X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=tree.cpp;h=ee17e6f37f6417d24546697e3247c21f34d1d0aa;hb=99bd02d78eaf21b01420c6f99e38efc7fa4e9e17;hp=2ad46f044f71556a9726c93a96d4fcd5ae42a6b6;hpb=d97b619c4297b1274c754d73a64792ba656b0a79;p=mothur.git

diff --git a/tree.cpp b/tree.cpp
index 2ad46f0..ee17e6f 100644
--- a/tree.cpp
+++ b/tree.cpp
@@ -13,9 +13,11 @@
 /*****************************************************************/
 Tree::Tree() {
 	try {
-	
 		globaldata = GlobalData::getInstance();
-		numLeaves = globaldata->gTreemap->getNumSeqs();
+		
+		if (globaldata->runParse == true) {  parseTreeFile();  globaldata->runParse = false;  }
+		
+		numLeaves = globaldata->Treenames.size();
 		numNodes = 2*numLeaves - 1;
 		
 		tree.resize(numNodes);
@@ -24,22 +26,18 @@ Tree::Tree() {
 		for (int i = 0; i < numNodes; i++) {
 			//initialize leaf nodes
 			if (i <= (numLeaves-1)) {
-				tree[i].setName(globaldata->gTreemap->namesOfSeqs[i]);
-				tree[i].setGroup(globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i]));
-				//the node knows its index
-				tree[i].setIndex(i);
+				tree[i].setName(globaldata->Treenames[i]);
+				tree[i].setGroup(globaldata->gTreemap->getGroup(globaldata->Treenames[i]));
 				//set pcount and pGroup for groupname to 1.
-				tree[i].pcount[globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i])] = 1;
-				tree[i].pGroups[globaldata->gTreemap->getGroup(globaldata->gTreemap->namesOfSeqs[i])] = 1;
+				tree[i].pcount[globaldata->gTreemap->getGroup(globaldata->Treenames[i])] = 1;
+				tree[i].pGroups[globaldata->gTreemap->getGroup(globaldata->Treenames[i])] = 1;
 				//Treemap knows name, group and index to speed up search
-				globaldata->gTreemap->setIndex(globaldata->gTreemap->namesOfSeqs[i], i);
+				globaldata->gTreemap->setIndex(globaldata->Treenames[i], i);
 	
 			//intialize non leaf nodes
 			}else if (i > (numLeaves-1)) {
 				tree[i].setName("");
 				tree[i].setGroup("");
-				//the node knows its index
-				tree[i].setIndex(i);
 			}
 		}
 	}
@@ -54,7 +52,8 @@ Tree::Tree() {
 }
 
 /*****************************************************************/
-
+Tree::~Tree() {}
+/*****************************************************************/
 int Tree::getIndex(string searchName) {
 	try {
 		//Treemap knows name, group and index to speed up search
@@ -155,10 +154,9 @@ void Tree::getCopy(Tree* copy) {
 
 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;
 		
@@ -166,7 +164,12 @@ map<string, int> Tree::mergeGroups(int i) {
 
 		//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++){
-			parsimony[it->first]++;
+			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];
@@ -179,7 +182,7 @@ map<string, int> Tree::mergeGroups(int i) {
 			for(it=parsimony.begin();it!=parsimony.end();it++){
 				if(it->second == 1){
 					parsimony.erase(it->first);
-					it--;
+//					it--;
 				}
 			}
 			//set one remaining groups to 1
@@ -201,6 +204,72 @@ map<string, int> Tree::mergeGroups(int i) {
 		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(); it++) {
+			if (inUsersGroups(it->first, g) != true) { tree[lc].pGroups.erase(it->first); }
+		}
+		
+		//loop through nodes groups removing the ones the user doesn't want
+		for (it = tree[rc].pGroups.begin(); it != tree[rc].pGroups.end(); it++) {
+			if (inUsersGroups(it->first, g) != true) { tree[rc].pGroups.erase(it->first); }
+		}
+
+		//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();it++){
+				if(it->second == 1){
+					parsimony.erase(it->first);
+				}
+			}
+			for(it=parsimony.begin();it!=parsimony.end();it++){
+				parsimony[it->first] = 1;
+			}
+		}		
+		
+		return parsimony;
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the Tree class function mergeGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
+
 
