X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=qualityscores.cpp;h=566b44c6bad55402fae1374aab0212547befe7cc;hb=e98d569d630c30d1ac3608eb6337bcec4765a724;hp=aafa5794e206b936793cf6ffce6109f66eebaf86;hpb=bdb5d82e2a73829b4e1fa42656ad9bcb57e3e948;p=mothur.git

diff --git a/qualityscores.cpp b/qualityscores.cpp
index aafa579..566b44c 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");
@@ -25,82 +26,104 @@ QualityScores::QualityScores(){
 
 /**************************************************************************************************/
 
-QualityScores::QualityScores(ifstream& qFile, int l){
+QualityScores::QualityScores(ifstream& qFile){
 	try {
 		
 		m = MothurOut::getInstance();
-
+		
 		seqName = "";
-		seqLength = l;
 		int score;
 		
 		qFile >> seqName; 
-
-		while (!qFile.eof())	{	char c = qFile.get(); if (c == 10 || c == 13 || c == -1){	break;	}	} // get rest of line 
-		m->gobble(qFile);
-		if (seqName == "") { m->mothurOut("Error reading quality file, name blank at position, " + toString(qFile.tellg())); m->mothurOutEndLine(); }
-		else {
-			seqName = seqName.substr(1); 
+		m->getline(qFile);
+		//cout << seqName << endl;	
+		if (seqName == "")	{
+			m->mothurOut("Error reading quality file, name blank at position, " + toString(qFile.tellg()));
+			m->mothurOutEndLine(); 
+		}
+		else{
+			seqName = seqName.substr(1);
 		}
 		
-		//m->getline(qFile, line);
-		//istringstream qualStream(line);
-	
-		//while(qualStream){
-		//	qualStream >> score;
-		//	qScores.push_back(score);
-		//}
-		//qScores.pop_back();
-		
-		//seqLength = qScores.size();	
-		
-		/*while(!in.eof()){	
-			string saveName = "";
-			string name = "";
-			string scores = "";
-			
-			in >> name; 
-			//cout << name << endl;		
-			if (name.length() != 0) { 
-				saveName = name.substr(1);
-				while (!in.eof())	{	
-					char c = in.get(); 
-					if (c == 10 || c == 13){	break;	}
-					else { name += c; }	
-				} 
-				m->gobble(in);
-			}
-			
-			while(in){
-				char letter= in.get();
-				if(letter == '>'){	in.putback(letter);	break;	}
-				else{ scores += letter; }
-			}
+		string qScoreString = m->getline(qFile);
+		//cout << qScoreString << endl;
+		while(qFile.peek() != '>' && qFile.peek() != EOF){
+			if (m->control_pressed) { break; }
+			string temp = m->getline(qFile);
+			//cout << temp << endl;
+			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);
 			
-		 //istringstream iss (scores,istringstream::in);
-		 
-		 //int count = 0; int tempScore;
-		 //while (iss) { iss >> tempScore; count++; }
-		 //cout << saveName << '\t' << count << endl;	
+			//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);
 			
-			m->gobble(in);
-		}*/		
-		
-		
-		
-		for(int i=0;i<seqLength;i++){
-			qFile >> score;
+			//cout << count << '\t' << score << endl;
 			qScores.push_back(score);
+			count++;
 		}
-		m->gobble(qFile);
+		//qScores.pop_back();
+		
+//		string scores = "";
+//		
+//		while(!qFile.eof()){	
+//			
+//			qFile >> seqName; 
+//			
+//			//get name
+//			if (seqName.length() != 0) { 
+//				seqName = seqName.substr(1);
+//				while (!qFile.eof())	{	
+//					char c = qFile.get(); 
+//					//gobble junk on line
+//					if (c == 10 || c == 13){	break;	}
+//				} 
+//				m->gobble(qFile);
+//			}
+//			
+//			//get scores
+//			while(qFile){
+//				char letter=qFile.get();
+//				if((letter == '>')){	qFile.putback(letter);	break;	}
+//				else if (isprint(letter)) { scores += letter; }
+//			}
+//			m->gobble(qFile);
+//			
+//			break;
+//		}
+//		
+//		//convert scores string to qScores
+//		istringstream qScoreStringStream(scores);
+//		
+//		int score;
+//		while(!qScoreStringStream.eof()){
+//			
+//			if (m->control_pressed) { break; }
+//			
+//			qScoreStringStream >> score;
+//			qScores.push_back(score);
+//		}
+//		
+//		qScores.pop_back();
 
+		seqLength = qScores.size();
+		//cout << "seqlength = " << seqLength << '\t' << count << endl;
+		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "QualityScores", "QualityScores");
 		exit(1);
 	}							
-
+	
 }
+
 /**************************************************************************************************/
 
 string QualityScores::getName(){
@@ -141,6 +164,9 @@ 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;		
@@ -197,9 +223,12 @@ 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) {
@@ -230,16 +259,17 @@ 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) {
@@ -263,36 +293,41 @@ bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int
 		
 		int end = windowSize;
 		int start = 0;
-		
+
 		if(seqLength < windowSize) {	return 0;	}
-		
-		while(start < seqLength){
+			
+		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 = 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);
 	}							
 	
@@ -343,17 +378,20 @@ void QualityScores::updateQScoreErrorMap(map<char, vector<int> >& qualErrorMap,
 	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;	}
+			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) {