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 Tree::Tree(int num) {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 m = MothurOut::getInstance();
  17
  18                 numLeaves = num;
  19                 numNodes = 2*numLeaves - 1;
  20
  21                 tree.resize(numNodes);
  22         }
  23         catch(exception& e) {
  24                 m->errorOut(e, "Tree", "Tree - numNodes");
  25                 exit(1);
  26         }
  27 }
  28 /*****************************************************************/
  29 Tree::Tree(string g) {
  30         try {
  31                 globaldata = GlobalData::getInstance();
  32                 m = MothurOut::getInstance();
  33
  34                 parseTreeFile();  globaldata->runParse = false;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "Tree", "Tree - just parse");
  38                 exit(1);
  39         }
  40 }
  41 /*****************************************************************/
  42 Tree::Tree() {
  43         try {
  44                 globaldata = GlobalData::getInstance();
  45                 m = MothurOut::getInstance();
  46
  47                 if (globaldata->runParse == true) {  parseTreeFile();  globaldata->runParse = false;  }
  48 //for(int i = 0; i <    globaldata->Treenames.size(); i++) { cout << i << '\t' << globaldata->Treenames[i] << endl;  }
  49                 numLeaves = globaldata->Treenames.size();
  50                 numNodes = 2*numLeaves - 1;
  51
  52                 tree.resize(numNodes);
  53
  54                 //initialize groupNodeInfo
  55                 for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
  56                         groupNodeInfo[globaldata->gTreemap->namesOfGroups[i]].resize(0);
  57                 }
  58
  59                 //initialize tree with correct number of nodes, name and group info.
  60                 for (int i = 0; i < numNodes; i++) {
  61                         //initialize leaf nodes
  62                         if (i <= (numLeaves-1)) {
  63                                 tree[i].setName(globaldata->Treenames[i]);
  64
  65                                 //save group info
  66                                 string group = globaldata->gTreemap->getGroup(globaldata->Treenames[i]);
  67                                 vector<string> tempGroups; tempGroups.push_back(group);
  68                                 tree[i].setGroup(tempGroups);
  69                                 groupNodeInfo[group].push_back(i);
  70
  71                                 //set pcount and pGroup for groupname to 1.
  72                                 tree[i].pcount[group] = 1;
  73                                 tree[i].pGroups[group] = 1;
  74
  75                                 //Treemap knows name, group and index to speed up search
  76                                 globaldata->gTreemap->setIndex(globaldata->Treenames[i], i);
  77
  78                         //intialize non leaf nodes
  79                         }else if (i > (numLeaves-1)) {
  80                                 tree[i].setName("");
  81                                 vector<string> tempGroups;
  82                                 tree[i].setGroup(tempGroups);
  83                         }
  84                 }
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "Tree", "Tree");
  88                 exit(1);
  89         }
  90 }
  91
  92 /*****************************************************************/
  93 Tree::~Tree() {}
  94 /*****************************************************************/
  95 void Tree::addNamesToCounts() {
  96         try {
  97                 //ex. seq1      seq2,seq3,se4
  98                 //              seq1 = pasture
  99                 //              seq2 = forest
 100                 //              seq4 = pasture
 101                 //              seq3 = ocean
 102
 103                 //before this function seq1.pcount = pasture -> 1
 104                 //after                            seq1.pcount = pasture -> 2, forest -> 1, ocean -> 1
 105
 106                 //before this function seq1.pgroups = pasture -> 1
 107                 //after                            seq1.pgroups = pasture -> 1 since that is the dominant group
 108
 109
 110                 //go through each leaf and update its pcounts and pgroups
 111
 112                 //float A = clock();
 113
 114                 for (int i = 0; i < numLeaves; i++) {
 115
 116                         string name = tree[i].getName();
 117
 118                         map<string, string>::iterator itNames = globaldata->names.find(name);
 119
 120                         if (itNames == globaldata->names.end()) { m->mothurOut(name + " is not in your name file, please correct."); m->mothurOutEndLine(); exit(1);  }
 121                         else {
 122                                 vector<string> dupNames;
 123                                 m->splitAtComma(globaldata->names[name], dupNames);
 124
 125                                 map<string, int>::iterator itCounts;
 126                                 int maxPars = 1;
 127                                 set<string> groupsAddedForThisNode;
 128                                 for (int j = 0; j < dupNames.size(); j++) {
 129
 130                                         string group = globaldata->gTreemap->getGroup(dupNames[j]);
 131
 132                                         if (dupNames[j] != name) {//you already added yourself in the constructor
 133
 134                                                 if (groupsAddedForThisNode.count(group) == 0)  {  groupNodeInfo[group].push_back(i);  groupsAddedForThisNode.insert(group);  } //if you have not already added this node for this group, then add it
 135
 136                                                 //update pcounts
 137                                                 itCounts = tree[i].pcount.find(group);
 138                                                 if (itCounts == tree[i].pcount.end()) { //new group, add it
 139                                                         tree[i].pcount[group] = 1;
 140                                                 }else {
 141                                                         tree[i].pcount[group]++;
 142                                                 }
 143
 144                                                 //update pgroups
 145                                                 itCounts = tree[i].pGroups.find(group);
 146                                                 if (itCounts == tree[i].pGroups.end()) { //new group, add it
 147                                                         tree[i].pGroups[group] = 1;
 148                                                 }else{
 149                                                         tree[i].pGroups[group]++;
 150                                                 }
 151
 152                                                 //keep highest group
 153                                                 if(tree[i].pGroups[group] > maxPars){
 154                                                         maxPars = tree[i].pGroups[group];
 155                                                 }
 156                                         }else {  groupsAddedForThisNode.insert(group);  } //add it so you don't add it to groupNodeInfo again
 157                                 }//end for
 158
 159                                 if (maxPars > 1) { //then we have some more dominant groups
 160                                         //erase all the groups that are less than maxPars because you found a more dominant group.
