X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=sharedrabundvector.cpp;h=9b2bb678ca7688777aa87a9f60015baba5d336d9;hp=142f787e24ac3663c19783874721299a071336cf;hb=df7e3ff9f68ef157b0328a2d353c3258c5d45d89;hpb=5a9f3cce3e2e053c249e3e97507ce180f0854d76

diff --git a/sharedrabundvector.cpp b/sharedrabundvector.cpp
index 142f787..9b2bb67 100644
--- a/sharedrabundvector.cpp
+++ b/sharedrabundvector.cpp
@@ -14,8 +14,7 @@
 
 
 /***********************************************************************/
-
-SharedRAbundVector::SharedRAbundVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0) {globaldata = GlobalData::getInstance();}
+SharedRAbundVector::SharedRAbundVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0) {} 
 /***********************************************************************/
 
 SharedRAbundVector::~SharedRAbundVector() {
@@ -26,7 +25,6 @@ SharedRAbundVector::~SharedRAbundVector() {
 /***********************************************************************/
 
 SharedRAbundVector::SharedRAbundVector(int n) : DataVector(), maxRank(0), numBins(n), numSeqs(0) {
-		globaldata = GlobalData::getInstance();
 		individual newGuy;
 		//initialize data
 		for (int i=0; i< n; i++) {
@@ -51,30 +49,84 @@ SharedRAbundVector::SharedRAbundVector(string id, vector<individual> rav) : Data
 		}
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
+		m->errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
 		exit(1);
 	}
 }
 
 
-/***********************************************************************/
+***********************************************************************/
 //reads a shared file
 SharedRAbundVector::SharedRAbundVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
 	try {
-		globaldata = GlobalData::getInstance();
-		
-		if (globaldata->gGroupmap == NULL) {  groupmap = new GroupMap(); }
+		m->clearAllGroups();
+		vector<string> allGroups;
 		
 		int num, inputData, count;
 		count = 0;  
 		string holdLabel, nextLabel, groupN;
 		individual newguy;
 		
-		for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
-		lookup.clear();
+		for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  lookup.clear();
+		
+		//are we at the beginning of the file??
+		if (m->saveNextLabel == "") {  
+			f >> label; 
+            
+			//is this a shared file that has headers
+			if (label == "label") { 
+				//gets "group"
+				f >> label; m->gobble(f);
+				
+				//gets "numOtus"
+				f >> label; m->gobble(f);
+				
+				//eat rest of line
+				label = m->getline(f); m->gobble(f);
+				
+				//parse labels to save
+				istringstream iStringStream(label);
+				m->binLabelsInFile.clear();
+				while(!iStringStream.eof()){
+					if (m->control_pressed) { break; }
+					string temp;
+					iStringStream >> temp;  m->gobble(iStringStream);
+		
+					m->binLabelsInFile.push_back(temp);
+				}
+				
+				f >> label >> groupN >> num;
+			}else {
+                //read in first row since you know there is at least 1 group.
+                f >> groupN >> num;
+                
+                //make binlabels because we don't have any
+                string snumBins = toString(num);
+                m->binLabelsInFile.clear();
+                for (int i = 0; i < num; i++) {  
+                    //if there is a bin label use it otherwise make one
+                    string binLabel = "Otu";
+                    string sbinNumber = toString(i+1);
+                    if (sbinNumber.length() < snumBins.length()) { 
+                        int diff = snumBins.length() - sbinNumber.length();
+                        for (int h = 0; h < diff; h++) { binLabel += "0"; }
+                    }
+                    binLabel += sbinNumber;
+                    m->binLabelsInFile.push_back(binLabel);
+                }
+            }
+		}else { 
+            label = m->saveNextLabel; 
+            
+            //read in first row since you know there is at least 1 group.
+            f >> groupN >> num;
+            
+            if (m->debug) { m->mothurOut("[DEBUG]: "+ groupN + '\t' + toString(num)); }
+        }
+		
+		//reset labels, currentLabels may have gotten changed as otus were eliminated because of group choices or sampling
+		m->currentBinLabels = m->binLabelsInFile;
 		
-		//read in first row since you know there is at least 1 group.
-		f >> label >> groupN >> num;
 		holdLabel = label;
 		
 		//add new vector to lookup
@@ -83,38 +135,30 @@ SharedRAbundVector::SharedRAbundVector(ifstream& f) : DataVector(), maxRank(0),
 		lookup[0]->setLabel(label);
 		lookup[0]->setGroup(groupN);
 		
