X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=phylosummary.cpp;fp=phylosummary.cpp;h=a9a170b86c77dff145079df269fd6b36813848ee;hb=a5af3313f0221a8b41aa362d72eadb60a4dd6e27;hp=4e94605e47caf6e09d099a4df32dfdd7677be676;hpb=2e77272b6f5987bd2bde592bb15262a8c69ea9e0;p=mothur.git

diff --git a/phylosummary.cpp b/phylosummary.cpp
index 4e94605..a9a170b 100644
--- a/phylosummary.cpp
+++ b/phylosummary.cpp
@@ -15,6 +15,7 @@ PhyloSummary::PhyloSummary(string refTfile, string groupFile){
 	try {
 		m = MothurOut::getInstance();
 		maxLevel = 0;
+		ignore = false;
 		
 		if (groupFile != "") {
 			groupmap = new GroupMap(groupFile);
@@ -42,6 +43,32 @@ PhyloSummary::PhyloSummary(string refTfile, string groupFile){
 		exit(1);
 	}
 }
+
+/**************************************************************************************************/
+
+PhyloSummary::PhyloSummary(string groupFile){
+	try {
+		m = MothurOut::getInstance();
+		maxLevel = 0;
+		ignore = true;
+		
+		if (groupFile != "") {
+			groupmap = new GroupMap(groupFile);
+			groupmap->readMap();
+		}else{
+			groupmap = NULL;
+		}
+		
+		tree.push_back(rawTaxNode("Root"));
+		tree[0].rank = "0";
+		
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "PhyloSummary", "PhyloSummary");
+		exit(1);
+	}
+}
 /**************************************************************************************************/
 
 void PhyloSummary::summarize(string userTfile){
@@ -128,9 +155,43 @@ int PhyloSummary::addSeqToTree(string seqName, string seqTaxonomy){
 				tree[childPointer->second].total++;
 
 				currentNode = childPointer->second;
-			}else{	//otherwise, error
-				m->mothurOut("Warning: cannot find taxon " + taxon + " in reference taxonomy tree at level " + toString(tree[currentNode].level) + " for " + seqName + ". This may cause totals of daughter levels not to add up in summary file."); m->mothurOutEndLine();
-				break;
+			}else{	
+				if (ignore) {
+						
+					tree.push_back(rawTaxNode(taxon));
+					int index = tree.size() - 1;
+				
+					tree[index].parent = currentNode;
+					tree[index].level = (level+1);
+					tree[index].total = 1;
+					tree[currentNode].children[taxon] = index;
+					
+					//initialize groupcounts
+					if (groupmap != NULL) {
+						for (int j = 0; j < groupmap->namesOfGroups.size(); j++) {
+							tree[index].groupCount[groupmap->namesOfGroups[j]] = 0;
+						}
+						
+						//find out the sequences group
+						string group = groupmap->getGroup(seqName);
+						
+						if (group == "not found") {  m->mothurOut(seqName + " is not in your groupfile, and will be included in the overall total, but not any group total."); m->mothurOutEndLine();  }
+						
+						//do you have a count for this group?
+						map<string, int>::iterator itGroup = tree[index].groupCount.find(group);
+						
+						//if yes, increment it - there should not be a case where we can't find it since we load group in read
+						if (itGroup != tree[index].groupCount.end()) {
+							tree[index].groupCount[group]++;
+						}						
+					}
+					
+					currentNode = index;
+					
+				}else{ //otherwise, error
+					m->mothurOut("Warning: cannot find taxon " + taxon + " in reference taxonomy tree at level " + toString(tree[currentNode].level) + " for " + seqName + ". This may cause totals of daughter levels not to add up in summary file."); m->mothurOutEndLine();
+					break;
+				}
 			}
 			
 			level++;
@@ -148,6 +209,117 @@ int PhyloSummary::addSeqToTree(string seqName, string seqTaxonomy){
 }
 /**************************************************************************************************/
 