 161                                         for(it=tree[i].pGroups.begin();it!=tree[i].pGroups.end();){
 162                                                 if(it->second < maxPars){
 163                                                         tree[i].pGroups.erase(it++);
 164                                                 }else { it++; }
 165                                         }
 166                                         //set one remaining groups to 1
 167                                         for(it=tree[i].pGroups.begin();it!=tree[i].pGroups.end();it++){
 168                                                 tree[i].pGroups[it->first] = 1;
 169                                         }
 170                                 }//end if
 171
 172                                 //update groups to reflect all the groups this node represents
 173                                 vector<string> nodeGroups;
 174                                 map<string, int>::iterator itGroups;
 175                                 for (itGroups = tree[i].pcount.begin(); itGroups != tree[i].pcount.end(); itGroups++) {
 176                                         nodeGroups.push_back(itGroups->first);
 177                                 }
 178                                 tree[i].setGroup(nodeGroups);
 179
 180                         }//end else
 181                 }//end for
 182
 183                 //float B = clock();
 184                 //cout << "addNamesToCounts\t" << (B - A) / CLOCKS_PER_SEC << endl;
 185
 186         }
 187         catch(exception& e) {
 188                 m->errorOut(e, "Tree", "addNamesToCounts");
 189                 exit(1);
 190         }
 191 }
 192 /*****************************************************************/
 193 int Tree::getIndex(string searchName) {
 194         try {
 195                 //Treemap knows name, group and index to speed up search
 196                 // getIndex function will return the vector index or -1 if seq is not found.
 197                 int index = globaldata->gTreemap->getIndex(searchName);
 198                 return index;
 199
 200         }
 201         catch(exception& e) {
 202                 m->errorOut(e, "Tree", "getIndex");
 203                 exit(1);
 204         }
 205 }
 206 /*****************************************************************/
 207
 208 void Tree::setIndex(string searchName, int index) {
 209         try {
 210                 //set index in treemap
 211                 globaldata->gTreemap->setIndex(searchName, index);
 212         }
 213         catch(exception& e) {
 214                 m->errorOut(e, "Tree", "setIndex");
 215                 exit(1);
 216         }
 217 }
 218 /*****************************************************************/
 219 int Tree::assembleTree() {
 220         try {
 221                 //float A = clock();
 222
 223                 //if user has given a names file we want to include that info in the pgroups and pcount info.
 224                 if(globaldata->names.size() != 0) {  addNamesToCounts();  }
 225
 226                 //build the pGroups in non leaf nodes to be used in the parsimony calcs.
 227                 for (int i = numLeaves; i < numNodes; i++) {
 228                         if (m->control_pressed) { return 1; }
 229
 230                         tree[i].pGroups = (mergeGroups(i));
 231                         tree[i].pcount = (mergeGcounts(i));
 232                 }
 233                 //float B = clock();
 234                 //cout << "assembleTree\t" << (B-A) / CLOCKS_PER_SEC << endl;
 235                 return 0;
 236         }
 237         catch(exception& e) {
 238                 m->errorOut(e, "Tree", "assembleTree");
 239                 exit(1);
 240         }
 241 }
 242 /*****************************************************************/
 243 int Tree::assembleTree(string n) {
 244         try {
 245
 246                 //build the pGroups in non leaf nodes to be used in the parsimony calcs.
 247                 for (int i = numLeaves; i < numNodes; i++) {
 248                         if (m->control_pressed) { return 1; }
 249
 250                         tree[i].pGroups = (mergeGroups(i));
 251                         tree[i].pcount = (mergeGcounts(i));
 252                 }
 253                 //float B = clock();
 254                 //cout << "assembleTree\t" << (B-A) / CLOCKS_PER_SEC << endl;
 255                 return 0;
 256         }
 257         catch(exception& e) {
 258                 m->errorOut(e, "Tree", "assembleTree");
 259                 exit(1);
 260         }
 261 }
 262 /*****************************************************************/
 263 void Tree::getSubTree(Tree* copy, vector<string> Groups) {
 264         try {
 265
 266                 //we want to select some of the leaf nodes to create the output tree
 267                 //go through the input Tree starting at parents of leaves
 268                 for (int i = 0; i < numNodes; i++) {
 269
 270                         //initialize leaf nodes
 271                         if (i <= (numLeaves-1)) {
 272                                 tree[i].setName(Groups[i]);
 273
 274                                 //save group info
 275                                 string group = globaldata->gTreemap->getGroup(Groups[i]);
 276                                 vector<string> tempGroups; tempGroups.push_back(group);
 277                                 tree[i].setGroup(tempGroups);
 278                                 groupNodeInfo[group].push_back(i);
 279
 280                                 //set pcount and pGroup for groupname to 1.