-		if (globaldata->gGroupmap == NULL) { 
-			//save group in groupmap
-			groupmap->namesOfGroups.push_back(groupN);
-			groupmap->groupIndex[groupN] = 0;
-		}
+		allGroups.push_back(groupN);
 		
 		//fill vector.  data = first sharedrabund in file
 		for(int i=0;i<num;i++){
 			f >> inputData;
+            if (m->debug) { m->mothurOut("[DEBUG]: OTU" + toString(i+1)+ '\t' +toString(inputData)); }
 			
-			lookup[0]->push_back(inputData, i, groupN); //abundance, bin, group
-			push_back(inputData, i, groupN);
-			numSeqs += inputData;
-			numBins++;
-			if (inputData > maxRank) { maxRank = inputData; }
+			lookup[0]->push_back(inputData, groupN); //abundance, bin, group
+			push_back(inputData, groupN);
 			
+			if (inputData > maxRank) { maxRank = inputData; }
 		}
 		
-		gobble(f);
-		
-		if (f.eof() != true) { f >> nextLabel; }
+		m->gobble(f);
 		
+		if (!(f.eof())) { f >> nextLabel; }
+	
 		//read the rest of the groups info in
 		while ((nextLabel == holdLabel) && (f.eof() != true)) {
 			f >> groupN >> num;
+            if (m->debug) { m->mothurOut("[DEBUG]: "+ groupN + '\t' + toString(num)); }
 			count++;
 			
-			if (globaldata->gGroupmap == NULL) { 
-				//save group in groupmap
-				groupmap->namesOfGroups.push_back(groupN);
-				groupmap->groupIndex[groupN] = count;
-			}
+			allGroups.push_back(groupN);
 			
 			//add new vector to lookup
 			temp = new SharedRAbundVector();
@@ -125,22 +169,20 @@ SharedRAbundVector::SharedRAbundVector(ifstream& f) : DataVector(), maxRank(0),
 			//fill vector.  
 			for(int i=0;i<num;i++){
 				f >> inputData;
-				lookup[count]->push_back(inputData, i, groupN); //abundance, bin, group
+                if (m->debug) { m->mothurOut("[DEBUG]: OTU" + toString(i+1)+ '\t' +toString(inputData)); }
+				
+				lookup[count]->push_back(inputData, groupN); //abundance, bin, group
 			}
 			
-			gobble(f);
+			m->gobble(f);
 				
 			if (f.eof() != true) { f >> nextLabel; }
 		}
-		
-		//put file pointer back since you are now at a new distance label
-		for (int i = 0; i < nextLabel.length(); i++) { f.unget();  }
-	
-		if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap;  }
-		
+		m->saveNextLabel = nextLabel;
+		m->setAllGroups(allGroups);
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
+		m->errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
 		exit(1);
 	}
 }
@@ -158,7 +200,7 @@ void SharedRAbundVector::set(int binNumber, int newBinSize, string groupname){
 		numSeqs += (newBinSize - oldBinSize);
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "set");
+		m->errorOut(e, "SharedRAbundVector", "set");
 		exit(1);
 	}
 }
@@ -174,6 +216,18 @@ int SharedRAbundVector::getAbundance(int index){
 	return data[index].abundance;
 	
 }
+/***********************************************************************/
+//returns vector of abundances 
+vector<int> SharedRAbundVector::getAbundances(){
+    vector<int> abunds;
+    for (int i = 0; i < data.size(); i++) {
+        abunds.push_back(data[i].abundance);
+    }
+    
+	return abunds;
+}
+
+
 /***********************************************************************/
 
 int SharedRAbundVector::numNZ(){
@@ -202,12 +256,22 @@ vector <individual> SharedRAbundVector::getData(){
 }
 /***********************************************************************/
 
-void SharedRAbundVector::push_back(int binSize, int otu, string groupName){
+void SharedRAbundVector::clear(){
+	numBins = 0;
+	maxRank = 0;
+	numSeqs = 0;
+	data.clear();
+	for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
+	lookup.clear();
+}
+/***********************************************************************/
+
+void SharedRAbundVector::push_back(int binSize, string groupName){
 	try {
 		individual newGuy;
 		newGuy.abundance = binSize;
 		newGuy.group = groupName;
-		newGuy.bin = otu;
+		newGuy.bin = data.size();
 		
 		data.push_back(newGuy);
 		numBins++;
@@ -219,7 +283,7 @@ void SharedRAbundVector::push_back(int binSize, int otu, string groupName){
 		numSeqs += binSize;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "push_back");
+		m->errorOut(e, "SharedRAbundVector", "push_back");
 		exit(1);
 	}
 }
@@ -243,7 +307,7 @@ void SharedRAbundVector::insert(int binSize, int otu, string groupName){
 		numSeqs += binSize;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "insert");
+		m->errorOut(e, "SharedRAbundVector", "insert");
 		exit(1);
 	}
 }
@@ -267,7 +331,7 @@ void SharedRAbundVector::push_front(int binSize, int otu, string groupName){
 		numSeqs += binSize;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "push_front");
+		m->errorOut(e, "SharedRAbundVector", "push_front");
 		exit(1);
 	}
 }
@@ -304,8 +368,54 @@ int SharedRAbundVector::size(){
 	return data.size();
 }
 
