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=numLeaves-1;i>=0;i--){
 313                         if(tree[i].pGroups.size() == 0){
 314                                 continue;
 315                         }
 316
 317                         int escape = 1;
 318                         int z;
 319
 320                         while(escape == 1){
 321                                 //get random index to switch with
 322                                 z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 323
 324                                 if(tree[z].pGroups.size() != 0){
 325                                         escape = 0;
 326                                 }
 327                         }
 328
 329                         //you only want to randomize the nodes that are from a group the user wants analyzed, so
 330                         //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.
 331                         bool treez, treei;
 332
 333                         //leaves have only one group so you can just set it to begin()
 334                         it = tree[z].pGroups.begin();
 335                         treez = inUsersGroups(it->first, globaldata->Groups);
 336
 337                         it = tree[i].pGroups.begin();
 338                         treei = inUsersGroups(it->first, globaldata->Groups);
 339
 340                         if ((treez == true) && (treei == true)) {
 341                                 //switches node i and node z's info.
 342                                 map<string,int> lib_hold = tree[z].pGroups;
 343                                 tree[z].pGroups = (tree[i].pGroups);
 344                                 tree[i].pGroups = (lib_hold);
 345
 346                                 tree[z].setGroup(tree[z].pGroups.begin()->first);
 347                                 tree[i].setGroup(tree[i].pGroups.begin()->first);
 348
 349                                 map<string,int> gcount_hold = tree[z].pcount;
 350                                 tree[z].pcount = (tree[i].pcount);
 351                                 tree[i].pcount = (gcount_hold);
 352                         }
 353                 }
 354         }
 355         catch(exception& e) {
 356                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 357                 exit(1);
 358         }
 359         catch(...) {
 360                 cout << "An unknown error has occurred in the Tree class function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 361                 exit(1);
 362         }
 363 }
 364 /**************************************************************************************************/
 365
 366 void Tree::randomBlengths()  {
 367         try {
 368                 for(int i=numNodes-1;i>=0;i--){
 369                         int z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 370
 371                         float bl_hold = tree[z].getBranchLength();
 372                         tree[z].setBranchLength(tree[i].getBranchLength());
 373                         tree[i].setBranchLength(bl_hold);
 374                 }
 375         }
 376         catch(exception& e) {
 377                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 378                 exit(1);
 379         }
 380         catch(...) {
 381                 cout << "An unknown error has occurred in the Tree class function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 382                 exit(1);
 383         }
 384 }
 385 /*************************************************************************************************/
 386 void Tree::assembleRandomUnifracTree() {
 387         randomLabels();
 388         assembleTree();
 389 }
 390
 391 /*************************************************************************************************/
 392 //for now it's just random topology but may become random labels as well later that why this is such a simple function now...
 393 void Tree::assembleRandomTree() {
 394         randomTopology();
 395         assembleTree();
 396 }
 397 /**************************************************************************************************/
 398
 399 void Tree::randomTopology() {
 400         try {
 401                 for(int i=0;i<numNodes;i++){
 402                         tree[i].setParent(-1);
 403                 }
 404                 for(int i=numLeaves;i<numNodes;i++){
 405                         tree[i].setChildren(-1, -1);
 406                 }
 407
 408                 for(int i=numLeaves;i<numNodes;i++){
 409                         int escape =0;
 410                         int rnd_index1, rnd_index2;
 411                         while(escape == 0){
 412                                 rnd_index1 = (int)(((double)rand() / (double) RAND_MAX)*i);
 413                                 if(tree[rnd_index1].getParent() == -1){escape = 1;}
 414                         }
 415
 416                         escape = 0;
 417                         while(escape == 0){
 418                                 rnd_index2 = (int)(((double)rand() / (double) RAND_MAX)*i);
 419                                 if(rnd_index2 != rnd_index1 && tree[rnd_index2].getParent() == -1){
 420                                         escape = 1;
 421                                 }
 422                         }
 423
 424                         tree[i].setChildren(rnd_index1,rnd_index2);
 425                         tree[i].setParent(-1);
 426                         tree[rnd_index1].setParent(i);
 427                         tree[rnd_index2].setParent(i);
 428                 }
 429         }
 430         catch(exception& e) {
 431                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 432                 exit(1);
 433         }
 434         catch(...) {
 435                 cout << "An unknown error has occurred in the Tree class function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 436                 exit(1);
 437         }
 438 }
 439
 440 /*****************************************************************/
 441 // This prints out the tree in Newick form.
 442 void Tree::createNewickFile() {
 443         try {
 444                 int root = findRoot();
 445                 filename = getRootName(globaldata->getTreeFile()) + "newick";
 446                 openOutputFile(filename, out);
 447
 448                 printBranch(root);
 449
 450                 // you are at the end of the tree
 451                 out << ";" << endl;
 452         }
 453         catch(exception& e) {
 454                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 455                 exit(1);
 456         }
 457         catch(...) {
 458                 cout << "An unknown error has occurred in the Tree class function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 459                 exit(1);
 460         }
 461 }
 462
 463 /*****************************************************************/
 464 //This function finds the index of the root node.
 465
 466 int Tree::findRoot() {
 467         try {
 468                 for (int i = 0; i < numNodes; i++) {
 469                         //you found the root
 470                         if (tree[i].getParent() == -1) { return i; }
 471                 }
 472                 return -1;
 473         }
 474         catch(exception& e) {
 475                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 476                 exit(1);
 477         }
 478         catch(...) {
 479                 cout << "An unknown error has occurred in the Tree class function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 480                 exit(1);
 481         }
 482 }
 483
 484 /*****************************************************************/
 485 void Tree::printBranch(int node) {
 486         try {
 487
 488                 // you are not a leaf
 489                 if (tree[node].getLChild() != -1) {
 490                         out << "(";
 491                         printBranch(tree[node].getLChild());
 492                         out << ",";
 493                         printBranch(tree[node].getRChild());
 494                         out << ")";
 495                 }else { //you are a leaf
 496                         tree[node].printNode();  //prints out name and branch length
 497                 }
 498
 499         }
 500         catch(exception& e) {
 501                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 502                 exit(1);
 503         }
 504         catch(...) {
 505                 cout << "An unknown error has occurred in the Tree class function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 506                 exit(1);
 507         }
 508 }
 509
 510 /*****************************************************************/
 511
 512 void Tree::setGroups() {
 513         try {
 514                 //if the user has not entered specific groups to analyze then do them all
 515                 if (globaldata->Groups.size() != 0) {
 516                         //check that groups are valid
 517                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 518                                 if (globaldata->gTreemap->isValidGroup(globaldata->Groups[i]) != true) {
 519                                         cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 520                                         // erase the invalid group from globaldata->Groups
 521                                         globaldata->Groups.erase (globaldata->Groups.begin()+i);
 522                                 }
 523                         }
 524
 525                         //if the user only entered invalid groups
 526                         if (globaldata->Groups.size() == 0) {
 527                                 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;
 528                                 for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 529                                         globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 530                                 }
 531                         }
 532
 533                 }else {
 534                         for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 535                                 globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 536                         }
 537                 }
 538         }
 539         catch(exception& e) {
 540                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 541                 exit(1);
 542         }
 543         catch(...) {
 544                 cout << "An unknown error has occurred in the Tree class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 545                 exit(1);
 546         }
 547
 548 }
 549
 550 /*****************************************************************/
 551
 552 void Tree::printTree() {
 553
 554         for(int i=0;i<numNodes;i++){
 555                 cout << i << '\t';
 556                 tree[i].printNode();
 557         }
 558
 559 }
 560
 561 /*****************************************************************/
 562
 563