Adds taxonomy spaces to summary.tax

source/classifier/classify.cpp

@@ -101,7 +101,6 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
         m->mothurOut("DONE."); m->mothurOutEndLine();
         m->mothurOut("It took " + toString(time(NULL) - start) + " seconds generate search database. "); m->mothurOutEndLine();
         
-        
         readTaxonomy(taxFile);
         
         //sanity check

source/classifier/phylosummary.cpp

@@ -292,10 +292,7 @@ int PhyloSummary::addSeqToTree(string seqName, string seqTaxonomy){
             
             if (level > maxLevel) { maxLevel = level; }
 			
-			if ((seqTaxonomy == "") && (level < maxLevel)) {  //if you think you are done and you are not.
-				for (int k = level; k < maxLevel; k++) {  seqTaxonomy += "unclassified;";   }
-			}
-		}
+        }
 		return 0;
 	}
 	catch(exception& e) {
@@ -368,9 +365,6 @@ int PhyloSummary::addSeqToTree(string seqTaxonomy, map<string, bool> containsGro
             
             if (level > maxLevel) { maxLevel = level; }
 			
-			if ((seqTaxonomy == "") && (level < maxLevel)) {  //if you think you are done and you are not.
-				for (int k = level; k < maxLevel; k++) {  seqTaxonomy += "unclassified;";   }
-			}
 		}
 		return 0;
 	}
@@ -755,13 +749,15 @@ void PhyloSummary::readTreeStruct(ifstream& in){
 		//read the tree file
 		for (int i = 0; i < tree.size(); i++) {
 	
-			in >> tree[i].level >> tree[i].name >> num; //num contains the number of children tree[i] has
-			
+			in >> tree[i].level >> num; m->gobble(in); //num contains the number of children tree[i] has
+            tree[i].name = m->getline(in); m->gobble(in);
+            
 			//set children
 			string childName;
 			int childIndex;
 			for (int j = 0; j < num; j++) {
-				in >> childName >> childIndex;
+				in >> childIndex; m->gobble(in);
+                childName = m->getline(in); m->gobble(in);
 				tree[i].children[childName] = childIndex;
 			}
 			

source/classifier/phylotree.cpp

@@ -46,7 +46,8 @@ PhyloTree::PhyloTree(ifstream& in, string filename){
         in >> maxLevel; m->gobble(in);
         
         for (int i = 0; i < tree.size(); i++) {
-            in >> tree[i].name >> tree[i].level >> tree[i].parent; m->gobble(in);
+            tree[i].name = m->getline(in); m->gobble(in);
+            in >> tree[i].level >> tree[i].parent; m->gobble(in);
         }
         
         //read genus nodes
@@ -454,11 +455,12 @@ void PhyloTree::print(ofstream& out, vector<TaxNode>& copy){
 				
 		for (int i = 0; i < copy.size(); i++) {
 				
-			out << copy[i].level << '\t'<< copy[i].name << '\t' << copy[i].children.size();
+			out << copy[i].level << '\t' << copy[i].children.size() << endl;
+            out << copy[i].name << endl;
 			
 			map<string,int>::iterator it;
 			for(it=copy[i].children.begin();it!=copy[i].children.end();it++){
-				out << '\t' << it->first << '\t' << it->second;
+				out << it->second << '\t' << it->first << endl;
 			}
 			out << endl;
 		}
@@ -485,7 +487,8 @@ void PhyloTree::printTreeNodes(string treefilename) {
         outTree << maxLevel << endl;
         
         for (int i = 0; i < tree.size(); i++) {
-            outTree << tree[i].name << '\t' << tree[i].level << '\t' << tree[i].parent << endl;
+            outTree << tree[i].name << endl;
+            outTree << tree[i].level << '\t' << tree[i].parent << endl;
         }
         
