fixed bug with phylo.diversity. not calculating as intended.

[mothur.git] / phylotree.cpp

diff --git a/phylotree.cpp b/phylotree.cpp
index 1c2deba8a303edd692470dd1deb8c4e60062ac15..6b32e27201b5103e077261117f21fe30b69eb31c 100644 (file)
--- a/phylotree.cpp
+++ b/phylotree.cpp
@@ -125,7 +125,7 @@ PhyloTree::PhyloTree(string tfile){
                
                #ifdef USE_MPI
                        int pid, num, processors;
-                       vector<long> positions;
+                       vector<unsigned long int> positions;
                        
                        MPI_Status status; 
                        MPI_File inMPI;
@@ -197,14 +197,22 @@ PhyloTree::PhyloTree(string tfile){
 
 /**************************************************************************************************/
 
-string PhyloTree::getNextTaxon(string& heirarchy){
+string PhyloTree::getNextTaxon(string& heirarchy, string seqname){
        try {
                string currentLevel = "";
                if(heirarchy != ""){
                        int pos = heirarchy.find_first_of(';');
-                       currentLevel=heirarchy.substr(0,pos);
-                       if (pos != (heirarchy.length()-1)) {  heirarchy=heirarchy.substr(pos+1);  }
-                       else { heirarchy = ""; }
+                       
+                       if (pos == -1) { //you can't find another ;
+                               currentLevel = heirarchy;
+                               heirarchy = "";
+                               m->mothurOut(seqname + " is missing a ;, please check for other errors."); m->mothurOutEndLine();
+                       }else{
+                               currentLevel=heirarchy.substr(0,pos);
+                               if (pos != (heirarchy.length()-1)) {  heirarchy=heirarchy.substr(pos+1);  }
+                               else { heirarchy = ""; }
+                       }
+                       
                }
                return currentLevel;
        }
@@ -228,16 +236,16 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
                
                tree[0].accessions.push_back(seqName);
                string taxon;// = getNextTaxon(seqTaxonomy);
-               
+       
                while(seqTaxonomy != ""){
                        
                        level++;
-                       
+               
                        if (m->control_pressed) { return 0; }
                        
                        //somehow the parent is getting one too many accnos
                        //use print to reassign the taxa id
-                       taxon = getNextTaxon(seqTaxonomy);
+                       taxon = getNextTaxon(seqTaxonomy, seqName);
                        
                        if (taxon == "") {  m->mothurOut(seqName + " has an error in the taxonomy.  This may be due to a ;;"); m->mothurOutEndLine(); if (currentNode != 0) {  uniqueTaxonomies[currentNode] = currentNode; } break;  }
                        
@@ -254,21 +262,13 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
                                tree[currentNode].children[taxon] = numNodes-1;
                                tree[numNodes-1].parent = currentNode;
                                
-                               //                      int numChildren = tree[currentNode].children.size();
-                               //                      string heirarchyID = tree[currentNode].heirarchyID;
-                               //                      tree[currentNode].accessions.push_back(seqName);
-                               
                                currentNode = tree[currentNode].children[taxon];
                                tree[currentNode].accessions.push_back(seqName);
                                name2Taxonomy[seqName] = currentNode;
-                               //                      tree[currentNode].level = level;
-                               //                      tree[currentNode].childNumber = numChildren;
-                               //                      tree[currentNode].heirarchyID = heirarchyID + '.' + toString(tree[currentNode].childNumber);
                        }
        
                        if (seqTaxonomy == "") {   uniqueTaxonomies[currentNode] = currentNode; }
                }
-
        }
        catch(exception& e) {
                m->errorOut(e, "PhyloTree", "addSeqToTree");
@@ -370,18 +370,18 @@ void PhyloTree::binUnclassified(string file){
                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) {
                                leafNodes[i] = i;
                        }
                }
-               
+       //cout << "maxLevel = " << maxLevel << endl;    
                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
@@ -394,7 +394,7 @@ void PhyloTree::binUnclassified(string file){
                        int currentNode = itLeaf->second;
                        
                        //this sequence is unclassified at some levels
-                       while(level <= maxLevel){
+                       while(level < maxLevel){
                
                                level++;
                        
@@ -434,19 +434,21 @@ void PhyloTree::fillOutTree(int index, vector<TaxNode>& copy) {
        try {
                map<string,int>::iterator it;
                
-               it = copy[index].children.find("unclassified");
-               if (it == copy[index].children.end()) { //no unclassified at this level
-                       string taxon = "unclassified";
-                       copy.push_back(TaxNode(taxon));
-                       copy[index].children[taxon] = copy.size()-1;
-                       copy[copy.size()-1].parent = index;
-                       copy[copy.size()-1].level = copy[index].level + 1;
-               }
+               if (copy[index].level < maxLevel) {
+                       it = copy[index].children.find("unclassified");
+                       if (it == copy[index].children.end()) { //no unclassified at this level
+                               string taxon = "unclassified";
+                               copy.push_back(TaxNode(taxon));
+                               copy[index].children[taxon] = copy.size()-1;
+                               copy[copy.size()-1].parent = index;
+                               copy[copy.size()-1].level = copy[index].level + 1;
+                       }
                
-               if (tree[index].level <= maxLevel) {
-                       for(it=tree[index].children.begin();it!=tree[index].children.end();it++){ //check your children
+                       
+                       for(it=copy[index].children.begin();it!=copy[index].children.end();it++){ //check your children
                                fillOutTree(it->second, copy);
                        }
+                       
                }
                
        }
@@ -598,7 +600,7 @@ bool PhyloTree::ErrorCheck(vector<string> templateFileNames){
                for (int i = 0; i < templateFileNames.size(); i++) {
                        itFind = taxonomyFileNames.find(templateFileNames[i]);
                        
-                       if (itFind != name2Taxonomy.end()) { //found it so erase it
+                       if (itFind != taxonomyFileNames.end()) { //found it so erase it
                                taxonomyFileNames.erase(itFind);
                        }else {
                                m->mothurOut(templateFileNames[i] + " is in your template file and is not in your taxonomy file. Please correct."); m->mothurOutEndLine();