 /**************************************************************************************************/
 
@@ -229,49 +298,35 @@ map<string,int> Tree::mergeGcounts(int position) {
 }
 /**************************************************************************************************/
 
-void Tree::randomLabels() {
+void Tree::randomLabels(vector<string> g) {
 	try {
 		
-		//set up the groups the user wants to include
-		setGroups();
-		
-		for(int i=numLeaves-1;i>=0;i--){
-			if(tree[i].pGroups.size() == 0){
-				continue;
-			}
-			
-			int escape = 1;
+		for(int i = 0; i < numLeaves; i++){
 			int z;
-		
-			while(escape == 1){
-				//get random index to switch with
-				z = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));	
-			
-				if(tree[z].pGroups.size() != 0){
-					escape = 0;
-				}
-			}
+			//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;
-			
-			//leaves have only one group so you can just set it to begin()
-			it = tree[z].pGroups.begin();
-			treez = inUsersGroups(it->first, globaldata->Groups);
-			
-			it = tree[i].pGroups.begin();
-			treei = inUsersGroups(it->first, globaldata->Groups);
+		
+			treez = inUsersGroups(tree[z].getGroup(), g);
+			treei = 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);
-		
-				tree[z].setGroup(tree[z].pGroups.begin()->first);
-				tree[i].setGroup(tree[i].pGroups.begin()->first);
-		
+				
+				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);
@@ -289,6 +344,39 @@ void Tree::randomLabels() {
 }
 /**************************************************************************************************/
 
+void Tree::randomLabels(string groupA, string groupB) {
+	try {
+		int numSeqsA = globaldata->gTreemap->seqsPerGroup[groupA];
+		int numSeqsB = globaldata->gTreemap->seqsPerGroup[groupB];
+
+		vector<string> randomGroups(numSeqsA+numSeqsB, groupA);
+		for(int i=numSeqsA;i<randomGroups.size();i++){
+			randomGroups[i] = groupB;
+		}
+		random_shuffle(randomGroups.begin(), randomGroups.end());
+				
+		int randomCounter = 0;				
+		for(int i=0;i<numLeaves;i++){
+			if(tree[i].getGroup() == groupA || tree[i].getGroup() == groupB){
+				tree[i].setGroup(randomGroups[randomCounter]);
+				tree[i].pcount.clear();
+				tree[i].pcount[randomGroups[randomCounter]] = 1;
+				tree[i].pGroups.clear();
+				tree[i].pGroups[randomGroups[randomCounter]] = 1;
+				randomCounter++;
+			}
+		}
+	}		
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the Tree class function randomLabels. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
+/**************************************************************************************************/
 void Tree::randomBlengths()  {
 	try {
 		for(int i=numNodes-1;i>=0;i--){
@@ -309,8 +397,13 @@ void Tree::randomBlengths()  {
 	}		
 }
 /*************************************************************************************************/
-void Tree::assembleRandomUnifracTree() {
-	randomLabels();
+void Tree::assembleRandomUnifracTree(vector<string> g) {
+	randomLabels(g);
+	assembleTree();
+}
+/*************************************************************************************************/
+void Tree::assembleRandomUnifracTree(string groupA, string groupB) {
+	randomLabels(groupA, groupB);
 	assembleTree();
 }
 
@@ -362,19 +455,54 @@ void Tree::randomTopology() {
 		exit(1);
 	}		
 }
+/*****************************************************************/
+void Tree::print(ostream& out) {
+	try {
+		int root = findRoot();
+		printBranch(root, out, "branch");
+		out << ";" << endl;
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the Tree class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
+/*****************************************************************/
+void Tree::printForBoot(ostream& out) {
+	try {
+		int root = findRoot();
+		printBranch(root, out, "boot");
+		out << ";" << endl;
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function printForBoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the Tree class function printForBoot. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
 
