working on pam

[mothur.git] / phylotree.cpp

diff --git a/phylotree.cpp b/phylotree.cpp
index 8a7c712b0f255db20da492fd95bc15d3116ce151..2e3136e1f42e1a93d15fcab921b2e612f55f3997 100644 (file)
--- a/phylotree.cpp
+++ b/phylotree.cpp
@@ -233,6 +233,48 @@ string PhyloTree::getNextTaxon(string& heirarchy, string seqname){
                exit(1);
        }
 }
+/**************************************************************************************************/
+
+vector<string> PhyloTree::getSeqs(string seqTaxonomy){
+       try {
+        string taxCopy = seqTaxonomy;
+        vector<string> names;
+        map<string, int>::iterator childPointer;
+               
+               int currentNode = 0;
+
+        m->removeConfidences(seqTaxonomy);
+        
+        string taxon;
+        while(seqTaxonomy != ""){
+                       
+                       if (m->control_pressed) { return names; }
+                       
+                       taxon = getNextTaxon(seqTaxonomy, "");
+            
+            if (m->debug) { m->mothurOut(taxon +'\n'); }
+                       
+                       if (taxon == "") {  m->mothurOut(taxCopy + " has an error in the taxonomy.  This may be due to a ;;"); m->mothurOutEndLine(); break;  }
+                       
+                       childPointer = tree[currentNode].children.find(taxon);
+                       
+                       if(childPointer != tree[currentNode].children.end()){   //if the node already exists, move on
+                               currentNode = childPointer->second;
+                       }
+                       else{                                                                                   //otherwise, error this taxonomy is not in tree
+                               m->mothurOut("[ERROR]: " + taxCopy + " is not in taxonomy tree, please correct."); m->mothurOutEndLine(); m->control_pressed = true; return names;
+                       }
+            
+                       if (seqTaxonomy == "") {   names = tree[currentNode].accessions;        }
+               }
+        
+        return names;
+    }
+       catch(exception& e) {
+               m->errorOut(e, "PhyloTree", "getSeqs");
+               exit(1);
+       }
+}
 
 /**************************************************************************************************/
 
@@ -259,6 +301,8 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
                        //somehow the parent is getting one too many accnos
                        //use print to reassign the taxa id
                        taxon = getNextTaxon(seqTaxonomy, seqName);
+            
+            if (m->debug) { m->mothurOut(seqName +'\t' + taxon +'\n'); }
                        
                        if (taxon == "") {  m->mothurOut(seqName + " has an error in the taxonomy.  This may be due to a ;;"); m->mothurOutEndLine(); if (currentNode != 0) {  uniqueTaxonomies.insert(currentNode); } break;  }
                        
@@ -341,6 +385,9 @@ void PhyloTree::assignHeirarchyIDs(int index){
                int counter = 1;
                
                for(it=tree[index].children.begin();it!=tree[index].children.end();it++){
+            
+            if (m->debug) { m->mothurOut(toString(index) +'\t' + tree[it->second].name +'\n'); }
+                
                        tree[it->second].heirarchyID = tree[index].heirarchyID + '.' + toString(counter);
                        counter++;
                        tree[it->second].level = tree[index].level + 1;
@@ -399,6 +446,8 @@ void PhyloTree::binUnclassified(string file){
                        }
                }
                
+        if (m->debug) { m->mothurOut("maxLevel = " + toString(maxLevel) +'\n'); }
+        
                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
@@ -409,11 +458,14 @@ void PhyloTree::binUnclassified(string file){
                        
                        int level = copy[itLeaf->second].level;
                        int currentNode = itLeaf->second;
+            
+            if (m->debug) { m->mothurOut(copy[currentNode].name +'\n'); }
                        
                        //this sequence is unclassified at some levels
                        while(level < maxLevel){
                
                                level++;
+                if (m->debug) { m->mothurOut("level = " + toString(level) +'\n'); }
                        
                                string taxon = "unclassified";