+++ /dev/null
-/*
- * tree.cpp
- * Mothur
- *
- * Created by Sarah Westcott on 1/22/09.
- * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
- *
- */
-
-#include "tree.h"
-
-/*****************************************************************/
-Tree::Tree(int num, TreeMap* t) : tmap(t) {
- try {
- m = MothurOut::getInstance();
-
- numLeaves = num;
- numNodes = 2*numLeaves - 1;
-
- tree.resize(numNodes);
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "Tree - numNodes");
- exit(1);
- }
-}
-/*****************************************************************/
-Tree::Tree(string g) { //do not use tree generated by this its just to extract the treenames, its a chicken before the egg thing that needs to be revisited.
- try {
- m = MothurOut::getInstance();
- parseTreeFile(); m->runParse = false;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "Tree - just parse");
- exit(1);
- }
-}
-/*****************************************************************/
-Tree::Tree(TreeMap* t) : tmap(t) {
- try {
- m = MothurOut::getInstance();
-
- if (m->runParse == true) { parseTreeFile(); m->runParse = false; }
-//for(int i = 0; i < globaldata->Treenames.size(); i++) { cout << i << '\t' << globaldata->Treenames[i] << endl; }
- numLeaves = m->Treenames.size();
- numNodes = 2*numLeaves - 1;
-
- tree.resize(numNodes);
-
- //initialize groupNodeInfo
- for (int i = 0; i < (tmap->getNamesOfGroups()).size(); i++) {
- groupNodeInfo[(tmap->getNamesOfGroups())[i]].resize(0);
- }
-
- //initialize tree with correct number of nodes, name and group info.
- for (int i = 0; i < numNodes; i++) {
- //initialize leaf nodes
- if (i <= (numLeaves-1)) {
- tree[i].setName(m->Treenames[i]);
-
- //save group info
- string group = tmap->getGroup(m->Treenames[i]);
-
- vector<string> tempGroups; tempGroups.push_back(group);
- tree[i].setGroup(tempGroups);
- groupNodeInfo[group].push_back(i);
-
- //set pcount and pGroup for groupname to 1.
- tree[i].pcount[group] = 1;
- tree[i].pGroups[group] = 1;
-
- //Treemap knows name, group and index to speed up search
- tmap->setIndex(m->Treenames[i], i);
-
- //intialize non leaf nodes
- }else if (i > (numLeaves-1)) {
- tree[i].setName("");
- vector<string> tempGroups;
- tree[i].setGroup(tempGroups);
- }
- }
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "Tree");
- exit(1);
- }
-}
-/*****************************************************************/
-Tree::Tree(TreeMap* t, vector< vector<double> >& sims) : tmap(t) {
- try {
- m = MothurOut::getInstance();
-
- if (m->runParse == true) { parseTreeFile(); m->runParse = false; }
- numLeaves = m->Treenames.size();
- numNodes = 2*numLeaves - 1;
-
- tree.resize(numNodes);
-
- //initialize groupNodeInfo
- for (int i = 0; i < (tmap->getNamesOfGroups()).size(); i++) {
- groupNodeInfo[(tmap->getNamesOfGroups())[i]].resize(0);
- }
-
- //initialize tree with correct number of nodes, name and group info.
- for (int i = 0; i < numNodes; i++) {
- //initialize leaf nodes
- if (i <= (numLeaves-1)) {
- tree[i].setName(m->Treenames[i]);
-
- //save group info
- string group = tmap->getGroup(m->Treenames[i]);
-
- vector<string> tempGroups; tempGroups.push_back(group);
- tree[i].setGroup(tempGroups);
- groupNodeInfo[group].push_back(i);
-
- //set pcount and pGroup for groupname to 1.
- tree[i].pcount[group] = 1;
- tree[i].pGroups[group] = 1;
-
- //Treemap knows name, group and index to speed up search
- tmap->setIndex(m->Treenames[i], i);
-
- //intialize non leaf nodes
- }else if (i > (numLeaves-1)) {
- tree[i].setName("");
- vector<string> tempGroups;
- tree[i].setGroup(tempGroups);
- }
- }
-
- //build tree from matrix
- //initialize indexes
- map<int, int> indexes; //maps row in simMatrix to vector index in the tree
- for (int g = 0; g < numLeaves; g++) { indexes[g] = g; }
-
- //do merges and create tree structure by setting parents and children
- //there are numGroups - 1 merges to do
- for (int i = 0; i < (numLeaves - 1); i++) {
- float largest = -1000.0;
-
- if (m->control_pressed) { break; }
-
- int row, column;
- //find largest value in sims matrix by searching lower triangle
- for (int j = 1; j < sims.size(); j++) {
- for (int k = 0; k < j; k++) {
- if (sims[j][k] > largest) { largest = sims[j][k]; row = j; column = k; }
- }
- }
-
- //set non-leaf node info and update leaves to know their parents
- //non-leaf
- tree[numLeaves + i].setChildren(indexes[row], indexes[column]);
-
- //parents
- tree[indexes[row]].setParent(numLeaves + i);
- tree[indexes[column]].setParent(numLeaves + i);
-
- //blength = distance / 2;
- float blength = ((1.0 - largest) / 2);
-
- //branchlengths
- tree[indexes[row]].setBranchLength(blength - tree[indexes[row]].getLengthToLeaves());
- tree[indexes[column]].setBranchLength(blength - tree[indexes[column]].getLengthToLeaves());
-
- //set your length to leaves to your childs length plus branchlength
- tree[numLeaves + i].setLengthToLeaves(tree[indexes[row]].getLengthToLeaves() + tree[indexes[row]].getBranchLength());
-
-
- //update index
- indexes[row] = numLeaves+i;
- indexes[column] = numLeaves+i;
-
- //remove highest value that caused the merge.
