working on windows paralellization, added trimOligos class to be used by trim.flows...

[mothur.git] / phylotree.cpp

diff --git a/phylotree.cpp b/phylotree.cpp
index d6c740ae791b494ea64951a03405b38535593152..6ca36361dc55fe6c204017486c98ebeea77e7a9c 100644 (file)
--- a/phylotree.cpp
+++ b/phylotree.cpp
@@ -54,22 +54,25 @@ PhyloTree::PhyloTree(ifstream& in, string filename){
                        istringstream iss (tempBuf,istringstream::in);
                        delete buffer;
                        
-                       iss >> numNodes; gobble(iss);
+                       //read version
+                       m->getline(iss); m->gobble(iss);
+                       
+                       iss >> numNodes; m->gobble(iss);
                        
                        tree.resize(numNodes);
                        
                        for (int i = 0; i < tree.size(); i++) {
-                               iss >> tree[i].name >> tree[i].level >> tree[i].parent; gobble(iss);
+                               iss >> tree[i].name >> tree[i].level >> tree[i].parent; m->gobble(iss);
                        }
                        
                        //read genus nodes
                        int numGenus = 0;
-                       iss >> numGenus; gobble(iss);
+                       iss >> numGenus; m->gobble(iss);
                        
                        int gnode, gsize;
                        totals.clear();
                        for (int i = 0; i < numGenus; i++) {
-                               iss >> gnode >> gsize; gobble(iss);
+                               iss >> gnode >> gsize; m->gobble(iss);
                                
                                uniqueTaxonomies[gnode] = gnode;
                                totals.push_back(gsize);
@@ -78,22 +81,25 @@ PhyloTree::PhyloTree(ifstream& in, string filename){
                        MPI_File_close(&inMPI);
                        
                #else
-                       in >> numNodes; gobble(in);
+                       //read version
+                       string line = m->getline(in); m->gobble(in);
+                       
+                       in >> numNodes; m->gobble(in);
                        
                        tree.resize(numNodes);
                        
                        for (int i = 0; i < tree.size(); i++) {
-                               in >> tree[i].name >> tree[i].level >> tree[i].parent; gobble(in);
+                               in >> tree[i].name >> tree[i].level >> tree[i].parent; m->gobble(in);
                        }
                        
                        //read genus nodes
                        int numGenus = 0;
-                       in >> numGenus; gobble(in);
+                       in >> numGenus; m->gobble(in);
                        
                        int gnode, gsize;
                        totals.clear();
                        for (int i = 0; i < numGenus; i++) {
-                               in >> gnode >> gsize; gobble(in);
+                               in >> gnode >> gsize; m->gobble(in);
                                
                                uniqueTaxonomies[gnode] = gnode;
                                totals.push_back(gsize);
@@ -125,7 +131,7 @@ PhyloTree::PhyloTree(string tfile){
                
                #ifdef USE_MPI
                        int pid, num, processors;
-                       vector<long> positions;
+                       vector<unsigned long long> positions;
                        
                        MPI_Status status; 
                        MPI_File inMPI;
@@ -138,7 +144,7 @@ PhyloTree::PhyloTree(string tfile){
                        MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 
                        if (pid == 0) {
-                               positions = setFilePosEachLine(tfile, num);
+                               positions = m->setFilePosEachLine(tfile, num);
                                
                                //send file positions to all processes
                                for(int i = 1; i < processors; i++) { 
@@ -173,19 +179,19 @@ PhyloTree::PhyloTree(string tfile){
                
                #else
                        ifstream in;
-                       openInputFile(tfile, in);
+                       m->openInputFile(tfile, in);
                        
                        //read in users taxonomy file and add sequences to tree
                        while(!in.eof()){
-                               in >> name >> tax; gobble(in);
-                               
+                               in >> name >> tax; m->gobble(in);
+                       
                                addSeqToTree(name, tax);
                        }
                        in.close();
                #endif
-               
+       
                assignHeirarchyIDs(0);
-               
+       
                //create file for summary if needed
                setUp(tfile);
        }
@@ -269,6 +275,8 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
        