 281                                 tree[i].pcount[group] = 1;
 282                                 tree[i].pGroups[group] = 1;
 283
 284                                 //Treemap knows name, group and index to speed up search
 285                                 globaldata->gTreemap->setIndex(Groups[i], i);
 286
 287                                 //intialize non leaf nodes
 288                         }else if (i > (numLeaves-1)) {
 289                                 tree[i].setName("");
 290                                 vector<string> tempGroups;
 291                                 tree[i].setGroup(tempGroups);
 292                         }
 293                 }
 294
 295                 set<int> removedLeaves;
 296                 for (int i = 0; i < copy->getNumLeaves(); i++) {
 297
 298                         if (removedLeaves.count(i) == 0) {
 299
 300                         //am I in the group
 301                         int parent = copy->tree[i].getParent();
 302
 303                         if (parent != -1) {
 304                                 if (m->inUsersGroups(copy->tree[i].getName(), Groups)) {
 305                                         //find my siblings name
 306                                         int parentRC = copy->tree[parent].getRChild();
 307                                         int parentLC = copy->tree[parent].getLChild();
 308
 309                                         //if I am the right child, then my sib is the left child
 310                                         int sibIndex = parentRC;
 311                                         if (parentRC == i) { sibIndex = parentLC; }
 312
 313                                         string sibsName = copy->tree[sibIndex].getName();
 314
 315                                         //if yes, is my sibling
 316                                         if ((m->inUsersGroups(sibsName, Groups)) || (sibsName == "")) {
 317                                                 //we both are okay no trimming required
 318                                         }else{
 319                                                 //i am, my sib is not, so remove sib by setting my parent to my grandparent
 320                                                 int grandparent = copy->tree[parent].getParent();
 321                                                 int grandparentLC = copy->tree[grandparent].getLChild();
 322                                                 int grandparentRC = copy->tree[grandparent].getRChild();
 323
 324                                                 //whichever of my granparents children was my parent now equals me
 325                                                 if (grandparentLC == parent) { grandparentLC = i; }
 326                                                 else { grandparentRC = i; }
 327
 328                                                 copy->tree[i].setParent(grandparent);
 329                                                 copy->tree[i].setBranchLength((copy->tree[i].getBranchLength()+copy->tree[parent].getBranchLength()));
 330                                                 copy->tree[grandparent].setChildren(grandparentLC, grandparentRC);
 331                                                 removedLeaves.insert(sibIndex);
 332                                         }
 333                                 }else{
 334                                         //find my siblings name
 335                                         int parentRC = copy->tree[parent].getRChild();
 336                                         int parentLC = copy->tree[parent].getLChild();
 337
 338                                         //if I am the right child, then my sib is the left child
 339                                         int sibIndex = parentRC;
 340                                         if (parentRC == i) { sibIndex = parentLC; }
 341
 342                                         string sibsName = copy->tree[sibIndex].getName();
 343
 344                                         //if no is my sibling
 345                                         if ((m->inUsersGroups(sibsName, Groups)) || (sibsName == "")) {
 346                                                 //i am not, but my sib is
 347                                                 int grandparent = copy->tree[parent].getParent();
 348                                                 int grandparentLC = copy->tree[grandparent].getLChild();
 349                                                 int grandparentRC = copy->tree[grandparent].getRChild();
 350
 351                                                 //whichever of my granparents children was my parent now equals my sib
 352                                                 if (grandparentLC == parent) { grandparentLC = sibIndex; }
 353                                                 else { grandparentRC = sibIndex; }
 354
 355                                                 copy->tree[sibIndex].setParent(grandparent);
 356                                                 copy->tree[sibIndex].setBranchLength((copy->tree[sibIndex].getBranchLength()+copy->tree[parent].getBranchLength()));
 357                                                 copy->tree[grandparent].setChildren(grandparentLC, grandparentRC);
 358                                                 removedLeaves.insert(i);
 359                                         }else{
 360                                                 //neither of us are, so we want to eliminate ourselves and our parent
 361                                                 //so set our parents sib to our great-grandparent
 362                                                 int parent = copy->tree[i].getParent();
 363                                                 int grandparent = copy->tree[parent].getParent();
 364
 365                                                 if (grandparent != -1) {
 366                                                         int greatgrandparent = copy->tree[grandparent].getParent();
 367                                                         int greatgrandparentLC = copy->tree[greatgrandparent].getLChild();
 368                                                         int greatgrandparentRC = copy->tree[greatgrandparent].getRChild();
 369
 370                                                         int grandparentLC = copy->tree[grandparent].getLChild();
 371                                                         int grandparentRC = copy->tree[grandparent].getRChild();
 372
 373                                                         int parentsSibIndex = grandparentLC;