- sims[row][column] = -1000.0;
- sims[column][row] = -1000.0;
-
- //merge values in simsMatrix
- for (int n = 0; n < sims.size(); n++) {
- //row becomes merge of 2 groups
- sims[row][n] = (sims[row][n] + sims[column][n]) / 2;
- sims[n][row] = sims[row][n];
- //delete column
- sims[column][n] = -1000.0;
- sims[n][column] = -1000.0;
- }
- }
-
- //adjust tree to make sure root to tip length is .5
- int root = findRoot();
- tree[root].setBranchLength((0.5 - tree[root].getLengthToLeaves()));
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "Tree");
- exit(1);
- }
-}
-/*****************************************************************/
-Tree::~Tree() {}
-/*****************************************************************/
-void Tree::addNamesToCounts(map<string, string> nameMap) {
- try {
- //ex. seq1 seq2,seq3,se4
- // seq1 = pasture
- // seq2 = forest
- // seq4 = pasture
- // seq3 = ocean
-
- //before this function seq1.pcount = pasture -> 1
- //after seq1.pcount = pasture -> 2, forest -> 1, ocean -> 1
-
- //before this function seq1.pgroups = pasture -> 1
- //after seq1.pgroups = pasture -> 1 since that is the dominant group
-
-
- //go through each leaf and update its pcounts and pgroups
-
- //float A = clock();
-
- for (int i = 0; i < numLeaves; i++) {
-
- string name = tree[i].getName();
-
- map<string, string>::iterator itNames = nameMap.find(name);
-
- if (itNames == nameMap.end()) { m->mothurOut(name + " is not in your name file, please correct."); m->mothurOutEndLine(); exit(1); }
- else {
- vector<string> dupNames;
- m->splitAtComma(nameMap[name], dupNames);
-
- map<string, int>::iterator itCounts;
- int maxPars = 1;
- set<string> groupsAddedForThisNode;
- for (int j = 0; j < dupNames.size(); j++) {
-
- string group = tmap->getGroup(dupNames[j]);
-
- if (dupNames[j] != name) {//you already added yourself in the constructor
-
- 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
-
- //update pcounts
- itCounts = tree[i].pcount.find(group);
- if (itCounts == tree[i].pcount.end()) { //new group, add it
- tree[i].pcount[group] = 1;
- }else {
- tree[i].pcount[group]++;
- }
-
- //update pgroups
- itCounts = tree[i].pGroups.find(group);
- if (itCounts == tree[i].pGroups.end()) { //new group, add it
- tree[i].pGroups[group] = 1;
- }else{
- tree[i].pGroups[group]++;
- }
-
- //keep highest group
- if(tree[i].pGroups[group] > maxPars){
- maxPars = tree[i].pGroups[group];
- }
- }else { groupsAddedForThisNode.insert(group); } //add it so you don't add it to groupNodeInfo again
- }//end for
-
- if (maxPars > 1) { //then we have some more dominant groups
- //erase all the groups that are less than maxPars because you found a more dominant group.
- for(it=tree[i].pGroups.begin();it!=tree[i].pGroups.end();){
- if(it->second < maxPars){
- tree[i].pGroups.erase(it++);
- }else { it++; }
- }
- //set one remaining groups to 1
- for(it=tree[i].pGroups.begin();it!=tree[i].pGroups.end();it++){
- tree[i].pGroups[it->first] = 1;
- }
- }//end if
-
- //update groups to reflect all the groups this node represents
- vector<string> nodeGroups;
- map<string, int>::iterator itGroups;
- for (itGroups = tree[i].pcount.begin(); itGroups != tree[i].pcount.end(); itGroups++) {
- nodeGroups.push_back(itGroups->first);
- }
- tree[i].setGroup(nodeGroups);
-
- }//end else
- }//end for
-
- //float B = clock();
- //cout << "addNamesToCounts\t" << (B - A) / CLOCKS_PER_SEC << endl;
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "addNamesToCounts");
- exit(1);
- }
-}
-/*****************************************************************/
-int Tree::getIndex(string searchName) {
- try {
- //Treemap knows name, group and index to speed up search
- // getIndex function will return the vector index or -1 if seq is not found.