+
 /***********************************************************************/
-void SharedRAbundVector::print(ostream& output){
+void SharedRAbundVector::printHeaders(ostream& output){
+	try {
+		string snumBins = toString(numBins);
+		output << "label\tGroup\tnumOtus\t";
+		if (m->sharedHeaderMode == "tax") {
+			for (int i = 0; i < numBins; i++) {  
+				
+				//if there is a bin label use it otherwise make one
+				string binLabel = "PhyloType";
+				string sbinNumber = toString(i+1);
+				if (sbinNumber.length() < snumBins.length()) { 
+					int diff = snumBins.length() - sbinNumber.length();
+					for (int h = 0; h < diff; h++) { binLabel += "0"; }
+				}
+				binLabel += sbinNumber;
+				if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
+				
+				output << binLabel << '\t'; 
+			}
+			output << endl;
+		}else {
+			for (int i = 0; i < numBins; i++) {  
+				//if there is a bin label use it otherwise make one
+				string binLabel = "Otu";
+				string sbinNumber = toString(i+1);
+				if (sbinNumber.length() < snumBins.length()) { 
+					int diff = snumBins.length() - sbinNumber.length();
+					for (int h = 0; h < diff; h++) { binLabel += "0"; }
+				}
+				binLabel += sbinNumber;
+				if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
+				
+				output << binLabel << '\t'; 
+			}
+			
+			output << endl;
+		}
+		m->printedHeaders = true;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "SharedRAbundVector", "printHeaders");
+		exit(1);
+	}
+}
+/***********************************************************************/
+void SharedRAbundVector::print(ostream& output) {
 	try {
 		output << numBins << '\t';
 	
@@ -313,7 +423,7 @@ void SharedRAbundVector::print(ostream& output){
 		output << endl;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "print");
+		m->errorOut(e, "SharedRAbundVector", "print");
 		exit(1);
 	}
 }
@@ -333,7 +443,7 @@ int SharedRAbundVector::getGroupIndex()  { return index; }
 void SharedRAbundVector::setGroupIndex(int vIndex)	{ index = vIndex; }
 /***********************************************************************/
 int SharedRAbundVector::getNumBins(){
-	return numBins;
+		return numBins;
 }
 
 /***********************************************************************/
@@ -358,23 +468,120 @@ vector<SharedRAbundVector*> SharedRAbundVector::getSharedRAbundVectors(){
 		SharedUtil* util;
 		util = new SharedUtil();
 		
-		util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
-
+		vector<string> Groups = m->getGroups();
+		vector<string> allGroups = m->getAllGroups();
+		util->setGroups(Groups, allGroups);
+		m->setGroups(Groups);
+		
+		bool remove = false;
 		for (int i = 0; i < lookup.size(); i++) {
 			//if this sharedrabund is not from a group the user wants then delete it.
-			if (util->isValidGroup(lookup[i]->getGroup(), globaldata->Groups) == false) { 
-				delete lookup[i]; 
+			if (util->isValidGroup(lookup[i]->getGroup(), m->getGroups()) == false) { 
+				remove = true;
+				delete lookup[i]; lookup[i] = NULL;
 				lookup.erase(lookup.begin()+i); 
 				i--; 
 			}
 		}
 		
