changed reading of name file to use buffered reads. note the splitAtWhiteSpace functi...

[mothur.git] / phylotree.cpp

diff --git a/phylotree.cpp b/phylotree.cpp
index 6ca36361dc55fe6c204017486c98ebeea77e7a9c..4ed3d8c1c02e1c5a1ee93b9dd6143309f5606f5c 100644 (file)
--- a/phylotree.cpp
+++ b/phylotree.cpp
@@ -20,6 +20,7 @@ PhyloTree::PhyloTree(){
                tree[0].heirarchyID = "0";
                maxLevel = 0;
                calcTotals = true;
+               addSeqToTree("unknown", "unknown;");
        }
        catch(exception& e) {
                m->errorOut(e, "PhyloTree", "PhyloTree");
@@ -127,7 +128,6 @@ PhyloTree::PhyloTree(string tfile){
                maxLevel = 0;
                calcTotals = true;
                string name, tax;
-
                
                #ifdef USE_MPI
                        int pid, num, processors;
@@ -191,7 +191,16 @@ PhyloTree::PhyloTree(string tfile){
                #endif
        
                assignHeirarchyIDs(0);
-       
+        
+        
+        string unknownTax = "unknown;";
+        //added last taxon until you get desired level
+               for (int i = 1; i < maxLevel; i++) {
+                       unknownTax += "unclassfied;";
+               }
+        
+        addSeqToTree("unknown", unknownTax);
+        
                //create file for summary if needed
                setUp(tfile);
        }
@@ -232,7 +241,6 @@ string PhyloTree::getNextTaxon(string& heirarchy, string seqname){
 
 int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
        try {
-                       
                numSeqs++;
                
                map<string, int>::iterator childPointer;
@@ -241,6 +249,8 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
                int level = 1;
                
                tree[0].accessions.push_back(seqName);
+               m->removeConfidences(seqTaxonomy);
+               
                string taxon;// = getNextTaxon(seqTaxonomy);
        
                while(seqTaxonomy != ""){
@@ -373,7 +383,7 @@ void PhyloTree::binUnclassified(string file){
                map<string, int>::iterator childPointer;
                
                vector<TaxNode> copy = tree;
-                       
+               
                //fill out tree
                fillOutTree(0, copy);
        
@@ -482,16 +492,16 @@ 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';
                        
                        map<string,int>::iterator it;
@@ -604,23 +614,27 @@ bool PhyloTree::ErrorCheck(vector<string> templateFileNames){
        try {
        
                bool okay = true;
+               templateFileNames.push_back("unknown");
                
                map<string, int>::iterator itFind;
                map<string, int> taxonomyFileNames = name2Taxonomy;
                
+        if (m->debug) { m->mothurOut("[DEBUG]: in error check. Numseqs in template = " + toString(templateFileNames.size()) + ". Numseqs in taxonomy = " + toString(taxonomyFileNames.size()) + ".\n"); }
+        
                for (int i = 0; i < templateFileNames.size(); i++) {
                        itFind = taxonomyFileNames.find(templateFileNames[i]);
                        
                        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();
+                               m->mothurOut("'" +templateFileNames[i] + "' is in your template file and is not in your taxonomy file. Please correct."); m->mothurOutEndLine();
                                okay = false;
                        }
                        
-                       templateFileNames.erase(templateFileNames.begin()+i);
-                       i--;
+                       //templateFileNames.erase(templateFileNames.begin()+i);
+                       //i--;
                }
+               templateFileNames.clear();
                
                if (taxonomyFileNames.size() > 0) { //there are names in tax file that are not in template
                        okay = false;