- int index = tmap->getIndex(searchName);
- return index;
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "getIndex");
- exit(1);
- }
-}
-/*****************************************************************/
-
-void Tree::setIndex(string searchName, int index) {
- try {
- //set index in treemap
- tmap->setIndex(searchName, index);
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "setIndex");
- exit(1);
- }
-}
-/*****************************************************************/
-int Tree::assembleTree(map<string, string> nameMap) {
- try {
- //save for later
- names = nameMap;
-
- //if user has given a names file we want to include that info in the pgroups and pcount info.
- if(nameMap.size() != 0) { addNamesToCounts(nameMap); }
-
- //build the pGroups in non leaf nodes to be used in the parsimony calcs.
- for (int i = numLeaves; i < numNodes; i++) {
- if (m->control_pressed) { return 1; }
-
- tree[i].pGroups = (mergeGroups(i));
- tree[i].pcount = (mergeGcounts(i));
- }
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "assembleTree");
- exit(1);
- }
-}
-/*****************************************************************
-int Tree::assembleTree(string n) {
- try {
-
- //build the pGroups in non leaf nodes to be used in the parsimony calcs.
- for (int i = numLeaves; i < numNodes; i++) {
- if (m->control_pressed) { return 1; }
-
- tree[i].pGroups = (mergeGroups(i));
- tree[i].pcount = (mergeGcounts(i));
- }
- //float B = clock();
- //cout << "assembleTree\t" << (B-A) / CLOCKS_PER_SEC << endl;
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "assembleTree");
- exit(1);
- }
-}
-/*****************************************************************/
-//assumes leaf node names are in groups and no names file - used by indicator command
-void Tree::getSubTree(Tree* Ctree, vector<string> Groups) {
- try {
-
- //copy Tree since we are going to destroy it
- Tree* copy = new Tree(tmap);
- copy->getCopy(Ctree);
- map<string, string> empty;
- copy->assembleTree(empty);
-
- //we want to select some of the leaf nodes to create the output tree
- //go through the input Tree starting at parents of leaves
- for (int i = 0; i < numNodes; i++) {
-
- //initialize leaf nodes
- if (i <= (numLeaves-1)) {
- tree[i].setName(Groups[i]);
-
- //save group info
- string group = tmap->getGroup(Groups[i]);
- vector<string> tempGroups; tempGroups.push_back(group);
- tree[i].setGroup(tempGroups);
- groupNodeInfo[group].push_back(i);
-
- //set pcount and pGroup for groupname to 1.
- tree[i].pcount[group] = 1;
- tree[i].pGroups[group] = 1;
-
- //Treemap knows name, group and index to speed up search
- tmap->setIndex(Groups[i], i);
-
- //intialize non leaf nodes
- }else if (i > (numLeaves-1)) {
- tree[i].setName("");
- vector<string> tempGroups;
- tree[i].setGroup(tempGroups);
- }
- }
-
- set<int> removedLeaves;
- for (int i = 0; i < copy->getNumLeaves(); i++) {
-
- if (removedLeaves.count(i) == 0) {
-
- //am I in the group
- int parent = copy->tree[i].getParent();
-
- if (parent != -1) {
-
- if (m->inUsersGroups(copy->tree[i].getName(), Groups)) {
- //find my siblings name
- int parentRC = copy->tree[parent].getRChild();
- int parentLC = copy->tree[parent].getLChild();
-
- //if I am the right child, then my sib is the left child
- int sibIndex = parentRC;
- if (parentRC == i) { sibIndex = parentLC; }
-
- string sibsName = copy->tree[sibIndex].getName();
-
- //if yes, is my sibling
- if ((m->inUsersGroups(sibsName, Groups)) || (sibsName == "")) {
- //we both are okay no trimming required
- }else{
- //i am, my sib is not, so remove sib by setting my parent to my grandparent
- int grandparent = copy->tree[parent].getParent();
- int grandparentLC = copy->tree[grandparent].getLChild();
- int grandparentRC = copy->tree[grandparent].getRChild();
-
- //whichever of my granparents children was my parent now equals me
- if (grandparentLC == parent) { grandparentLC = i; }
