X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=qualityscores.cpp;h=26492245e2b9805144125052d57045a834bc3f95;hp=dd5d6de8da69a000882eadd86f8d2c50f853454f;hb=a8e2df1b96a57f5f29576b08361b86a96a8eff4f;hpb=e0fbf58358a72f20352cf2a43922ab6b5bdf0cf8

diff --git a/qualityscores.cpp b/qualityscores.cpp
index dd5d6de..2649224 100644
--- a/qualityscores.cpp
+++ b/qualityscores.cpp
@@ -16,6 +16,7 @@ QualityScores::QualityScores(){
 		m = MothurOut::getInstance();
 		seqName = "";
 		seqLength = -1;
+		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "QualityScores", "QualityScores");
@@ -25,47 +26,92 @@ QualityScores::QualityScores(){
 
 /**************************************************************************************************/
 
-QualityScores::QualityScores(ifstream& qFile, int l){
+QualityScores::QualityScores(ifstream& qFile){
 	try {
 		
 		m = MothurOut::getInstance();
 
-		seqName = "";
-		seqLength = l;
 		int score;
+		seqName = getSequenceName(qFile);
 		
-		string line;
-		getline(qFile, line); gobble(qFile);
-		istringstream nameStream(line);
-	
-		nameStream >> seqName;
-		seqName = seqName.substr(1); 
-
-		//getline(qFile, line);
-		//istringstream qualStream(line);
-	
-		//while(qualStream){
-		//	qualStream >> score;
-		//	qScores.push_back(score);
-		//}
-		//qScores.pop_back();
+        if (m->debug) { m->mothurOut("[DEBUG]: name = '" + seqName + "'\n.");  }
+        
+		if (!m->control_pressed) {
+            string qScoreString = m->getline(qFile);
+            
+            if (m->debug) { m->mothurOut("[DEBUG]: scores = '" + qScoreString + "'\n.");  }
+            
+            while(qFile.peek() != '>' && qFile.peek() != EOF){
+                if (m->control_pressed) { break; }
+                string temp = m->getline(qFile);
+                if (m->debug) { m->mothurOut("[DEBUG]: scores = '" + temp + "'\n.");  }
+                qScoreString +=  ' ' + temp;
+            }
+            //cout << "done reading " << endl;	
+            istringstream qScoreStringStream(qScoreString);
+            int count = 0;
+            while(!qScoreStringStream.eof()){
+                if (m->control_pressed) { break; }
+                string temp;
+                qScoreStringStream >> temp;  m->gobble(qScoreStringStream);
+                
+                if (m->debug) { m->mothurOut("[DEBUG]: score " + toString(qScores.size()) + " = '" + temp + "'\n.");  }
+                
+                //check temp to make sure its a number
+                if (!m->isContainingOnlyDigits(temp)) { m->mothurOut("[ERROR]: In sequence " + seqName + "'s quality scores, expected a number and got " + temp + ", setting score to 0."); m->mothurOutEndLine(); temp = "0"; }
+                convert(temp, score);
+                
+                //cout << count << '\t' << score << endl;
+                qScores.push_back(score);
+                count++;
+            }
+        }
 		
-		//seqLength = qScores.size();	
+		seqLength = qScores.size();
+		//cout << "seqlength = " << seqLength << '\t' << count << endl;
 		
-		for(int i=0;i<seqLength;i++){
-			qFile >> score;
-			qScores.push_back(score);
-		}
-		gobble(qFile);
-
 	}
 	catch(exception& e) {
 		m->errorOut(e, "QualityScores", "QualityScores");
 		exit(1);
 	}							
-
+	
+}
+//********************************************************************************************************************
+string QualityScores::getSequenceName(ifstream& qFile) {
+	try {
+		string name = "";
+		
+        qFile >> name;
+        m->getline(qFile);
+		
+		if (name.length() != 0) { 
+            
+			name = name.substr(1); 
+            
+            m->checkName(name);
+            
+        }else{ m->mothurOut("Error in reading your qfile, at position " + toString(qFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
+        
+		return name;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "QualityScores", "getSequenceName");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+void QualityScores::setName(string name) {
+	try {
+      
+        m->checkName(name);   
+        seqName = name;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "QualityScores", "setName");
+		exit(1);
+	}
 }
-
 /**************************************************************************************************/
 
 string QualityScores::getName(){
@@ -106,13 +152,18 @@ void QualityScores::trimQScores(int start, int end){
 	try {
 		vector<int> hold;
 		
+
+		//cout << seqName << '\t' << start << '\t' << end << '\t' << qScores.size() << endl;
+		//for (int i = 0; i < qScores.size(); i++) { cout << qScores[i] << end; }
 		if(end == -1){		
 			hold = vector<int>(qScores.begin()+start, qScores.end());
 			qScores = hold;		
 		}
 		if(start == -1){
-			hold = vector<int>(qScores.begin(), qScores.begin()+end);	//not sure if indexing is correct
-			qScores = hold;		
+			if(qScores.size() > end){
+				hold = vector<int>(qScores.begin(), qScores.begin()+end);
+				qScores = hold;		
+			}
 		}
 
 		seqLength = qScores.size();
@@ -160,9 +211,12 @@ bool QualityScores::stripQualThreshold(Sequence& sequence, double qThreshold){
 			}
 		}
 		