 374                                                         if (grandparentRC == parent) { parentsSibIndex = grandparentLC; }
 375
 376                                                         //whichever of my greatgrandparents children was my grandparent
 377                                                         if (greatgrandparentLC == grandparent) { greatgrandparentLC = parentsSibIndex; }
 378                                                         else { greatgrandparentRC = parentsSibIndex; }
 379
 380                                                         copy->tree[parentsSibIndex].setParent(greatgrandparent);
 381                                                         copy->tree[parentsSibIndex].setBranchLength((copy->tree[parentsSibIndex].getBranchLength()+copy->tree[grandparent].getBranchLength()));
 382                                                         copy->tree[greatgrandparent].setChildren(greatgrandparentLC, greatgrandparentRC);
 383                                                 }else{
 384                                                         copy->tree[parent].setChildren(-1, -1);
 385                                                         cout << "issues with making subtree" << endl;
 386                                                 }
 387                                                 removedLeaves.insert(sibIndex);
 388                                                 removedLeaves.insert(i);
 389                                         }
 390                                 }
 391                         }
 392                         }
 393                 }
 394
 395                 int root = 0;
 396                 for (int i = 0; i < copy->getNumNodes(); i++) {
 397                         //you found the root
 398                         if (copy->tree[i].getParent() == -1) { root = i; break; }
 399                 }
 400
 401                 int nextSpot = numLeaves;
 402                 populateNewTree(copy->tree, root, nextSpot);
 403
 404
 405         }
 406         catch(exception& e) {
 407                 m->errorOut(e, "Tree", "getCopy");
 408                 exit(1);
 409         }
 410 }
 411 /*****************************************************************/
 412 int Tree::populateNewTree(vector<Node>& oldtree, int node, int& index) {
 413         try {
 414
 415                 if (oldtree[node].getLChild() != -1) {
 416                         int rc = populateNewTree(oldtree, oldtree[node].getLChild(), index);
 417                         int lc = populateNewTree(oldtree, oldtree[node].getRChild(), index);
 418
 419                         tree[index].setChildren(lc, rc);
 420                         index++;
 421
 422                         return (index-1);
 423                 }else { //you are a leaf
 424                         int indexInNewTree = globaldata->gTreemap->getIndex(oldtree[node].getName());
 425
 426                         tree[indexInNewTree].setParent(index);
 427                         return indexInNewTree;
 428
 429                 }
 430         }
 431         catch(exception& e) {
 432                 m->errorOut(e, "Tree", "populateNewTree");
 433                 exit(1);
 434         }
 435 }
 436 /*****************************************************************/
 437 void Tree::getCopy(Tree* copy) {
 438         try {
 439
 440                 //for each node in the tree copy its info
 441                 for (int i = 0; i < numNodes; i++) {
 442                         //copy name
 443                         tree[i].setName(copy->tree[i].getName());
 444
 445                         //copy group
 446                         tree[i].setGroup(copy->tree[i].getGroup());
 447
 448                         //copy branch length
 449                         tree[i].setBranchLength(copy->tree[i].getBranchLength());
 450
 451                         //copy parent
 452                         tree[i].setParent(copy->tree[i].getParent());
 453
 454                         //copy children
 455                         tree[i].setChildren(copy->tree[i].getLChild(), copy->tree[i].getRChild());
 456
 457                         //copy index in node and tmap
 458                         tree[i].setIndex(copy->tree[i].getIndex());
 459                         setIndex(copy->tree[i].getName(), getIndex(copy->tree[i].getName()));
 460
 461                         //copy pGroups
 462                         tree[i].pGroups = copy->tree[i].pGroups;
 463
 464                         //copy pcount
 465                         tree[i].pcount = copy->tree[i].pcount;
 466                 }
 467
 468                 groupNodeInfo = copy->groupNodeInfo;
 469
 470         }
 471         catch(exception& e) {
 472                 m->errorOut(e, "Tree", "getCopy");
 473                 exit(1);
 474         }
 475 }
 476 /*****************************************************************/
 477 //returns a map with a groupname and the number of times that group was seen in the children
 478 //for instance if your children are white and black then it would return a map with 2 entries
 479 // p[white] = 1 and p[black] = 1.  Now go up a level and merge that with a node who has p[white] = 1
 480 //and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
 481
 482 map<string, int> Tree::mergeGroups(int i) {
 483         try {
 484                 int lc = tree[i].getLChild();
 485                 int rc = tree[i].getRChild();
 486
 487                 //set parsimony groups to left child
 488                 map<string,int> parsimony = tree[lc].pGroups;
 489
 490                 int maxPars = 1;
 491
 492                 //look at right child groups and update maxPars if right child has something higher for that group.
 493                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
 494                         it2 = parsimony.find(it->first);
 495                         if (it2 != parsimony.end()) {
 496                                 parsimony[it->first]++;
 497                         }else {
 498                                 parsimony[it->first] = 1;
 499                         }
 500
 501                         if(parsimony[it->first] > maxPars){
 502                                 maxPars = parsimony[it->first];
 503                         }
 504                 }
 505
 506                 // this is true if right child had a greater parsimony for a certain group
 507                 if(maxPars > 1){
 508                         //erase all the groups that are only 1 because you found something with 2.
