopology() {
 427         try {
 428                 for(int i=0;i<numNodes;i++){
 429                         tree[i].setParent(-1);
 430                 }
 431                 for(int i=numLeaves;i<numNodes;i++){
 432                         tree[i].setChildren(-1, -1);
 433                 }
 434
 435                 for(int i=numLeaves;i<numNodes;i++){
 436                         int escape =0;
 437                         int rnd_index1, rnd_index2;
 438                         while(escape == 0){
 439                                 rnd_index1 = (int)(((double)rand() / (double) RAND_MAX)*i);
 440                                 if(tree[rnd_index1].getParent() == -1){escape = 1;}
 441                         }
 442
 443                         escape = 0;
 444                         while(escape == 0){
 445                                 rnd_index2 = (int)(((double)rand() / (double) RAND_MAX)*i);
 446                                 if(rnd_index2 != rnd_index1 && tree[rnd_index2].getParent() == -1){
 447                                         escape = 1;
 448                                 }
 449                         }
 450
 451                         tree[i].setChildren(rnd_index1,rnd_index2);
 452                         tree[i].setParent(-1);
 453                         tree[rnd_index1].setParent(i);
 454                         tree[rnd_index2].setParent(i);
 455                 }
 456         }
 457         catch(exception& e) {
 458                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 459                 exit(1);
 460         }
 461         catch(...) {
 462                 cout << "An unknown error has occurred in the Tree class function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 463                 exit(1);
 464         }
 465 }
 466
 467 /*****************************************************************/
 468 // This prints out the tree in Newick form.
 469 void Tree::createNewickFile(string f) {
 470         try {
 471                 int root = findRoot();
 472                 //filename = getRootName(globaldata->getTreeFile()) + "newick";
 473                 filename = f;
 474                 openOutputFile(filename, out);
 475
 476                 printBranch(root);
 477
 478                 // you are at the end of the tree
 479                 out << ";" << endl;
 480                 out.close();
 481         }
 482         catch(exception& e) {
 483                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 484                 exit(1);
 485         }
 486         catch(...) {
 487                 cout << "An unknown error has occurred in the Tree class function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 488                 exit(1);
 489         }
 490 }
 491
 492 /*****************************************************************/
 493 //This function finds the index of the root node.
 494
 495 int Tree::findRoot() {
 496         try {
 497                 for (int i = 0; i < numNodes; i++) {
 498                         //you found the root
 499                         if (tree[i].getParent() == -1) { return i; }
 500                 }
 501                 return -1;
 502         }
 503         catch(exception& e) {
 504                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 505                 exit(1);
 506         }
 507         catch(...) {
 508                 cout << "An unknown error has occurred in the Tree class function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 509                 exit(1);
 510         }
 511 }
 512
 513 /*****************************************************************/
 514 void Tree::printBranch(int node) {
 515         try {
 516
 517                 // you are not a leaf
 518                 if (tree[node].getLChild() != -1) {
 519                         out << "(";
 520                         printBranch(tree[node].getLChild());
 521                         out << ",";
 522                         printBranch(tree[node].getRChild());
 523                         out << ")";
 524                 }else { //you are a leaf
 525                         out << tree[node].getGroup() << ":" << tree[node].getBranchLength();
 526                 }
 527
 528         }
 529         catch(exception& e) {
 530                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 531                 exit(1);
 532         }
 533         catch(...) {
 534                 cout << "An unknown error has occurred in the Tree class function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 535                 exit(1);
 536         }
 537 }
 538
 539 /*****************************************************************/
 540
 541 void Tree::setGroups() {
 542         try {
 543                 //if the user has not entered specific groups to analyze then do them all
 544                 if (globaldata->Groups.size() != 0) {
 545                         //check that groups are valid
 546                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 547                                 if (globaldata->gTreemap->isValidGroup(globaldata->Groups[i]) != true) {
 548                                         cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 549                                         // erase the invalid group from globaldata->Groups
 550                                         globaldata->Groups.erase (globaldata->Groups.begin()+i);
 551                                 }
 552                         }
 553
 554                         //if the user only entered invalid groups
 555                         if (globaldata->Groups.size() == 0) {
 556                                 cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
 557                                 for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 558                                         globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 559                                 }
 560                         }
 561
 562                 }else {
 563                         for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 564                                 globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 565                         }
 566                 }
 567         }
 568         catch(exception& e) {
 569                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 570                 exit(1);
 571         }
 572         catch(...) {
 573                 cout << "An unknown error has occurred in the Tree class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 574                 exit(1);
 575         }
 576
 577 }
 578
 579 /*****************************************************************/
 580
 581 void Tree::printTree() {
 582
 583         for(int i=0;i<numNodes;i++){
 584                 cout << i << '\t';
 585                 tree[i].printNode();
 586         }
 587
 588 }
 589
 590 /*****************************************************************/
 591
 592