- else { grandparentRC = i; }
-
- copy->tree[i].setParent(grandparent);
- copy->tree[i].setBranchLength((copy->tree[i].getBranchLength()+copy->tree[parent].getBranchLength()));
- if (grandparent != -1) {
- copy->tree[grandparent].setChildren(grandparentLC, grandparentRC);
- }
- removedLeaves.insert(sibIndex);
- }
- }else{
- //find my siblings name
- int parentRC = copy->tree[parent].getRChild();
- int parentLC = copy->tree[parent].getLChild();
-
- //if I am the right child, then my sib is the left child
- int sibIndex = parentRC;
- if (parentRC == i) { sibIndex = parentLC; }
-
- string sibsName = copy->tree[sibIndex].getName();
-
- //if no is my sibling
- if ((m->inUsersGroups(sibsName, Groups)) || (sibsName == "")) {
- //i am not, but my sib is
- int grandparent = copy->tree[parent].getParent();
- int grandparentLC = copy->tree[grandparent].getLChild();
- int grandparentRC = copy->tree[grandparent].getRChild();
-
- //whichever of my granparents children was my parent now equals my sib
- if (grandparentLC == parent) { grandparentLC = sibIndex; }
- else { grandparentRC = sibIndex; }
-
- copy->tree[sibIndex].setParent(grandparent);
- copy->tree[sibIndex].setBranchLength((copy->tree[sibIndex].getBranchLength()+copy->tree[parent].getBranchLength()));
- if (grandparent != -1) {
- copy->tree[grandparent].setChildren(grandparentLC, grandparentRC);
- }
- removedLeaves.insert(i);
- }else{
- //neither of us are, so we want to eliminate ourselves and our parent
- //so set our parents sib to our great-grandparent
- int parent = copy->tree[i].getParent();
- int grandparent = copy->tree[parent].getParent();
- int parentsSibIndex;
- if (grandparent != -1) {
- int greatgrandparent = copy->tree[grandparent].getParent();
- int greatgrandparentLC, greatgrandparentRC;
- if (greatgrandparent != -1) {
- greatgrandparentLC = copy->tree[greatgrandparent].getLChild();
- greatgrandparentRC = copy->tree[greatgrandparent].getRChild();
- }
-
- int grandparentLC = copy->tree[grandparent].getLChild();
- int grandparentRC = copy->tree[grandparent].getRChild();
-
- parentsSibIndex = grandparentLC;
- if (grandparentLC == parent) { parentsSibIndex = grandparentRC; }
-
- //whichever of my greatgrandparents children was my grandparent
- if (greatgrandparentLC == grandparent) { greatgrandparentLC = parentsSibIndex; }
- else { greatgrandparentRC = parentsSibIndex; }
-
- copy->tree[parentsSibIndex].setParent(greatgrandparent);
- copy->tree[parentsSibIndex].setBranchLength((copy->tree[parentsSibIndex].getBranchLength()+copy->tree[grandparent].getBranchLength()));
- if (greatgrandparent != -1) {
- copy->tree[greatgrandparent].setChildren(greatgrandparentLC, greatgrandparentRC);
- }
- }else{
- copy->tree[parent].setParent(-1);
- //cout << "issues with making subtree" << endl;
- }
- removedLeaves.insert(sibIndex);
- removedLeaves.insert(i);
- }
- }
- }
- }
- }
-
- int root = 0;
- for (int i = 0; i < copy->getNumNodes(); i++) {
- //you found the root
- if (copy->tree[i].getParent() == -1) { root = i; break; }
- }
-
- int nextSpot = numLeaves;
- populateNewTree(copy->tree, root, nextSpot);
-
- delete copy;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "getSubTree");
- exit(1);
- }
-}
-/*****************************************************************/
-//assumes nameMap contains unique names as key or is empty.
-//assumes numLeaves defined in tree constructor equals size of seqsToInclude and seqsToInclude only contains unique seqs.
-int Tree::getSubTree(Tree* copy, vector<string> seqsToInclude, map<string, string> nameMap) {
- try {
-
- if (numLeaves != seqsToInclude.size()) { m->mothurOut("[ERROR]: numLeaves does not equal numUniques, cannot create subtree.\n"); m->control_pressed = true; return 0; }
-
- getSubTree(copy, seqsToInclude);
- if (nameMap.size() != 0) { addNamesToCounts(nameMap); }
-
- //build the pGroups in non leaf nodes to be used in the parsimony calcs.