 509                         for(it=parsimony.begin();it!=parsimony.end();){
 510                                 if(it->second == 1){
 511                                         parsimony.erase(it++);
 512                                 }else { it++; }
 513                         }
 514                         //set one remaining groups to 1
 515                         //so with our above example p[white] = 2 would be left and it would become p[white] = 1
 516                         for(it=parsimony.begin();it!=parsimony.end();it++){
 517                                 parsimony[it->first] = 1;
 518                         }
 519
 520                 }
 521
 522                 return parsimony;
 523         }
 524         catch(exception& e) {
 525                 m->errorOut(e, "Tree", "mergeGroups");
 526                 exit(1);
 527         }
 528 }
 529 /*****************************************************************/
 530 //returns a map with a groupname and the number of times that group was seen in the children
 531 //for instance if your children are white and black then it would return a map with 2 entries
 532 // p[white] = 1 and p[black] = 1.  Now go up a level and merge that with a node who has p[white] = 1
 533 //and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
 534
 535 map<string, int> Tree::mergeUserGroups(int i, vector<string> g) {
 536         try {
 537
 538                 int lc = tree[i].getLChild();
 539                 int rc = tree[i].getRChild();
 540
 541                 //loop through nodes groups removing the ones the user doesn't want
 542                 for(it=tree[lc].pGroups.begin();it!=tree[lc].pGroups.end();){
 543                                 if (m->inUsersGroups(it->first, g) != true) {
 544                                         tree[lc].pGroups.erase(it++);
 545                                 }else { it++; }
 546                 }
 547
 548                 //loop through nodes groups removing the ones the user doesn't want
 549                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();){
 550                                 if (m->inUsersGroups(it->first, g) != true) {
 551                                         tree[rc].pGroups.erase(it++);
 552                                 }else { it++; }
 553                 }
 554
 555                 //set parsimony groups to left child
 556                 map<string,int> parsimony = tree[lc].pGroups;
 557
 558                 int maxPars = 1;
 559
 560                 //look at right child groups and update maxPars if right child has something higher for that group.
 561                 for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
 562                         it2 = parsimony.find(it->first);
 563                         if (it2 != parsimony.end()) {
 564                                 parsimony[it->first]++;
 565                         }else {
 566                                 parsimony[it->first] = 1;
 567                         }
 568
 569                         if(parsimony[it->first] > maxPars){
 570                                 maxPars = parsimony[it->first];
 571                         }
 572                 }
 573
 574                 // this is true if right child had a greater parsimony for a certain group
 575                 if(maxPars > 1){
 576                         //erase all the groups that are only 1 because you found something with 2.
 577                         for(it=parsimony.begin();it!=parsimony.end();){
 578                                 if(it->second == 1){
 579                                         parsimony.erase(it++);
 580                                 }else { it++; }
 581                         }
 582
 583                         for(it=parsimony.begin();it!=parsimony.end();it++){
 584                                 parsimony[it->first] = 1;
 585                         }
 586                 }
 587
 588                 return parsimony;
 589         }
 590         catch(exception& e) {
 591                 m->errorOut(e, "Tree", "mergeUserGroups");
 592                 exit(1);
 593         }
 594 }
 595
 596
 597 /**************************************************************************************************/
 598
 599 map<string,int> Tree::mergeGcounts(int position) {
 600         try{
 601                 map<string,int>::iterator pos;
 602
 603                 int lc = tree[position].getLChild();
 604                 int rc = tree[position].getRChild();
 605
 606                 map<string,int> sum = tree[lc].pcount;
 607
 608                 for(it=tree[rc].pcount.begin();it!=tree[rc].pcount.end();it++){
 609                         sum[it->first] += it->second;
 610                 }
 611                 return sum;
 612         }
 613         catch(exception& e) {
 614                 m->errorOut(e, "Tree", "mergeGcounts");
 615                 exit(1);
 616         }
 617 }
 618 /**************************************************************************************************/
 619
 620 void Tree::randomLabels(vector<string> g) {
 621         try {
 622
 623                 //initialize groupNodeInfo
 624                 for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
 625                         groupNodeInfo[globaldata->gTreemap->namesOfGroups[i]].resize(0);
 626                 }
 627
 628                 for(int i = 0; i < numLeaves; i++){
 629                         int z;
 630                         //get random index to switch with
 631                         z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 632
 633                         //you only want to randomize the nodes that are from a group the user wants analyzed, so
 634                         //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.
 635                         bool treez, treei;
 636
 637                         treez = m->inUsersGroups(tree[z].getGroup(), g);
 638                         treei = m->inUsersGroups(tree[i].getGroup(), g);
 639
 640                         if ((treez == true) && (treei == true)) {
 641                                 //switches node i and node z's info.