         //print genus nodes

source/classifier/taxonomyequalizer.cpp

@@ -44,7 +44,8 @@ TaxEqualizer::TaxEqualizer(string tfile, int c, string o) : cutoff(c), outputDir
 			
 				if (m->control_pressed) {  break; }
 				
-				in >> name >> tax;   m->gobble(in);
+                in >> name;   m->gobble(in);
+                tax = m->getline(in); m->gobble(in);
 				
 				if (containsConfidence) {  m->removeConfidences(tax);	}
 				
@@ -78,7 +79,8 @@ int TaxEqualizer::getHighestLevel(ifstream& in) {
 		string name, tax;
 		
 		while (in) {
-			in >> name >> tax;   m->gobble(in);
+            in >> name;   m->gobble(in);
+            tax = m->getline(in); m->gobble(in);
 		
 			//count levels in this taxonomy
 			int thisLevel = 0;

source/commands/classifyseqscommand.cpp

@@ -721,7 +721,8 @@ int ClassifySeqsCommand::execute(){
             while (!inTax.eof()) {
                 if (m->control_pressed) { outputTypes.clear(); if (ct != NULL) { delete ct; }  if (groupMap != NULL) { delete groupMap; } delete taxaSum; for (int i = 0; i < outputNames.size(); i++) {	m->mothurRemove(outputNames[i]);	} delete classify; return 0; }
                 
-                inTax >> name >> taxon; m->gobble(inTax);
+                inTax >> name; m->gobble(inTax);
+                taxon = m->getline(inTax); m->gobble(inTax);
                 
                 string newTax = m->addUnclassifieds(taxon, maxLevel, probs);
                 

source/commands/summarytaxcommand.cpp

@@ -220,7 +220,9 @@ SummaryTaxCommand::SummaryTaxCommand(string option)  {
 
 int SummaryTaxCommand::execute(){
 	try{
-		
+        
+        int maxLevel = findMaxLevel(taxfile);
+        
 		if (abort == true) { if (calledHelp) { return 0; }  return 2;	}
 		int start = time(NULL);
 		
@@ -241,63 +243,45 @@ int SummaryTaxCommand::execute(){
 		if (m->control_pressed) { if (groupMap != NULL) { delete groupMap; } if (ct != NULL) { delete ct; } delete taxaSum; return 0; }
 		
 		int numSeqs = 0;
-		if ((threshold == 0) && ((namefile == "") || (countfile != ""))) { numSeqs = taxaSum->summarize(taxfile);  }
-        else if (threshold != 0) {
-            ifstream in;
-            m->openInputFile(taxfile, in);
+        map<string, vector<string> > nameMap;
+        map<string, vector<string> >::iterator itNames;
+        if (namefile != "") { m->readNames(namefile, nameMap); }
+		
+        ifstream in;
+        m->openInputFile(taxfile, in);
+        
+        string name, taxon;
+        while(!in.eof()){
             
-            string name, taxon;
-            while(!in.eof()){
-                
-                if (m->control_pressed) { break; }
-                
-                in >> name >> taxon; m->gobble(in);
-                
-                if (threshold != 0) {  taxon = processTaxMap(taxon);  }
-                
-                //cout << taxon << endl;
+            if (m->control_pressed) { break; }
+            
+            in >> name; m->gobble(in);
+            taxon = m->getline(in); m->gobble(in);
+            
+            string newTax = m->addUnclassifieds(taxon, maxLevel, true);
+            
+            if (threshold != 0) {  newTax = processTaxMap(newTax);  }
+            
+            //add sequence to summary, countfile info included from Phylosummary constructor
+            if (namefile != "") {
+                itNames = nameMap.find(name);
                 