                        if (seqTaxonomy == "") {   uniqueTaxonomies[currentNode] = currentNode; }
                }
+               
+               return 0;
        }
        catch(exception& e) {
                m->errorOut(e, "PhyloTree", "addSeqToTree");
@@ -346,12 +354,7 @@ void PhyloTree::setUp(string tfile){
                        if (pid == 0) {  binUnclassified(taxFileNameTest);  }
                
                #else
-                       //create file needed for summary if it doesn't exist
-                       ifstream FileTest(taxFileNameTest.c_str());
-                       
-                       if (!FileTest) { 
-                               binUnclassified(taxFileNameTest); 
-                       }
+                       binUnclassified(taxFileNameTest); 
                #endif
        }
        catch(exception& e) {
@@ -364,17 +367,17 @@ void PhyloTree::binUnclassified(string file){
        try {
        
                ofstream out;
-               openOutputFile(file, out);
+               m->openOutputFile(file, out);
                
                map<string, int>::iterator itBin;
                map<string, int>::iterator childPointer;
                
                vector<TaxNode> copy = tree;
-                               
+                       
                //fill out tree
                fillOutTree(0, copy);
-               
-               //get leaf nodes that may need externsion
+       
+               //get leaf nodes that may need extension
                for (int i = 0; i < copy.size(); i++) {  
 
                        if (copy[i].children.size() == 0) {
@@ -383,7 +386,7 @@ void PhyloTree::binUnclassified(string file){
                }
                
                int copyNodes = copy.size();
-               
+       
                //go through the seqs and if a sequence finest taxon is not the same level as the most finely defined taxon then classify it as unclassified where necessary
                map<int, int>::iterator itLeaf;
                for (itLeaf = leafNodes.begin(); itLeaf != leafNodes.end(); itLeaf++) {
@@ -394,7 +397,7 @@ void PhyloTree::binUnclassified(string file){
                        int currentNode = itLeaf->second;
                        
                        //this sequence is unclassified at some levels
-                       while(level <= maxLevel){
+                       while(level < maxLevel){
                
                                level++;
                        
@@ -432,6 +435,7 @@ void PhyloTree::binUnclassified(string file){
 /**************************************************************************************************/
 void PhyloTree::fillOutTree(int index, vector<TaxNode>& copy) {
        try {
+       
                map<string,int>::iterator it;
                
                it = copy[index].children.find("unclassified");
@@ -443,12 +447,12 @@ void PhyloTree::fillOutTree(int index, vector<TaxNode>& copy) {
                        copy[copy.size()-1].level = copy[index].level + 1;
                }
                
-               if (tree[index].level <= maxLevel) {
+               if (tree[index].level < maxLevel) {
                        for(it=tree[index].children.begin();it!=tree[index].children.end();it++){ //check your children
                                fillOutTree(it->second, copy);
                        }
                }
-               
+
        }
        catch(exception& e) {
                m->errorOut(e, "PhyloTree", "fillOutTree");
@@ -478,8 +482,14 @@ string PhyloTree::getFullTaxonomy(string seqName) {
 
 void PhyloTree::print(ofstream& out, vector<TaxNode>& copy){
        try {
+       
+               //output mothur version
+               out << "#" << m->getVersion() << endl;
+               
                out << copy.size() << endl;
                
+               out << maxLevel << endl;
+               
                for (int i = 0; i < copy.size(); i++) {
        
                        out << copy[i].level << '\t'<< copy[i].name << '\t' << copy[i].children.size() << '\t';
@@ -511,7 +521,10 @@ void PhyloTree::printTreeNodes(string treefilename) {
                #endif
 
                        ofstream outTree;
-                       openOutputFile(treefilename, outTree);
+                       m->openOutputFile(treefilename, outTree);
+                       
+                       //output mothur version
+                       outTree << "#" << m->getVersion() << endl;
                        
                        //print treenodes
                        outTree << tree.size() << endl;