 642                                 map<string,int> lib_hold = tree[z].pGroups;
 643                                 tree[z].pGroups = (tree[i].pGroups);
 644                                 tree[i].pGroups = (lib_hold);
 645
 646                                 vector<string> zgroup = tree[z].getGroup();
 647                                 tree[z].setGroup(tree[i].getGroup());
 648                                 tree[i].setGroup(zgroup);
 649
 650                                 string zname = tree[z].getName();
 651                                 tree[z].setName(tree[i].getName());
 652                                 tree[i].setName(zname);
 653
 654                                 map<string,int> gcount_hold = tree[z].pcount;
 655                                 tree[z].pcount = (tree[i].pcount);
 656                                 tree[i].pcount = (gcount_hold);
 657                         }
 658
 659                         for (int k = 0; k < (tree[i].getGroup()).size(); k++) {  groupNodeInfo[(tree[i].getGroup())[k]].push_back(i); }
 660                         for (int k = 0; k < (tree[z].getGroup()).size(); k++) {  groupNodeInfo[(tree[z].getGroup())[k]].push_back(z); }
 661                 }
 662         }
 663         catch(exception& e) {
 664                 m->errorOut(e, "Tree", "randomLabels");
 665                 exit(1);
 666         }
 667 }
 668 /**************************************************************************************************
 669
 670 void Tree::randomLabels(string groupA, string groupB) {
 671         try {
 672                 int numSeqsA = globaldata->gTreemap->seqsPerGroup[groupA];
 673                 int numSeqsB = globaldata->gTreemap->seqsPerGroup[groupB];
 674
 675                 vector<string> randomGroups(numSeqsA+numSeqsB, groupA);
 676                 for(int i=numSeqsA;i<randomGroups.size();i++){
 677                         randomGroups[i] = groupB;
 678                 }
 679                 random_shuffle(randomGroups.begin(), randomGroups.end());
 680
 681                 int randomCounter = 0;
 682                 for(int i=0;i<numLeaves;i++){
 683                         if(tree[i].getGroup() == groupA || tree[i].getGroup() == groupB){
 684                                 tree[i].setGroup(randomGroups[randomCounter]);
 685                                 tree[i].pcount.clear();
 686                                 tree[i].pcount[randomGroups[randomCounter]] = 1;
 687                                 tree[i].pGroups.clear();
 688                                 tree[i].pGroups[randomGroups[randomCounter]] = 1;
 689                                 randomCounter++;
 690                         }
 691                 }
 692         }
 693         catch(exception& e) {
 694                 m->errorOut(e, "Tree", "randomLabels");
 695                 exit(1);
 696         }
 697 }
 698 /**************************************************************************************************/
 699 void Tree::randomBlengths()  {
 700         try {
 701                 for(int i=numNodes-1;i>=0;i--){
 702                         int z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
 703
 704                         float bl_hold = tree[z].getBranchLength();
 705                         tree[z].setBranchLength(tree[i].getBranchLength());
 706                         tree[i].setBranchLength(bl_hold);
 707                 }
 708         }
 709         catch(exception& e) {
 710                 m->errorOut(e, "Tree", "randomBlengths");
 711                 exit(1);
 712         }
 713 }
 714 /*************************************************************************************************/
 715 void Tree::assembleRandomUnifracTree(vector<string> g) {
 716         randomLabels(g);
 717         assembleTree("noNameCounts");
 718 }
 719 /*************************************************************************************************/
 720 void Tree::assembleRandomUnifracTree(string groupA, string groupB) {
 721
 722         vector<string> temp; temp.push_back(groupA); temp.push_back(groupB);
 723         randomLabels(temp);
 724         assembleTree("noNameCounts");
 725 }
 726
 727 /*************************************************************************************************/
 728 //for now it's just random topology but may become random labels as well later that why this is such a simple function now...
 729 void Tree::assembleRandomTree() {
 730         randomTopology();
 731         assembleTree();
 732 }
 733 /**************************************************************************************************/
 734
 735 void Tree::randomTopology() {
 736         try {
 737                 for(int i=0;i<numNodes;i++){
 738                         tree[i].setParent(-1);
 739                 }
 740                 for(int i=numLeaves;i<numNodes;i++){
 741                         tree[i].setChildren(-1, -1);
 742                 }
 743
 744                 for(int i=numLeaves;i<numNodes;i++){
 745                         int escape =0;
 746                         int rnd_index1, rnd_index2;
 747                         while(escape == 0){
 748                                 rnd_index1 = (int)(((double)rand() / (double) RAND_MAX)*i);
 749                                 if(tree[rnd_index1].getParent() == -1){escape = 1;}
 750                         }
 751
 752                         escape = 0;
 753                         while(escape == 0){
 754                                 rnd_index2 = (int)(((double)rand() / (double) RAND_MAX)*i);
 755                                 if(rnd_index2 != rnd_index1 && tree[rnd_index2].getParent() == -1){
 756                                         escape = 1;
 757                                 }
 758                         }
 759
 760                         tree[i].setChildren(rnd_index1,rnd_index2);
 761                         tree[i].setParent(-1);
 762                         tree[rnd_index1].setParent(i);
 763                         tree[rnd_index2].setParent(i);
 764                 }
 765         }
 766         catch(exception& e) {
 767                 m->errorOut(e, "Tree", "randomTopology");
 768                 exit(1);
 769         }
 770 }
 771 /*****************************************************************/
 772 void Tree::print(ostream& out) {
 773         try {
 774                 int root = findRoot();
 775                 printBranch(root, out, "branch");
 776                 out << ";" << endl;
 777         }
 778         catch(exception& e) {
 779                 m->errorOut(e, "Tree", "print");
 780                 exit(1);
 781         }
 782 }
 783 /*****************************************************************/
 784 void Tree::print(ostream& out, string mode) {
 785         try {
 786                 int root = findRoot();
 787                 printBranch(root, out, mode);
 788                 out << ";" << endl;
 789         }
 790         catch(exception& e) {
 791                 m->errorOut(e, "Tree", "print");
 792                 exit(1);
 793         }
 794 }
 795 /*****************************************************************/
 796 // This prints out the tree in Newick form.
