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
 159                 int lc = tree[i].getLChild();
 160                 int rc = tree[i].getRChild();
 161
 162                 //set parsimony groups to left child
 163                 map<string,int> parsimony = tree[lc].pGroups;
 164
 165                 int maxPars = 1;
 166
 167                 //look at right child groups and update maxPars if right child has something higher for that group.
 168                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
 169                         parsimony[it->first]++;
 170
 171                         if(parsimony[it->first] > maxPars){
 172                                 maxPars = parsimony[it->first];
 173                         }
 174                 }
 175
 176                 // this is true if right child had a greater parsimony for a certain group
 177                 if(maxPars > 1){
 178                         //erase all the groups that are only 1 because you found something with 2.
 179                         for(it=parsimony.begin();it!=parsimony.end();it++){
 180                                 if(it->second == 1){
 181                                         parsimony.erase(it->first);
 182                                         it--;
 183                                 }
 184                         }
 185                         //set one remaining groups to 1
 186                         //so with our above example p[white] = 2 would be left and it would become p[white] = 1
 187                         for(it=parsimony.begin();it!=parsimony.end();it++){
 188                                 parsimony[it->first] = 1;
 189                         }
 190
 191                 }
 192
 193                 return parsimony;
 194         }
 195         catch(exception& e) {
 196                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 197                 exit(1);
 198         }
 199         catch(...) {
 200                 cout << "An unknown error has occurred in the Tree class function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 201                 exit(1);
 202         }
 203 }
 204
 205 /**************************************************************************************************/
 206
 207 map<string,int> Tree::mergeGcounts(int position) {
 208         try{
 209                 map<string,int>::iterator pos;
 210
 211                 int lc = tree[position].getLChild();
 212                 int rc = tree[position].getRChild();
 213
 214                 map<string,int> sum = tree[lc].pcount;
 215
 216                 for(it=tree[rc].pcount.begin();it!=tree[rc].pcount.end();it++){
 217                         sum[it->first] += it->second;
 218                 }
 219                 return sum;
 220         }
 221         catch(exception& e) {
 222                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGcounts. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 223                 exit(1);
 224         }
 225         catch(...) {
 226                 cout << "An unknown error has occurred in the Tree class function mergeGcounts. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 227                 exit(1);
 228         }
 229 }
 230 /**************************************************************************************************/
 231
 232 void Tree::randomLabels() {
 233         try {
 234
 235                 //set up the groups the user wants to include
 236                 setGroups();
 237
 238                 for(int i=numLeaves-1;i>=0;i--){
 239                         if(tree[i].pGroups.size() == 0){
 240                                 continue;
 241                         }
 242
 243                         int escape = 1;
 244                         int z;
 245
 246                         while(escape == 1){
 247                                 //get random index to switch with
 248                                 z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 249
 250                                 if(tree[z].pGroups.size() != 0){
 251                                         escape = 0;
 252                                 }
 253                         }
 254
 255                         //you only want to randomize the nodes that are from a group the user wants analyzed, so
 256                         //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.
 257                         bool treez, treei;
 258
 259                         //leaves have only one group so you can just set it to begin()
 260                         it = tree[z].pGroups.begin();
 261                         treez = inUsersGroups(it->first, globaldata->Groups);
 262
 263                         it = tree[i].pGroups.begin();
 264                         treei = inUsersGroups(it->first, globaldata->Groups);
 265
 266                         if ((treez == true) && (treei == true)) {
 267                                 //switches node i and node z's info.
 268                                 map<string,int> lib_hold = tree[z].pGroups;
 269                                 tree[z].pGroups = (tree[i].pGroups);
 270                                 tree[i].pGroups = (lib_hold);
 271
 272                                 tree[z].setGroup(tree[z].pGroups.begin()->first);
 273                                 tree[i].setGroup(tree[i].pGroups.begin()->first);
 274
 275                                 map<string,int> gcount_hold = tree[z].pcount;
 276                                 tree[z].pcount = (tree[i].pcount);
 277                                 tree[i].pcount = (gcount_hold);
 278                         }
 279                 }
 280         }
 281         catch(exception& e) {
 282                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 283                 exit(1);
 284         }
 285         catch(...) {
 286                 cout << "An unknown error has occurred in the Tree class function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 287                 exit(1);
 288         }
 289 }
 290 /**************************************************************************************************/
 291
 292 void Tree::randomBlengths()  {
 293         try {
 294                 for(int i=numNodes-1;i>=0;i--){
 295                         int z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 296
 297                         float bl_hold = tree[z].getBranchLength();
 298                         tree[z].setBranchLength(tree[i].getBranchLength());
 299                         tree[i].setBranchLength(bl_hold);
 300                 }
 301         }
 302         catch(exception& e) {
 303                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 304                 exit(1);
 305         }
 306         catch(...) {
 307                 cout << "An unknown error has occurred in the Tree class function randomBlengths. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 308                 exit(1);
 309         }
 310 }
 311 /*************************************************************************************************/
 312 void Tree::assembleRandomUnifracTree() {
 313         randomLabels();
 314         assembleTree();
 315 }
 316
 317 /*************************************************************************************************/
 318 //for now it's just random topology but may become random labels as well later that why this is such a simple function now...