+int PhyloSummary::addSeqToTree(string seqTaxonomy, vector<string> names){
+	try {
+		numSeqs++;
+		
+		map<string, int>::iterator childPointer;
+		
+		int currentNode = 0;
+		string taxon;
+		
+		int level = 0;
+		
+		while (seqTaxonomy != "") {
+			
+			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);
+			
+			childPointer = tree[currentNode].children.find(taxon);
+			
+			if(childPointer != tree[currentNode].children.end()){	//if the node already exists, update count and move on
+				if (groupmap != NULL) {
+					
+					map<string, bool> containsGroup; 
+					for (int j = 0; j < groupmap->namesOfGroups.size(); j++) {
+						containsGroup[groupmap->namesOfGroups[j]] = false;
+					}
+					
+					for (int k = 0; k < names.size(); k++) {
+						//find out the sequences group
+						string group = groupmap->getGroup(names[k]);
+					
+						if (group == "not found") {  m->mothurOut(names[k] + " is not in your groupfile, and will be included in the overall total, but not any group total."); m->mothurOutEndLine();  }
+						else {
+							containsGroup[group] = true;
+						}
+					}
+					
+					for (map<string, bool>::iterator itGroup = containsGroup.begin(); itGroup != containsGroup.end(); itGroup++) {
+						if (itGroup->second == true) {
+							tree[childPointer->second].groupCount[itGroup->first]++;
+						}
+					}
+					
+				}
+				
+				tree[childPointer->second].total++;
+				
+				currentNode = childPointer->second;
+			}else{	
+				if (ignore) {
+					
+					tree.push_back(rawTaxNode(taxon));
+					int index = tree.size() - 1;
+					
+					tree[index].parent = currentNode;
+					tree[index].level = (level+1);
+					tree[index].total = 1;
+					tree[currentNode].children[taxon] = index;
+					
+					//initialize groupcounts
+					if (groupmap != NULL) {
+						map<string, bool> containsGroup; 
+						for (int j = 0; j < groupmap->namesOfGroups.size(); j++) {
+							tree[index].groupCount[groupmap->namesOfGroups[j]] = 0;
+							containsGroup[groupmap->namesOfGroups[j]] = false;
+						}
+						
+						
+						for (int k = 0; k < names.size(); k++) {
+							//find out the sequences group
+							string group = groupmap->getGroup(names[k]);
+							
+							if (group == "not found") {  m->mothurOut(names[k] + " is not in your groupfile, and will be included in the overall total, but not any group total."); m->mothurOutEndLine();  }
+							else {
+								containsGroup[group] = true;
+							}
+						}
+						
+						for (map<string, bool>::iterator itGroup = containsGroup.begin(); itGroup != containsGroup.end(); itGroup++) {
+							if (itGroup->second == true) {
+								tree[index].groupCount[itGroup->first]++;
+							}
+						}
+					}
+					
+					currentNode = index;
+					
+				}else{ //otherwise, error
+					m->mothurOut("Warning: cannot find taxon " + taxon + " in reference taxonomy tree at level " + toString(tree[currentNode].level) + ". This may cause totals of daughter levels not to add up in summary file."); m->mothurOutEndLine();
+					break;
+				}
+			}
+			
+			level++;
+			
+			if ((seqTaxonomy == "") && (level < maxLevel)) {  //if you think you are done and you are not.
+				for (int k = level; k < maxLevel; k++) {  seqTaxonomy += "unclassified;";   }
+			}
+		}
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "PhyloSummary", "addSeqToTree");
+		exit(1);
+	}
+}
+
+/**************************************************************************************************/
+
 void PhyloSummary::assignRank(int index){
 	try {
 		map<string,int>::iterator it;
@@ -169,6 +341,9 @@ void PhyloSummary::assignRank(int index){
 
 void PhyloSummary::print(ofstream& out){
 	try {
+		
+		if (ignore) { assignRank(0); }
+		
 		//print labels
 		out << "taxlevel\t rankID\t taxon\t daughterlevels\t total\t";
 		if (groupmap != NULL) {