 797 void Tree::createNewickFile(string f) {
 798         try {
 799                 int root = findRoot();
 800                 //filename = m->getRootName(globaldata->getTreeFile()) + "newick";
 801                 filename = f;
 802
 803                 m->openOutputFile(filename, out);
 804
 805                 printBranch(root, out, "branch");
 806
 807                 // you are at the end of the tree
 808                 out << ";" << endl;
 809                 out.close();
 810         }
 811         catch(exception& e) {
 812                 m->errorOut(e, "Tree", "createNewickFile");
 813                 exit(1);
 814         }
 815 }
 816
 817 /*****************************************************************/
 818 //This function finds the index of the root node.
 819
 820 int Tree::findRoot() {
 821         try {
 822                 for (int i = 0; i < numNodes; i++) {
 823                         //you found the root
 824                         if (tree[i].getParent() == -1) { return i; }
 825                         //cout << "i = " << i << endl;
 826                         //cout << "i's parent = " << tree[i].getParent() << endl;
 827                 }
 828                 return -1;
 829         }
 830         catch(exception& e) {
 831                 m->errorOut(e, "Tree", "findRoot");
 832                 exit(1);
 833         }
 834 }
 835 /*****************************************************************/
 836 void Tree::printBranch(int node, ostream& out, string mode) {
 837 try {
 838
 839 // you are not a leaf
 840                 if (tree[node].getLChild() != -1) {
 841                         out << "(";
 842                         printBranch(tree[node].getLChild(), out, mode);
 843                         out << ",";
 844                         printBranch(tree[node].getRChild(), out, mode);
 845                         out << ")";
 846                         if (mode == "branch") {
 847                                 //if there is a branch length then print it
 848                                 if (tree[node].getBranchLength() != -1) {
 849                                         out << ":" << tree[node].getBranchLength();
 850                                 }
 851                         }else if (mode == "boot") {
 852                                 //if there is a label then print it
 853                                 if (tree[node].getLabel() != -1) {
 854                                         out << tree[node].getLabel();
 855                                 }
 856                         }else if (mode == "both") {
 857                                 if (tree[node].getLabel() != -1) {
 858                                         out << tree[node].getLabel();
 859                                 }
 860                                 //if there is a branch length then print it
 861                                 if (tree[node].getBranchLength() != -1) {
 862                                         out << ":" << tree[node].getBranchLength();
 863                                 }
 864                         }
 865                 }else { //you are a leaf
 866                         string leafGroup = globaldata->gTreemap->getGroup(tree[node].getName());
 867
 868                         if (mode == "branch") {
 869                                 out << leafGroup;
 870                                 //if there is a branch length then print it
 871                                 if (tree[node].getBranchLength() != -1) {
 872                                         out << ":" << tree[node].getBranchLength();
 873                                 }
 874                         }else if (mode == "boot") {
 875                                 out << leafGroup;
 876                                 //if there is a label then print it
 877                                 if (tree[node].getLabel() != -1) {
 878                                         out << tree[node].getLabel();
 879                                 }
 880                         }else if (mode == "both") {
 881                                 out << tree[node].getName();
 882                                 if (tree[node].getLabel() != -1) {
 883                                         out << tree[node].getLabel();
 884                                 }
 885                                 //if there is a branch length then print it
 886                                 if (tree[node].getBranchLength() != -1) {
 887                                         out << ":" << tree[node].getBranchLength();
 888                                 }
 889                         }
 890                 }
 891
 892         }
 893         catch(exception& e) {
 894                 m->errorOut(e, "Tree", "printBranch");
 895                 exit(1);
 896         }
 897 }
 898
 899 /*****************************************************************/
 900
 901 void Tree::printTree() {
 902
 903         for(int i=0;i<numNodes;i++){
 904                 cout << i << '\t';
 905                 tree[i].printNode();
 906         }
 907
 908 }
 909
 910 /*****************************************************************/
 911 //this code is a mess and should be rethought...-slw
 912 void Tree::parseTreeFile() {
 913
 914         //only takes names from the first tree and assumes that all trees use the same names.