- for (int i = numLeaves; i < numNodes; i++) {
- if (m->control_pressed) { return 1; }
-
- tree[i].pGroups = (mergeGroups(i));
- tree[i].pcount = (mergeGcounts(i));
- }
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "getSubTree");
- exit(1);
- }
-}
-/*****************************************************************/
-int Tree::populateNewTree(vector<Node>& oldtree, int node, int& index) {
- try {
-
- if (oldtree[node].getLChild() != -1) {
- int rc = populateNewTree(oldtree, oldtree[node].getLChild(), index);
- int lc = populateNewTree(oldtree, oldtree[node].getRChild(), index);
-
- tree[index].setChildren(lc, rc);
- tree[rc].setParent(index);
- tree[lc].setParent(index);
-
- tree[index].setBranchLength(oldtree[node].getBranchLength());
- tree[rc].setBranchLength(oldtree[oldtree[node].getLChild()].getBranchLength());
- tree[lc].setBranchLength(oldtree[oldtree[node].getRChild()].getBranchLength());
-
- return (index++);
- }else { //you are a leaf
- int indexInNewTree = tmap->getIndex(oldtree[node].getName());
- return indexInNewTree;
- }
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "populateNewTree");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::getCopy(Tree* copy) {
- try {
-
- //for each node in the tree copy its info
- for (int i = 0; i < numNodes; i++) {
- //copy name
- tree[i].setName(copy->tree[i].getName());
-
- //copy group
- tree[i].setGroup(copy->tree[i].getGroup());
-
- //copy branch length
- tree[i].setBranchLength(copy->tree[i].getBranchLength());
-
- //copy parent
- tree[i].setParent(copy->tree[i].getParent());
-
- //copy children
- tree[i].setChildren(copy->tree[i].getLChild(), copy->tree[i].getRChild());
-
- //copy index in node and tmap
- tree[i].setIndex(copy->tree[i].getIndex());
- setIndex(copy->tree[i].getName(), getIndex(copy->tree[i].getName()));
-
- //copy pGroups
- tree[i].pGroups = copy->tree[i].pGroups;
-
- //copy pcount
- tree[i].pcount = copy->tree[i].pcount;
- }
-
- groupNodeInfo = copy->groupNodeInfo;
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "getCopy");
- exit(1);
- }
-}
-/*****************************************************************/
-//returns a map with a groupname and the number of times that group was seen in the children
-//for instance if your children are white and black then it would return a map with 2 entries
-// p[white] = 1 and p[black] = 1. Now go up a level and merge that with a node who has p[white] = 1
-//and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
-
-map<string, int> Tree::mergeGroups(int i) {
- try {
- int lc = tree[i].getLChild();
- int rc = tree[i].getRChild();
-
- //set parsimony groups to left child
- map<string,int> parsimony = tree[lc].pGroups;
-
- int maxPars = 1;
-
- //look at right child groups and update maxPars if right child has something higher for that group.
- for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
- it2 = parsimony.find(it->first);
- if (it2 != parsimony.end()) {
- parsimony[it->first]++;
- }else {
- parsimony[it->first] = 1;
- }
-
- if(parsimony[it->first] > maxPars){
- maxPars = parsimony[it->first];
- }
- }
-
- // this is true if right child had a greater parsimony for a certain group
- if(maxPars > 1){
- //erase all the groups that are only 1 because you found something with 2.
- for(it=parsimony.begin();it!=parsimony.end();){
- if(it->second == 1){
- parsimony.erase(it++);
- }else { it++; }
- }
- //set one remaining groups to 1
- //so with our above example p[white] = 2 would be left and it would become p[white] = 1
- for(it=parsimony.begin();it!=parsimony.end();it++){
- parsimony[it->first] = 1;
- }
-
- }
-
- return parsimony;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "mergeGroups");
- exit(1);
- }
-}
-/*****************************************************************/
-//returns a map with a groupname and the number of times that group was seen in the children
-//for instance if your children are white and black then it would return a map with 2 entries
-// p[white] = 1 and p[black] = 1. Now go up a level and merge that with a node who has p[white] = 1
-//and you get p[white] = 2, p[black] = 1, but you erase the p[black] because you have a p value higher than 1.
-
-map<string, int> Tree::mergeUserGroups(int i, vector<string> g) {
- try {
-
- int lc = tree[i].getLChild();
- int rc = tree[i].getRChild();
-
- //loop through nodes groups removing the ones the user doesn't want
- for(it=tree[lc].pGroups.begin();it!=tree[lc].pGroups.end();){
- if (m->inUsersGroups(it->first, g) != true) {
- tree[lc].pGroups.erase(it++);
- }else { it++; }
- }
-
- //loop through nodes groups removing the ones the user doesn't want
- for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();){
- if (m->inUsersGroups(it->first, g) != true) {
- tree[rc].pGroups.erase(it++);
- }else { it++; }
- }
-
- //set parsimony groups to left child
- map<string,int> parsimony = tree[lc].pGroups;
-
- int maxPars = 1;
-
- //look at right child groups and update maxPars if right child has something higher for that group.
- for(it=tree[rc].pGroups.begin();it!=tree[rc].pGroups.end();it++){
- it2 = parsimony.find(it->first);
- if (it2 != parsimony.end()) {
- parsimony[it->first]++;
- }else {
- parsimony[it->first] = 1;
- }
-
- if(parsimony[it->first] > maxPars){
- maxPars = parsimony[it->first];
- }
- }
-
- // this is true if right child had a greater parsimony for a certain group
- if(maxPars > 1){
- //erase all the groups that are only 1 because you found something with 2.
- for(it=parsimony.begin();it!=parsimony.end();){
- if(it->second == 1){
- parsimony.erase(it++);
- }else { it++; }
- }
-
- for(it=parsimony.begin();it!=parsimony.end();it++){
- parsimony[it->first] = 1;
- }
- }
-
- return parsimony;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "mergeUserGroups");
- exit(1);
- }
-}
-
-
-/**************************************************************************************************/
-
-map<string,int> Tree::mergeGcounts(int position) {
- try{
- map<string,int>::iterator pos;
-
- int lc = tree[position].getLChild();
- int rc = tree[position].getRChild();
-
- map<string,int> sum = tree[lc].pcount;
-
- for(it=tree[rc].pcount.begin();it!=tree[rc].pcount.end();it++){
- sum[it->first] += it->second;
- }
- return sum;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "mergeGcounts");
- exit(1);
- }
-}
-/**************************************************************************************************/
-void Tree::randomLabels(vector<string> g) {
- try {
-
- //initialize groupNodeInfo
- for (int i = 0; i < (tmap->getNamesOfGroups()).size(); i++) {
- groupNodeInfo[(tmap->getNamesOfGroups())[i]].resize(0);
- }
-
- for(int i = 0; i < numLeaves; i++){
- int z;
- //get random index to switch with
- z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
-
- //you only want to randomize the nodes that are from a group the user wants analyzed, so
- //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.