 319 void Tree::assembleRandomTree() {
 320         randomTopology();
 321         assembleTree();
 322 }
 323 /**************************************************************************************************/
 324
 325 void Tree::randomTopology() {
 326         try {
 327                 for(int i=0;i<numNodes;i++){
 328                         tree[i].setParent(-1);
 329                 }
 330                 for(int i=numLeaves;i<numNodes;i++){
 331                         tree[i].setChildren(-1, -1);
 332                 }
 333
 334                 for(int i=numLeaves;i<numNodes;i++){
 335                         int escape =0;
 336                         int rnd_index1, rnd_index2;
 337                         while(escape == 0){
 338                                 rnd_index1 = (int)(((double)rand() / (double) RAND_MAX)*i);
 339                                 if(tree[rnd_index1].getParent() == -1){escape = 1;}
 340                         }
 341
 342                         escape = 0;
 343                         while(escape == 0){
 344                                 rnd_index2 = (int)(((double)rand() / (double) RAND_MAX)*i);
 345                                 if(rnd_index2 != rnd_index1 && tree[rnd_index2].getParent() == -1){
 346                                         escape = 1;
 347                                 }
 348                         }
 349
 350                         tree[i].setChildren(rnd_index1,rnd_index2);
 351                         tree[i].setParent(-1);
 352                         tree[rnd_index1].setParent(i);
 353                         tree[rnd_index2].setParent(i);
 354                 }
 355         }
 356         catch(exception& e) {
 357                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 358                 exit(1);
 359         }
 360         catch(...) {
 361                 cout << "An unknown error has occurred in the Tree class function randomTopology. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 362                 exit(1);
 363         }
 364 }
 365
 366 /*****************************************************************/
 367 // This prints out the tree in Newick form.
 368 void Tree::createNewickFile() {
 369         try {
 370                 int root = findRoot();
 371                 filename = getRootName(globaldata->getTreeFile()) + "newick";
 372                 openOutputFile(filename, out);
 373
 374                 printBranch(root);
 375
 376                 // you are at the end of the tree
 377                 out << ";" << endl;
 378         }
 379         catch(exception& e) {
 380                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 381                 exit(1);
 382         }
 383         catch(...) {
 384                 cout << "An unknown error has occurred in the Tree class function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 385                 exit(1);
 386         }
 387 }
 388
 389 /*****************************************************************/
 390 //This function finds the index of the root node.
 391
 392 int Tree::findRoot() {
 393         try {
 394                 for (int i = 0; i < numNodes; i++) {
 395                         //you found the root
 396                         if (tree[i].getParent() == -1) { return i; }
 397                 }
 398                 return -1;
 399         }
 400         catch(exception& e) {
 401                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 402                 exit(1);
 403         }
 404         catch(...) {
 405                 cout << "An unknown error has occurred in the Tree class function findRoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 406                 exit(1);
 407         }
 408 }
 409
 410 /*****************************************************************/
 411 void Tree::printBranch(int node) {
 412         try {
 413
 414                 // you are not a leaf
 415                 if (tree[node].getLChild() != -1) {
 416                         out << "(";
 417                         printBranch(tree[node].getLChild());
 418                         out << ",";
 419                         printBranch(tree[node].getRChild());
 420                         out << ")";
 421                 }else { //you are a leaf
 422                         tree[node].printNode(out);  //prints out name and branch length
 423                 }
 424
 425         }
 426         catch(exception& e) {
 427                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 428                 exit(1);
 429         }
 430         catch(...) {
 431                 cout << "An unknown error has occurred in the Tree class function printBranch. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 432                 exit(1);
 433         }
 434 }
 435
 436 /*****************************************************************/
 437 void Tree::setGroups() {
 438         try {
 439                 //if the user has not entered specific groups to analyze then do them all
 440                 if (globaldata->Groups.size() != 0) {
 441                         //check that groups are valid
 442                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 443                                 if (globaldata->gTreemap->isValidGroup(globaldata->Groups[i]) != true) {
 444                                         cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 445                                         // erase the invalid group from globaldata->Groups
 446                                         globaldata->Groups.erase (globaldata->Groups.begin()+i);
 447                                 }
 448                         }
 449
 450                         //if the user only entered invalid groups
 451                         if (globaldata->Groups.size() == 0) {
 452                                 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;
 453                                 for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 454                                         globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 455                                 }
 456                         }
 457
 458                 }else {
 459                         for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 460                                 globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
 461                         }
 462                 }
 463         }
 464         catch(exception& e) {
 465                 cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 466                 exit(1);
 467         }
 468         catch(...) {
 469                 cout << "An unknown error has occurred in the Tree class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 470                 exit(1);
 471         }
 472
 473 }
 474