 915         try {
 916                 string filename = globaldata->getTreeFile();
 917                 ifstream filehandle;
 918                 m->openInputFile(filename, filehandle);
 919                 int c, comment;
 920                 comment = 0;
 921                 int done = 1;
 922
 923                 //ifyou are not a nexus file
 924                 if((c = filehandle.peek()) != '#') {
 925                         while((c = filehandle.peek()) != ';') {
 926                                 while ((c = filehandle.peek()) != ';') {
 927                                         // get past comments
 928                                         if(c == '[') {
 929                                                 comment = 1;
 930                                         }
 931                                         if(c == ']'){
 932                                                 comment = 0;
 933                                         }
 934                                         if((c == '(') && (comment != 1)){ break; }
 935                                         filehandle.get();
 936                                 }
 937
 938                                 done = readTreeString(filehandle);
 939                                 if (done == 0) { break; }
 940                         }
 941                 //ifyou are a nexus file
 942                 }else if((c = filehandle.peek()) == '#') {
 943                         string holder = "";
 944
 945                         // get past comments
 946                         while(holder != "translate" && holder != "Translate"){
 947                                 if(holder == "[" || holder == "[!"){
 948                                         comment = 1;
 949                                 }
 950                                 if(holder == "]"){
 951                                         comment = 0;
 952                                 }
 953                                 filehandle >> holder;
 954
 955                                 //if there is no translate then you must read tree string otherwise use translate to get names
 956                                 if((holder == "tree") && (comment != 1)){
 957                                         //pass over the "tree rep.6878900 = "
 958                                         while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF)) {;}
 959
 960                                         if(c == EOF) { break; }
 961                                         filehandle.putback(c);  //put back first ( of tree.
 962                                         done = readTreeString(filehandle);
 963
 964                                         break;
 965                                 }
 966
 967                                 if (done == 0) { break;  }
 968                         }
 969
 970                         //use nexus translation rather than parsing tree to save time
 971                         if((holder == "translate") || (holder == "Translate")) {
 972
 973                                 string number, name, h;
 974                                 h = ""; // so it enters the loop the first time
 975                                 while((h != ";") && (number != ";")) {
 976                                         filehandle >> number;
 977                                         filehandle >> name;
 978
 979                                         //c = , until done with translation then c = ;
 980                                         h = name.substr(name.length()-1, name.length());
 981                                         name.erase(name.end()-1);  //erase the comma
 982                                         globaldata->Treenames.push_back(number);
 983                                 }
 984                                 if(number == ";") { globaldata->Treenames.pop_back(); }  //in case ';' from translation is on next line instead of next to last name
 985                         }
 986                 }
 987                 filehandle.close();
 988
 989                 //for (int i = 0; i < globaldata->Treenames.size(); i++) {
 990 //cout << globaldata->Treenames[i] << endl; }
 991 //cout << globaldata->Treenames.size() << endl;
 992         }
 993         catch(exception& e) {
 994                 m->errorOut(e, "Tree", "parseTreeFile");
 995                 exit(1);
 996         }
 997 }
 998 /*******************************************************/
 999
1000 /*******************************************************/
1001 int Tree::readTreeString(ifstream& filehandle)  {
1002         try {
1003                 int c;
1004                 string name;  //, k
1005
1006                 while((c = filehandle.peek()) != ';') {
1007 //k = c;
1008 //cout << " at beginning of while " <<  k << endl;
1009                         if(c == ')')  {
1010                                 //to pass over labels in trees
1011                                 c=filehandle.get();
1012                                 while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); }
1013                                 filehandle.putback(c);
1014                         }
1015                         if(c == ';') { return 0; }
1016                         if(c == -1) { return 0; }
1017                         //if you are a name
1018                         if((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != '\t') && (c != 32)) { //32 is space
1019                                 name = "";
1020                                 c = filehandle.get();
1021                         //k = c;
1022 //cout << k << endl;
1023                                 while ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != 32) && (c != '\t')) {
1024                                         name += c;
1025                                         c = filehandle.get();
1026                         //k = c;
1027 //cout << " in name while " << k << endl;
1028                                 }
1029
1030 //cout << "name = " << name << endl;
1031                                 globaldata->Treenames.push_back(name);
1032                                 filehandle.putback(c);
1033 //k = c;
1034 //cout << " after putback" <<  k << endl;
1035                         }
1036
1037                         if(c  == ':') { //read until you reach the end of the branch length
1038                                 while ((c != '(') && (c != ')') && (c != ',') && (c != ';') && (c != '\n') && (c != '\t') && (c != 32)) {
1039                                         c = filehandle.get();
1040         //k = c;
1041         //cout << " in branch while " << k << endl;
1042                                 }
1043                                 filehandle.putback(c);
1044                         }
1045
1046                         c = filehandle.get();
1047 //k = c;
1048         //cout << " here after get " << k << endl;
1049                         if(c == ';') { return 0; }
1050                         if(c == ')') { filehandle.putback(c); }
1051         //k = c;
1052 //cout << k << endl;
1053
1054                 }
1055                 return 0;
1056         }
1057         catch(exception& e) {
1058                 m->errorOut(e, "Tree", "readTreeString");
1059                 exit(1);
1060         }
1061 }
1062
1063 /*******************************************************/
1064
1065 /*******************************************************/
1066