- bool treez, treei;
-
- treez = m->inUsersGroups(tree[z].getGroup(), g);
- treei = m->inUsersGroups(tree[i].getGroup(), g);
-
- if ((treez == true) && (treei == true)) {
- //switches node i and node z's info.
- map<string,int> lib_hold = tree[z].pGroups;
- tree[z].pGroups = (tree[i].pGroups);
- tree[i].pGroups = (lib_hold);
-
- vector<string> zgroup = tree[z].getGroup();
- tree[z].setGroup(tree[i].getGroup());
- tree[i].setGroup(zgroup);
-
- string zname = tree[z].getName();
- tree[z].setName(tree[i].getName());
- tree[i].setName(zname);
-
- map<string,int> gcount_hold = tree[z].pcount;
- tree[z].pcount = (tree[i].pcount);
- tree[i].pcount = (gcount_hold);
- }
-
- for (int k = 0; k < (tree[i].getGroup()).size(); k++) { groupNodeInfo[(tree[i].getGroup())[k]].push_back(i); }
- for (int k = 0; k < (tree[z].getGroup()).size(); k++) { groupNodeInfo[(tree[z].getGroup())[k]].push_back(z); }
- }
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "randomLabels");
- exit(1);
- }
-}
-/**************************************************************************************************/
-void Tree::randomBlengths() {
- try {
- for(int i=numNodes-1;i>=0;i--){
- int z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
-
- float bl_hold = tree[z].getBranchLength();
- tree[z].setBranchLength(tree[i].getBranchLength());
- tree[i].setBranchLength(bl_hold);
- }
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "randomBlengths");
- exit(1);
- }
-}
-/*************************************************************************************************/
-void Tree::assembleRandomUnifracTree(vector<string> g) {
- randomLabels(g);
- map<string, string> empty;
- assembleTree(empty);
-}
-/*************************************************************************************************/
-void Tree::assembleRandomUnifracTree(string groupA, string groupB) {
- vector<string> temp; temp.push_back(groupA); temp.push_back(groupB);
- randomLabels(temp);
- map<string, string> empty;
- assembleTree(empty);
-}
-
-/*************************************************************************************************/
-//for now it's just random topology but may become random labels as well later that why this is such a simple function now...
-void Tree::assembleRandomTree() {
- randomTopology();
- map<string, string> empty;
- assembleTree(empty);
-}
-/**************************************************************************************************/
-
-void Tree::randomTopology() {
- try {
- for(int i=0;i<numNodes;i++){
- tree[i].setParent(-1);
- }
- for(int i=numLeaves;i<numNodes;i++){
- tree[i].setChildren(-1, -1);
- }
-
- for(int i=numLeaves;i<numNodes;i++){
- int escape =0;
- int rnd_index1, rnd_index2;
- while(escape == 0){
- rnd_index1 = (int)(((double)rand() / (double) RAND_MAX)*i);
- if(tree[rnd_index1].getParent() == -1){escape = 1;}
- }
-
- escape = 0;
- while(escape == 0){
- rnd_index2 = (int)(((double)rand() / (double) RAND_MAX)*i);
- if(rnd_index2 != rnd_index1 && tree[rnd_index2].getParent() == -1){
- escape = 1;
- }
- }
-
- tree[i].setChildren(rnd_index1,rnd_index2);
- tree[i].setParent(-1);
- tree[rnd_index1].setParent(i);
- tree[rnd_index2].setParent(i);
- }
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "randomTopology");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::print(ostream& out) {
- try {
- int root = findRoot();
- printBranch(root, out, "branch");
- out << ";" << endl;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "print");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::print(ostream& out, map<string, string> nameMap) {
- try {
- int root = findRoot();
- printBranch(root, out, nameMap);
- out << ";" << endl;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "print");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::print(ostream& out, string mode) {
- try {
- int root = findRoot();
- printBranch(root, out, mode);
- out << ";" << endl;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "print");
- exit(1);
- }
-}
-/*****************************************************************/
-// This prints out the tree in Newick form.
-void Tree::createNewickFile(string f) {
- try {
- int root = findRoot();
-
- filename = f;
-
- m->openOutputFile(filename, out);
-
- printBranch(root, out, "branch");
-
- // you are at the end of the tree
- out << ";" << endl;
- out.close();
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "createNewickFile");
- exit(1);
- }
-}
-
-/*****************************************************************/
-//This function finds the index of the root node.
