modified seqerror and qualscores

[mothur.git] / qualityscores.cpp

diff --git a/qualityscores.cpp b/qualityscores.cpp
index 4b315b1c397d3a42921b0071253065b26b4befe6..fca3d9f4c9cd17947116fbdb9a872bfce863fe47 100644 (file)
--- 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");
@@ -29,34 +30,36 @@ QualityScores::QualityScores(ifstream& qFile){
        try {
                
                m = MothurOut::getInstance();
-
+               
                seqName = "";
-               seqLength = -1;
                int score;
                
-               string line;
-               getline(qFile, line);
-               istringstream nameStream(line);
-       
-               nameStream >> seqName;
-               seqName = seqName.substr(1); 
-
-               getline(qFile, line);
-               istringstream qualStream(line);
-       
-               while(qualStream){
-                       qualStream >> score;
+               qFile >> seqName; 
+               m->getline(qFile);
+               
+               if (seqName == "")      {
+                       m->mothurOut("Error reading quality file, name blank at position, " + toString(qFile.tellg()));
+                       m->mothurOutEndLine(); 
+               }
+               else{
+                       seqName = seqName.substr(1);
+               }
+               
+               string qScoreString = m->getline(qFile);
+               
+               istringstream qScoreStringStream(qScoreString);
+               while(!qScoreStringStream.eof()){
+                       qScoreStringStream >> score;
                        qScores.push_back(score);
                }
                qScores.pop_back();
-               
-               seqLength = qScores.size();     
+               seqLength = qScores.size();
        }
        catch(exception& e) {
                m->errorOut(e, "QualityScores", "QualityScores");
                exit(1);
        }                                                       
-
+       
 }
 
 /**************************************************************************************************/
@@ -104,8 +107,10 @@ void QualityScores::trimQScores(int start, int 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();
@@ -153,9 +158,12 @@ 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) {
@@ -193,9 +201,11 @@ bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshol
                
                if(end == -1){  end = seqLength;        }
                
+               
                sequence.setUnaligned(rawSequence.substr(0,end));
                trimQScores(-1, end);
                
+               
                return 1;
        }
        catch(exception& e) {
@@ -213,24 +223,25 @@ 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;
                
-
+               if(seqLength < windowSize) {    return 0;       }
+                       
                while(start < 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;
@@ -238,7 +249,6 @@ bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int
                        if(end >= seqLength){   end = seqLength - 1;    }
                }
                
-               
                if(end == -1){  end = seqLength;        }
                
                sequence.setUnaligned(rawSequence.substr(0,end));
@@ -287,7 +297,67 @@ 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;   }
+                       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;i--){
+                       reverseMap[index++][qScores[i]] += weight;
+               }
+               
+       }       
+       catch(exception& e) {
+               m->errorOut(e, "QualityScores", "updateForwardMap");
                exit(1);
        }
 }