X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=qualityscores.cpp;h=3a1687bf9e174d77ddded9eaf331d6f1a78da734;hb=9a808b553f14a25f9c5e94da92a7077a57054e8e;hp=33ca1728052e4db59372fc2d8129e5bcbaf729c5;hpb=7f1aca4ed9e6db70de82e80ca4771f7680b21d26;p=mothur.git

diff --git a/qualityscores.cpp b/qualityscores.cpp
index 33ca172..3a1687b 100644
--- a/qualityscores.cpp
+++ b/qualityscores.cpp
@@ -32,22 +32,22 @@ QualityScores::QualityScores(ifstream& qFile){
 		m = MothurOut::getInstance();
 
 		int score;
-		seqName = getSequenceName(qFile);
+		seqName = getSequenceName(qFile); m->gobble(qFile);
 		
         if (m->debug) { m->mothurOut("[DEBUG]: name = '" + seqName + "'\n.");  }
         
 		if (!m->control_pressed) {
-            string qScoreString = m->getline(qFile);
+            string qScoreString = m->getline(qFile); m->gobble(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.");  }
+                string temp = m->getline(qFile); m->gobble(qFile);
+                //if (m->debug) { m->mothurOut("[DEBUG]: scores = '" + temp + "'\n.");  }
                 qScoreString +=  ' ' + temp;
             }
-            //cout << "done reading " << endl;	
+            //cout << "done reading " << endl;
             istringstream qScoreStringStream(qScoreString);
             int count = 0;
             while(!qScoreStringStream.eof()){
@@ -55,7 +55,7 @@ QualityScores::QualityScores(ifstream& qFile){
                 string temp;
                 qScoreStringStream >> temp;  m->gobble(qScoreStringStream);
                 
-                if (m->debug) { m->mothurOut("[DEBUG]: score " + toString(qScores.size()) + " = '" + temp + "'\n.");  }
+                //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"; }
@@ -68,7 +68,7 @@ QualityScores::QualityScores(ifstream& qFile){
         }
 		
 		seqLength = qScores.size();
-		//cout << "seqlength = " << seqLength << '\t' << count << endl;
+		//cout << "seqlength = " << seqLength  << endl;
 		
 	}
 	catch(exception& e) {
@@ -89,9 +89,7 @@ string QualityScores::getSequenceName(ifstream& qFile) {
             
 			name = name.substr(1); 
             
-            for (int i = 0; i < name.length(); i++) {
-                if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
-            }
+            m->checkName(name);
             
         }else{ m->mothurOut("Error in reading your qfile, at position " + toString(qFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
         
@@ -106,10 +104,7 @@ string QualityScores::getSequenceName(ifstream& qFile) {
 void QualityScores::setName(string name) {
 	try {
       
-        for (int i = 0; i < name.length(); i++) {
-            if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
-        }     
-    
+        m->checkName(name);   
         seqName = name;
 	}
 	catch(exception& e) {
@@ -135,7 +130,7 @@ string QualityScores::getName(){
 void QualityScores::printQScores(ofstream& qFile){
 	try {
 		
-		double aveQScore = calculateAverage();
+		double aveQScore = calculateAverage(false);
 		
 		qFile << '>' << seqName << '\t' << aveQScore << endl;
 		
@@ -205,7 +200,7 @@ bool QualityScores::stripQualThreshold(Sequence& sequence, double qThreshold){
 		
 		if(seqName != sequence.getName()){
 			m->mothurOut("sequence name mismatch btwn fasta: " + sequence.getName() + " and qual file: " + seqName);
-			m->mothurOutEndLine();	
+			m->mothurOutEndLine();	m->control_pressed = true;
 		}
 		
 		int end;
@@ -233,7 +228,7 @@ bool QualityScores::stripQualThreshold(Sequence& sequence, double qThreshold){
 
 /**************************************************************************************************/
 
-bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshold){
+bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshold, bool logTransform){
 	try {
 		string rawSequence = sequence.getUnaligned();
 		int seqLength = sequence.getNumBases();
@@ -245,12 +240,22 @@ bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshol
 		
 		int end = -1;
 		double rollingSum = 0.0000;
+        double value = 0.0;
 		
 		for(int i=0;i<seqLength;i++){
-
-			rollingSum += (double)qScores[i];
+            
+            if (logTransform)   {
+                rollingSum += (double)pow(10.0, qScores[i]);
+                value = log10(rollingSum / (double)(i+1));
+                
+            } //Sum 10^Q
+            else                {
+                rollingSum += (double)qScores[i];
+                value = rollingSum / (double)(i+1);
+            }
 			
-			if(rollingSum / (double)(i+1) < qThreshold){
+			
+			if(value < qThreshold){
 				end = i;
 				break;
 			}
@@ -274,7 +279,7 @@ bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshol
 
 /**************************************************************************************************/
 
-bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int windowSize, double qThreshold){
+bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int windowSize, double qThreshold, bool logTransform){
 	try {
 		string rawSequence = sequence.getUnaligned();
 		int seqLength = sequence.getNumBases();
@@ -293,9 +298,12 @@ bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int
 			double windowSum = 0.0000;
 
 			for(int i=start;i<end;i++){
-				windowSum += qScores[i];
+                if (logTransform)   {  windowSum += pow(10.0, qScores[i]);  }
+                else                {  windowSum += qScores[i];             }
 			}
-			double windowAverage = windowSum / (double)(end-start);
+			double windowAverage = 0.0;
+            if (logTransform)   { windowAverage = log10(windowSum / (double)(end-start)); }
+            else                { windowAverage = windowSum / (double)(end-start);      }
 				
 			if(windowAverage < qThreshold){
 				end = end - stepSize;
@@ -328,21 +336,25 @@ bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int
 
 /**************************************************************************************************/
 
-double QualityScores::calculateAverage(){
+double QualityScores::calculateAverage(bool logTransform){
 	
 	double aveQScore = 0.0000;
 	
 	for(int i=0;i<seqLength;i++){
-		aveQScore += (double) qScores[i];
+        if (logTransform)   {  aveQScore += pow(10.0, qScores[i]);  }
+        else                {  aveQScore += qScores[i];             }
 	}
 	aveQScore /= (double) seqLength;
+    
+    if (logTransform)   {  aveQScore = log10(aveQScore /(double) seqLength);  }
+    else                {  aveQScore /= (double) seqLength;                 }
 	
 	return aveQScore;
 }
 
 /**************************************************************************************************/
 
-bool QualityScores::cullQualAverage(Sequence& sequence, double qAverage){
+bool QualityScores::cullQualAverage(Sequence& sequence, double qAverage, bool logTransform){
 	try {
 		string rawSequence = sequence.getUnaligned();
 		bool success = 0;	//guilty until proven innocent
@@ -352,7 +364,7 @@ bool QualityScores::cullQualAverage(Sequence& sequence, double qAverage){
 			m->mothurOutEndLine();	
 		} 
 			
-		double aveQScore = calculateAverage();
+		double aveQScore = calculateAverage(logTransform);
 		
 		if(aveQScore >= qAverage)	{	success = 1;	}
 		else						{	success = 0;	}