-
-int Tree::findRoot() {
- try {
- for (int i = 0; i < numNodes; i++) {
- //you found the root
- if (tree[i].getParent() == -1) { return i; }
- //cout << "i = " << i << endl;
- //cout << "i's parent = " << tree[i].getParent() << endl;
- }
- return -1;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "findRoot");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::printBranch(int node, ostream& out, map<string, string> names) {
-try {
-
-// you are not a leaf
- if (tree[node].getLChild() != -1) {
- out << "(";
- printBranch(tree[node].getLChild(), out, names);
- out << ",";
- printBranch(tree[node].getRChild(), out, names);
- out << ")";
-
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- out << ":" << tree[node].getBranchLength();
- }
-
- }else { //you are a leaf
- map<string, string>::iterator itNames = names.find(tree[node].getName());
-
- string outputString = "";
- if (itNames != names.end()) {
-
- vector<string> dupNames;
- m->splitAtComma((itNames->second), dupNames);
-
- if (dupNames.size() == 1) {
- outputString += tree[node].getName();
- if (tree[node].getBranchLength() != -1) {
- outputString += ":" + toString(tree[node].getBranchLength());
- }
- }else {
- outputString += "(";
-
- for (int u = 0; u < dupNames.size()-1; u++) {
- outputString += dupNames[u];
-
- if (tree[node].getBranchLength() != -1) {
- outputString += ":" + toString(0.0);
- }
- outputString += ",";
- }
-
- outputString += dupNames[dupNames.size()-1];
- if (tree[node].getBranchLength() != -1) {
- outputString += ":" + toString(0.0);
- }
-
- outputString += ")";
- if (tree[node].getBranchLength() != -1) {
- outputString += ":" + toString(tree[node].getBranchLength());
- }
- }
- }else {
- outputString = tree[node].getName();
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- outputString += ":" + toString(tree[node].getBranchLength());
- }
-
- m->mothurOut("[ERROR]: " + tree[node].getName() + " is not in your namefile, please correct."); m->mothurOutEndLine();
- }
-
- out << outputString;
- }
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "printBranch");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::printBranch(int node, ostream& out, string mode) {
- try {
-
- // you are not a leaf
- if (tree[node].getLChild() != -1) {
- out << "(";
- printBranch(tree[node].getLChild(), out, mode);
- out << ",";
- printBranch(tree[node].getRChild(), out, mode);
- out << ")";
- if (mode == "branch") {
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- out << ":" << tree[node].getBranchLength();
- }
- }else if (mode == "boot") {
- //if there is a label then print it
- if (tree[node].getLabel() != -1) {
- out << tree[node].getLabel();
- }
- }else if (mode == "both") {
- if (tree[node].getLabel() != -1) {
- out << tree[node].getLabel();
- }
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- out << ":" << tree[node].getBranchLength();
- }
- }
- }else { //you are a leaf
- string leafGroup = tmap->getGroup(tree[node].getName());
-
- if (mode == "branch") {
- out << leafGroup;
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- out << ":" << tree[node].getBranchLength();
- }
- }else if (mode == "boot") {
- out << leafGroup;
- //if there is a label then print it
- if (tree[node].getLabel() != -1) {
- out << tree[node].getLabel();
- }
- }else if (mode == "both") {
- out << tree[node].getName();
- if (tree[node].getLabel() != -1) {
- out << tree[node].getLabel();
- }
- //if there is a branch length then print it
- if (tree[node].getBranchLength() != -1) {
- out << ":" << tree[node].getBranchLength();
- }
- }
- }
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "printBranch");
- exit(1);
- }
-}
-/*****************************************************************/
-void Tree::printBranch(int node, ostream& out, string mode, vector<Node>& theseNodes) {
- try {
-
- // you are not a leaf
- if (theseNodes[node].getLChild() != -1) {
- out << "(";
- printBranch(theseNodes[node].getLChild(), out, mode);
- out << ",";
- printBranch(theseNodes[node].getRChild(), out, mode);
- out << ")";
- if (mode == "branch") {
- //if there is a branch length then print it
- if (theseNodes[node].getBranchLength() != -1) {
- out << ":" << theseNodes[node].getBranchLength();
- }
- }else if (mode == "boot") {
- //if there is a label then print it
- if (theseNodes[node].getLabel() != -1) {
- out << theseNodes[node].getLabel();
- }
- }else if (mode == "both") {
- if (theseNodes[node].getLabel() != -1) {
- out << theseNodes[node].getLabel();
- }
- //if there is a branch length then print it
- if (theseNodes[node].getBranchLength() != -1) {
- out << ":" << theseNodes[node].getBranchLength();
- }
- }
- }else { //you are a leaf
- string leafGroup = tmap->getGroup(theseNodes[node].getName());
-
- if (mode == "branch") {
- out << leafGroup;
- //if there is a branch length then print it
- if (theseNodes[node].getBranchLength() != -1) {
- out << ":" << theseNodes[node].getBranchLength();
- }
- }else if (mode == "boot") {
- out << leafGroup;
- //if there is a label then print it
- if (theseNodes[node].getLabel() != -1) {
- out << theseNodes[node].getLabel();
- }
- }else if (mode == "both") {
- out << theseNodes[node].getName();
- if (theseNodes[node].getLabel() != -1) {
- out << theseNodes[node].getLabel();
- }
- //if there is a branch length then print it
- if (theseNodes[node].getBranchLength() != -1) {
- out << ":" << theseNodes[node].getBranchLength();
- }
- }
- }
-
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "printBranch");
- exit(1);
- }
-}
-/*****************************************************************/
-
-void Tree::printTree() {
-
- for(int i=0;i<numNodes;i++){
- cout << i << '\t';
- tree[i].printNode();
- }
-
-}
-
-/*****************************************************************/
-//this code is a mess and should be rethought...-slw
-void Tree::parseTreeFile() {
-
- //only takes names from the first tree and assumes that all trees use the same names.