-                //add sequence to summary, countfile info included from Phylosummary constructor
-                taxaSum->addSeqToTree(name, taxon);
+                if (itNames == nameMap.end()) {
+                    m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
+                }else{
+                    for (int i = 0; i < itNames->second.size(); i++) {
+                        taxaSum->addSeqToTree(itNames->second[i], newTax);  //add it as many times as there are identical seqs
+                    }
+                    itNames->second.clear();
+                    nameMap.erase(itNames->first);
+                }
+            }else {
+                taxaSum->addSeqToTree(name, newTax);
             }
-            in.close();
+            
         }
-		else if (namefile != "") {
-			map<string, vector<string> > nameMap;
-			map<string, vector<string> >::iterator itNames;
-			m->readNames(namefile, nameMap);
-			
-			if (m->control_pressed) { if (groupMap != NULL) { delete groupMap; } if (ct != NULL) { delete ct; } delete taxaSum; return 0; }
-			
-			ifstream in;
-			m->openInputFile(taxfile, in);
-			
-			//read in users taxonomy file and add sequences to tree
-			string name, taxon;
-			
-			while(!in.eof()){
-                
-                if (m->control_pressed) { break; }
-                
-				in >> name >> taxon; m->gobble(in);
-                
-                if (threshold != 0) {  taxon = processTaxMap(taxon);  }
-				
-				itNames = nameMap.find(name);
-				
-				if (itNames == nameMap.end()) { 
-					m->mothurOut("[ERROR]: " + name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
-				}else{
-					for (int i = 0; i < itNames->second.size(); i++) { 
-						numSeqs++;
-						taxaSum->addSeqToTree(itNames->second[i], taxon);  //add it as many times as there are identical seqs
-					}
-					itNames->second.clear();
-					nameMap.erase(itNames->first);
-				}
-			}
-			in.close();
-        }else if (threshold == 0) { numSeqs = taxaSum->summarize(taxfile);  }
+        in.close();
+        
 		
 		if (m->control_pressed) {  if (groupMap != NULL) { delete groupMap; } if (ct != NULL) { delete ct; } delete taxaSum; return 0; }
 		
@@ -382,10 +366,10 @@ string SummaryTaxCommand::processTaxMap(string tax) {
         if (spot != 0) {
             newTax = "";
             for (int i = 0; i < taxons.size(); i++) {  newTax += taxons[i] + ";";  }
-            for (int i = spot; i < taxLength; i++) {
-                if(tax[i] == ';'){   newTax += "unclassified;"; }
-                m->removeConfidences(newTax);
-            }
+            //for (int i = spot; i < taxLength; i++) {
+                //if(tax[i] == ';'){   newTax += "unclassified;"; }
+                //m->removeConfidences(newTax);
+            //}
         }else { m->removeConfidences(tax); newTax = tax; } //leave tax alone
         
         return newTax;
@@ -395,6 +379,21 @@ string SummaryTaxCommand::processTaxMap(string tax) {
         exit(1);
     }
 }
+/**************************************************************************************************/
+int SummaryTaxCommand::findMaxLevel(string file) {
+    try{
+        GroupMap* groupMap = NULL;
+        PhyloSummary taxaSum(groupMap, false, -1);
+        
+        taxaSum.summarize(file);
+       
+        return taxaSum.getMaxLevel();
+    }
+    catch(exception& e) {
+        m->errorOut(e, "SummaryTaxCommand", "findMaxLevel");
+        exit(1);
+    }
+}
 /**************************************************************************************/
 
 

source/commands/summarytaxcommand.h

@@ -41,6 +41,7 @@ class SummaryTaxCommand : public Command {
 		map<string, int> nameMap;
     
         string processTaxMap(string);
+        int findMaxLevel(string);
 };
 
 /**************************************************************************************************/

source/mothurout.cpp

@@ -2480,78 +2480,39 @@ vector<string> MothurOut::splitWhiteSpaceWithQuotes(string input){
 	}
 }
 //**********************************************************************************************************************
-int MothurOut::readTax(string namefile, map<string, string>& taxMap, bool removeConfidence) {
+int MothurOut::readTax(string taxfile, map<string, string>& taxMap, bool removeConfidence) {
 	try {
         //open input file
 		ifstream in;
-		openInputFile(namefile, in);
+		openInputFile(taxfile, in);
         
-        string rest = "";
-        char buffer[4096];
-        bool pairDone = false;
-        bool columnOne = true;
-        string firstCol, secondCol;
         bool error = false;
+        string name, taxonomy;
         
 		while (!in.eof()) {
 			if (control_pressed) { break; }
 			
-            in.read(buffer, 4096);
-            vector<string> pieces = splitWhiteSpace(rest, buffer, in.gcount());
+            in >> name; gobble(in);
+            taxonomy = getline(in); gobble(in);
             