+		//every score passed
+		if (end == (seqLength-1)) { end = seqLength; }
+		
 		sequence.setUnaligned(rawSequence.substr(0,end));
 		trimQScores(-1, end);
-	
+		
 		return 1;
 	}
 	catch(exception& e) {
@@ -193,16 +247,17 @@ bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshol
 			
 			if(rollingSum / (double)(i+1) < qThreshold){
 				end = i;
-//				cout << i+1 << '\t' << seqName << '\t' << rollingSum / (double)(i+1) << endl;
 				break;
 			}
 		}
 		
 		if(end == -1){	end = seqLength;	}
 		
+		
 		sequence.setUnaligned(rawSequence.substr(0,end));
 		trimQScores(-1, end);
 		
+		
 		return 1;
 	}
 	catch(exception& e) {
@@ -220,41 +275,47 @@ bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int
 		int seqLength = sequence.getNumBases();
 		
 		if(seqName != sequence.getName()){
-			m->mothurOut("sequence name mismatch btwn fasta: " + sequence.getName() + " and qual file: " + seqName);
-			m->mothurOutEndLine();	
+			m->mothurOut("sequence name mismatch between fasta: " + sequence.getName() + " and qual file: " + seqName);
+			m->mothurOutEndLine();
 		}
 		
 		int end = windowSize;
 		int start = 0;
-		
 
-		while(start < seqLength){
+		if(seqLength < windowSize) {	return 0;	}
+			
+		while((start+windowSize) < seqLength){
 			double windowSum = 0.0000;
 
 			for(int i=start;i<end;i++){
-				windowSum += qScores[i];				
+				windowSum += qScores[i];
 			}
 			double windowAverage = windowSum / (double)(end-start);
-			
+				
 			if(windowAverage < qThreshold){
-				end = start;
+				end = end - stepSize;
 				break;
 			}
+			
 			start += stepSize;
 			end = start + windowSize;
-			if(end >= seqLength){	end = seqLength - 1;	}
+				
+			if(end >= seqLength){	end = seqLength;	}
+				
 		}
-		
-		
+	
 		if(end == -1){	end = seqLength;	}
 		
+		//failed first window
+		if (end < windowSize) { return 0; }
+			
 		sequence.setUnaligned(rawSequence.substr(0,end));
 		trimQScores(-1, end);
 		
 		return 1;
 	}
 	catch(exception& e) {
-		m->errorOut(e, "QualityScores", "flipQScores");
+		m->errorOut(e, "QualityScores", "stripQualWindowAverage");
 		exit(1);
 	}							
 	
@@ -294,7 +355,69 @@ bool QualityScores::cullQualAverage(Sequence& sequence, double qAverage){
 		return success;
 	}
 	catch(exception& e) {
-		m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
+		m->errorOut(e, "QualityScores", "cullQualAverage");
+		exit(1);
+	}
+}
+
+/**************************************************************************************************/
+
+void QualityScores::updateQScoreErrorMap(map<char, vector<int> >& qualErrorMap, string errorSeq, int start, int stop, int weight){
+	try {
+
+		int seqLength = errorSeq.size();
+		
+		int qIndex = start - 1;
+
+		for(int i=0;i<seqLength;i++){
+			
+			if(errorSeq[i] == 'm')		{	qualErrorMap['m'][qScores[qIndex]] += weight;	}
+			else if(errorSeq[i] == 's')	{	qualErrorMap['s'][qScores[qIndex]] += weight;	}
+			else if(errorSeq[i] == 'i')	{	qualErrorMap['i'][qScores[qIndex]] += weight;	}
+			else if(errorSeq[i] == 'a')	{	qualErrorMap['a'][qScores[qIndex]] += weight;	/*if(qScores[qIndex] != 0){	cout << qIndex << '\t';		}*/	}
+			else if(errorSeq[i] == 'd')	{	/*	there are no qScores for deletions	*/		}
+
+			if(errorSeq[i] != 'd')		{	qIndex++;	}
+						
+			if(qIndex > stop){	break;	}
+		}	
+	}
+	catch(exception& e) {
+		m->errorOut(e, "QualityScores", "updateQScoreErrorMap");
+		exit(1);
+	}
+}
+
+/**************************************************************************************************/
+
+void QualityScores::updateForwardMap(vector<vector<int> >& forwardMap, int start, int stop, int weight){
+	try {
+		
+		int index = 0;
+		for(int i=start-1;i<stop;i++){
+			forwardMap[index++][qScores[i]] += weight;
+		}
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "QualityScores", "updateForwardMap");
+		exit(1);
+	}
+}
+
+/**************************************************************************************************/
+
+void QualityScores::updateReverseMap(vector<vector<int> >& reverseMap, int start, int stop, int weight){
+	try {
+		
+		int index = 0;
+		for(int i=stop-1;i>=start-1;i--){
+			reverseMap[index++][qScores[i]] += weight;
+		}
+		
+	}	
+	catch(exception& e) {
+		m->errorOut(e, "QualityScores", "updateReverseMap");
 		exit(1);
 	}
 }