 		delete util;
+		
+		if (remove) { eliminateZeroOTUS(lookup); }
 	
 		return lookup;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getSharedRAbundVectors");
+		m->errorOut(e, "SharedRAbundVector", "getSharedRAbundVectors");
+		exit(1);
+	}
+}
+//**********************************************************************************************************************
+int SharedRAbundVector::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
+		try {
+			
+			vector<SharedRAbundVector*> newLookup;
+			for (int i = 0; i < thislookup.size(); i++) {
+				SharedRAbundVector* temp = new SharedRAbundVector();
+				temp->setLabel(thislookup[i]->getLabel());
+				temp->setGroup(thislookup[i]->getGroup());
+				newLookup.push_back(temp);
+			}
+			
+			//for each bin
+			vector<string> newBinLabels;
+			string snumBins = toString(thislookup[0]->getNumBins());
+			for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
+				if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
+				
+				//look at each sharedRabund and make sure they are not all zero
+				bool allZero = true;
+				for (int j = 0; j < thislookup.size(); j++) {
+					if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
+				}
+				
+				//if they are not all zero add this bin
+				if (!allZero) {
+					for (int j = 0; j < thislookup.size(); j++) {
+						newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
+					}
+					
+					//if there is a bin label use it otherwise make one
+					string binLabel = "Otu";
+					string sbinNumber = toString(i+1);
+					if (sbinNumber.length() < snumBins.length()) { 
+						int diff = snumBins.length() - sbinNumber.length();
+						for (int h = 0; h < diff; h++) { binLabel += "0"; }
+					}
+					binLabel += sbinNumber; 
+					if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
+					
+					newBinLabels.push_back(binLabel);
+				}
+			}
+			
+			for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
+			
+			thislookup = newLookup;
+			m->currentBinLabels = newBinLabels;
+			
+			return 0;
+			
+		}
+		catch(exception& e) {
+			m->errorOut(e, "SharedRAbundVector", "eliminateZeroOTUS");
+			exit(1);
+		}
+	}
+	
+/***********************************************************************/
+vector<SharedRAbundFloatVector*> SharedRAbundVector::getSharedRAbundFloatVectors(vector<SharedRAbundVector*> thislookup){
+	try {
+		vector<SharedRAbundFloatVector*> newLookupFloat;	
+		for (int i = 0; i < lookup.size(); i++) {
+			SharedRAbundFloatVector* temp = new SharedRAbundFloatVector();
+			temp->setLabel(thislookup[i]->getLabel());
+			temp->setGroup(thislookup[i]->getGroup());
+			newLookupFloat.push_back(temp);
+		}
+		
+		for (int i = 0; i < thislookup.size(); i++) {
+			
+			for (int j = 0; j < thislookup[i]->getNumBins(); j++) {
+				
+				if (m->control_pressed) { return newLookupFloat; }
+				
+				int abund = thislookup[i]->getAbundance(j);
+				
+				float relabund = abund / (float) thislookup[i]->getNumSeqs();
+				
+				newLookupFloat[i]->push_back(relabund, thislookup[i]->getGroup());
+			}
+		}
+		
+		return newLookupFloat;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "SharedRAbundVector", "getSharedRAbundVectors");
 		exit(1);
 	}
 }
@@ -394,7 +601,7 @@ RAbundVector SharedRAbundVector::getRAbundVector() {
 		return rav;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getRAbundVector");
+		m->errorOut(e, "SharedRAbundVector", "getRAbundVector");
 		exit(1);
 	}
 }
@@ -409,7 +616,7 @@ RAbundVector SharedRAbundVector::getRAbundVector2() {
 		return rav;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getRAbundVector2");
+		m->errorOut(e, "SharedRAbundVector", "getRAbundVector2");
 		exit(1);
 	}
 }
@@ -431,7 +638,7 @@ SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 		return sav;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getSharedSAbundVector");
+		m->errorOut(e, "SharedRAbundVector", "getSharedSAbundVector");
 		exit(1);
 	}
 }
@@ -450,7 +657,7 @@ SAbundVector SharedRAbundVector::getSAbundVector() {
 		return sav;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getSAbundVector");		
+		m->errorOut(e, "SharedRAbundVector", "getSAbundVector");		
 		exit(1);
 	}
 }
@@ -474,7 +681,7 @@ SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 		return ov;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getSharedOrderVector");
+		m->errorOut(e, "SharedRAbundVector", "getSharedOrderVector");
 		exit(1);
 	}
 }
@@ -483,19 +690,19 @@ SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 OrderVector SharedRAbundVector::getOrderVector(map<string,int>* nameMap = NULL) {
 	try {
 		OrderVector ov;
-	
-		for(int i=0;i<data.size();i++){
+		for(int i=0;i<numBins;i++){
 			for(int j=0;j<data[i].abundance;j++){
 				ov.push_back(i);
 			}
 		}
 		random_shuffle(ov.begin(), ov.end());
-
+		
 		ov.setLabel(label);	
+
 		return ov;
 	}
 	catch(exception& e) {
-		errorOut(e, "SharedRAbundVector", "getOrderVector");
+		m->errorOut(e, "SharedRAbundVector", "getOrderVector");
 		exit(1);
 	}
 }