-            for (int i = 0; i < pieces.size(); i++) {
-                if (columnOne) {  firstCol = pieces[i]; columnOne=false; }
-                else  { secondCol = pieces[i]; pairDone = true; columnOne=true; }
-                
-                if (pairDone) { 
-                    checkName(firstCol);
-                    //are there confidence scores, if so remove them
-                    if (removeConfidence) { if (secondCol.find_first_of('(') != -1) {  removeConfidences(secondCol);	} }
-                    map<string, string>::iterator itTax = taxMap.find(firstCol);
-                    
-                    if(itTax == taxMap.end()) {
-                        bool ignore = false;
-                        if (secondCol != "") { if (secondCol[secondCol.length()-1] != ';') { mothurOut("[ERROR]: " + firstCol + " is missing the final ';', ignoring.\n"); ignore=true; }
-                        }
-                        if (!ignore) { taxMap[firstCol] = secondCol; }
-                        if (debug) {  mothurOut("[DEBUG]: name = '" + firstCol + "' tax = '" + secondCol + "'\n");  }
-                    }else {
-                        mothurOut("[ERROR]: " + firstCol + " is already in your taxonomy file, names must be unique.\n"); error = true;
-                    }
-                    pairDone = false; 
+            checkName(name);
+            //are there confidence scores, if so remove them
+            if (removeConfidence) {  if (taxonomy.find_first_of('(') != -1) {  removeConfidences(taxonomy);	} }
+            map<string, string>::iterator itTax = taxMap.find(name);
+            
+            if(itTax == taxMap.end()) {
+                bool ignore = false;
+                if (taxonomy != "") { if (taxonomy[taxonomy.length()-1] != ';') { mothurOut("[ERROR]: " + name + " is missing the final ';', ignoring.\n"); ignore=true; }
                 }
+                if (!ignore) { taxMap[name] = taxonomy; }
+                if (debug) {  mothurOut("[DEBUG]: name = '" + name + "' tax = '" + taxonomy + "'\n");  }
+            }else {
+                mothurOut("[ERROR]: " + name + " is already in your taxonomy file, names must be unique./n"); error = true;
             }
-		}
-		in.close();
-        
-        if (rest != "") {
-            vector<string> pieces = splitWhiteSpace(rest);
-            
-            for (int i = 0; i < pieces.size(); i++) {
-                if (columnOne) {  firstCol = pieces[i]; columnOne=false; }
-                else  { secondCol = pieces[i]; pairDone = true; columnOne=true; }
-                
-                if (pairDone) { 
-                    checkName(firstCol);
-                    //are there confidence scores, if so remove them
-                    if (removeConfidence) {  if (secondCol.find_first_of('(') != -1) {  removeConfidences(secondCol);	} }
-                    map<string, string>::iterator itTax = taxMap.find(firstCol);
-                    
-                    if(itTax == taxMap.end()) {
-                        bool ignore = false;
-                        if (secondCol != "") { if (secondCol[secondCol.length()-1] != ';') { mothurOut("[ERROR]: " + firstCol + " is missing the final ';', ignoring.\n"); ignore=true; }
-                        }
-                        if (!ignore) { taxMap[firstCol] = secondCol; }
-                        if (debug) {  mothurOut("[DEBUG]: name = '" + firstCol + "' tax = '" + secondCol + "'\n");  }
-                    }else {
-                        mothurOut("[ERROR]: " + firstCol + " is already in your taxonomy file, names must be unique./n"); error = true;
-                    }
 
-                    pairDone = false; 
-                }
-            } 
         }
-		
+		in.close();
+        
         if (error) { control_pressed = true; }
         if (debug) {  mothurOut("[DEBUG]: numSeqs saved = '" + toString(taxMap.size()) + "'\n"); }
 		return taxMap.size();