 /*****************************************************************/
 // This prints out the tree in Newick form.
-void Tree::createNewickFile() {
+void Tree::createNewickFile(string f) {
 	try {
 		int root = findRoot();
-		filename = getRootName(globaldata->getTreeFile()) + "newick";
+		//filename = getRootName(globaldata->getTreeFile()) + "newick";
+		filename = f;
+
 		openOutputFile(filename, out);
 		
-		printBranch(root);
+		printBranch(root, out, "branch");
 		
 		// you are at the end of the tree
 		out << ";" << endl;
+		out.close();
 	}
 	catch(exception& e) {
 		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function createNewickFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -394,6 +522,8 @@ int Tree::findRoot() {
 		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;
 	}
@@ -408,18 +538,40 @@ int Tree::findRoot() {
 }
 
 /*****************************************************************/
-void Tree::printBranch(int node) {
+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());
+			printBranch(tree[node].getLChild(), out, mode);
 			out << ",";
-			printBranch(tree[node].getRChild());
+			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 { //you are a leaf
-			tree[node].printNode(out);  //prints out name and branch length
+			out << tree[node].getGroup(); 
+			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();
+				}
+			}
 		}
 		
 	}
@@ -434,41 +586,154 @@ void Tree::printBranch(int node) {
 }
 
 /*****************************************************************/
-void Tree::setGroups() {
+
+void Tree::printTree() {
+	
+	for(int i=0;i<numNodes;i++){
+		cout << i << '\t';
+		tree[i].printNode();
+	}
+	
+}
+
+/*****************************************************************/
+
+void Tree::parseTreeFile() {
+	
+	//only takes names from the first tree and assumes that all trees use the same names.
 	try {
-		//if the user has not entered specific groups to analyze then do them all
-		if (globaldata->Groups.size() != 0) {
-			//check that groups are valid
-			for (int i = 0; i < globaldata->Groups.size(); i++) {
-				if (globaldata->gTreemap->isValidGroup(globaldata->Groups[i]) != true) {
-					cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
-					// erase the invalid group from globaldata->Groups
-					globaldata->Groups.erase (globaldata->Groups.begin()+i);
+		string filename = globaldata->getTreeFile();
+		ifstream filehandle;
+		openInputFile(filename, filehandle);
+		int c, comment;
+		comment = 0;
+		
+		//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();
+				}
+
+				readTreeString(filehandle); 
+			}
+		//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; 
+	
+				//ifthere 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.
+					readTreeString(filehandle);	
+					break;
 				}
 			}
 			
-			//if the user only entered invalid groups
-			if (globaldata->Groups.size() == 0) { 
-				cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl; 
-				for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
-					globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
+			//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
+					globaldata->Treenames.push_back(number);
 				}
+				if(number == ";") { globaldata->Treenames.pop_back(); }  //in case ';' from translation is on next line instead of next to last name
 			}
-					
-		}else {
-			for (int i = 0; i < globaldata->gTreemap->namesOfGroups.size(); i++) {
-				globaldata->Groups.push_back(globaldata->gTreemap->namesOfGroups[i]);
+		}
+		filehandle.close();
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the Tree class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
+/*******************************************************/
+
+/*******************************************************/
+void Tree::readTreeString(ifstream& filehandle)	{
+	try {
+		int c;
+		string name; //k
+		
+		while((c = filehandle.peek()) != ';') { 
+				//ifyou 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;
+				globaldata->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();
+			if(c == ';') { break; }
+		//	k = c;
+//cout << k << endl;
+
 		}
 	}
 	catch(exception& e) {
-		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		cout << "Standard Error: " << e.what() << " has occurred in the Tree class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 		exit(1);
 	}
 	catch(...) {
-		cout << "An unknown error has occurred in the Tree class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		cout << "An unknown error has occurred in the Tree class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 		exit(1);
 	}		
+}	
 
-}
+/*******************************************************/
+
+/*******************************************************/