added classify.shared command and random forest files. added count file to pcr.seqs...

[mothur.git] / qualityscores.cpp

diff --git a/qualityscores.cpp b/qualityscores.cpp
index 719bf3dc6c44cd3880ebc8b1c4d3dc352a448ee1..566b44c6bad55402fae1374aab0212547befe7cc 100644 (file)
--- a/qualityscores.cpp
+++ b/qualityscores.cpp
@@ -31,12 +31,12 @@ QualityScores::QualityScores(ifstream& qFile){
                
                m = MothurOut::getInstance();
                
-               /*seqName = "";
+               seqName = "";
                int score;
                
                qFile >> seqName; 
                m->getline(qFile);
-               
+               //cout << seqName << endl;      
                if (seqName == "")      {
                        m->mothurOut("Error reading quality file, name blank at position, " + toString(qFile.tellg()));
                        m->mothurOutEndLine(); 
@@ -44,63 +44,77 @@ QualityScores::QualityScores(ifstream& qFile){
                else{
                        seqName = seqName.substr(1);
                }
-               cout << seqName << endl;        
+               
                string qScoreString = m->getline(qFile);
-               cout << qScoreString << endl;   
+               //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; }
-                       qScoreStringStream >> score;
-                       qScores.push_back(score);
-                       cout << score << '\t' << count << endl;
-                       count++;
-               }
-               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()){
+                       string temp;
+                       qScoreStringStream >> temp;  m->gobble(qScoreStringStream);
                        
-                       if (m->control_pressed) { break; }
+                       //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);
                        
-                       qScoreStringStream >> score;
+                       //cout << count << '\t' << score << endl;
                        qScores.push_back(score);
+                       count++;
                }
-               
-               qScores.pop_back();
+               //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) {
@@ -150,6 +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;         
@@ -276,35 +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];
                        }
                        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);
        }                                                       
        
@@ -363,10 +386,11 @@ void QualityScores::updateQScoreErrorMap(map<char, vector<int> >& qualErrorMap,
                        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;  }
                }       
        }