- try {
- string filename = m->getTreeFile();
- ifstream filehandle;
- m->openInputFile(filename, filehandle);
- int c, comment;
- comment = 0;
- int done = 1;
-
- //ifyou are not a nexus file
- if((c = filehandle.peek()) != '#') {
- while((c = filehandle.peek()) != ';') {
- while ((c = filehandle.peek()) != ';') {
- // get past comments
- if(c == '[') {
- comment = 1;
- }
- if(c == ']'){
- comment = 0;
- }
- if((c == '(') && (comment != 1)){ break; }
- filehandle.get();
- }
-
- done = readTreeString(filehandle);
- if (done == 0) { break; }
- }
- //ifyou are a nexus file
- }else if((c = filehandle.peek()) == '#') {
- string holder = "";
-
- // get past comments
- while(holder != "translate" && holder != "Translate"){
- if(holder == "[" || holder == "[!"){
- comment = 1;
- }
- if(holder == "]"){
- comment = 0;
- }
- filehandle >> holder;
-
- //if there is no translate then you must read tree string otherwise use translate to get names
- if((holder == "tree") && (comment != 1)){
- //pass over the "tree rep.6878900 = "
- while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF)) {;}
-
- if(c == EOF) { break; }
- filehandle.putback(c); //put back first ( of tree.
- done = readTreeString(filehandle);
-
- break;
- }
-
- if (done == 0) { break; }
- }
-
- //use nexus translation rather than parsing tree to save time
- if((holder == "translate") || (holder == "Translate")) {
-
- string number, name, h;
- h = ""; // so it enters the loop the first time
- while((h != ";") && (number != ";")) {
- filehandle >> number;
- filehandle >> name;
-
- //c = , until done with translation then c = ;
- h = name.substr(name.length()-1, name.length());
- name.erase(name.end()-1); //erase the comma
- m->Treenames.push_back(number);
- }
- if(number == ";") { m->Treenames.pop_back(); } //in case ';' from translation is on next line instead of next to last name
- }
- }
- filehandle.close();
-
- //for (int i = 0; i < globaldata->Treenames.size(); i++) {
-//cout << globaldata->Treenames[i] << endl; }
-//cout << globaldata->Treenames.size() << endl;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "parseTreeFile");
- exit(1);
- }
-}
-/*******************************************************/
-
-/*******************************************************/
-int Tree::readTreeString(ifstream& filehandle) {
- try {
- int c;
- string name; //, k
-
- while((c = filehandle.peek()) != ';') {
-//k = c;
-//cout << " at beginning of while " << k << endl;
- if(c == ')') {
- //to pass over labels in trees
- c=filehandle.get();
- while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); }
- filehandle.putback(c);
- }
- if(c == ';') { return 0; }
- if(c == -1) { return 0; }
- //if you are a name
- if((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != '\t') && (c != 32)) { //32 is space
- name = "";
- c = filehandle.get();
- //k = c;
-//cout << k << endl;
- while ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != 32) && (c != '\t')) {
- name += c;
- c = filehandle.get();
- //k = c;
-//cout << " in name while " << k << endl;
- }
-
-//cout << "name = " << name << endl;
- m->Treenames.push_back(name);
- filehandle.putback(c);
-//k = c;
-//cout << " after putback" << k << endl;
- }
-
- if(c == ':') { //read until you reach the end of the branch length
- while ((c != '(') && (c != ')') && (c != ',') && (c != ';') && (c != '\n') && (c != '\t') && (c != 32)) {
- c = filehandle.get();
- //k = c;
- //cout << " in branch while " << k << endl;
- }
- filehandle.putback(c);
- }
-
- c = filehandle.get();
-//k = c;
- //cout << " here after get " << k << endl;
- if(c == ';') { return 0; }
- if(c == ')') { filehandle.putback(c); }
- //k = c;
-//cout << k << endl;
-
- }
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "Tree", "readTreeString");
- exit(1);
- }
-}
-
-/*******************************************************/
-
-/*******************************************************/
-