Sequential quality score calculation is now in place in the refactored recalibrator and matches the quality scores calculated by the old recalibrator exactly; at least on the small sets of data used so far. Validation, documentation, and optimization work is on going.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1985 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/CovariateCounterWalker.java

@@ -126,7 +126,7 @@ public class CovariateCounterWalker extends LocusWalker<Integer, PrintStream> {
 
     private void updateDataFromRead(SAMRecord read, int offset, ReferenceContext ref) {
 
-        ArrayList<Comparable<?>> key = new ArrayList<Comparable<?>>();
+        List<Comparable<?>> key = new ArrayList<Comparable<?>>();
         Comparable<?> keyElement; // preallocate for use in for loop below
         boolean badKey = false;
         for( Covariate covariate : requestedCovariates ) {

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/CycleCovariate.java

@@ -21,7 +21,7 @@ public class CycleCovariate implements Covariate {
 
     public Comparable<?> getValue(SAMRecord read, int offset, char[] refBases) {
         //BUGBUG: assumes Solexia platform
-        int cycle = offset;
+        Integer cycle = offset;
         if( read.getReadNegativeStrandFlag() ) {
             cycle = read.getReadLength() - (offset + 1);
         }

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/MappingQualityCovariate.java

@@ -13,7 +13,7 @@ public class MappingQualityCovariate implements Covariate {
     }
 
     public Comparable<?> getValue(SAMRecord read, int offset, char[] refBases) {
-        return read.getMappingQuality();
+        return (Integer)(read.getMappingQuality());
     }
     
     public Comparable<?> getValue(String str) {

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/MinimumNQSCovariate.java

@@ -32,7 +32,7 @@ public class MinimumNQSCovariate implements Covariate {
             }
         }
 
-        int minQual = quals[0];
+        Integer minQual = (int)(quals[0]);
         for ( int qual : quals ) {
             if( qual < minQual ) {
                 minQual = qual;

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/QualityScoreCovariate.java

@@ -32,7 +32,7 @@ public class QualityScoreCovariate implements Covariate {
             }
         }
 
-        return quals[offset];
+        return ((Integer)((int)quals[offset]));
     }
     
     public Comparable<?> getValue(String str) {

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/RecalDataManager.java

@@ -75,6 +75,7 @@ public class RecalDataManager {
             newKey.add( key.get(1) ); //                                    and quality score
             collapsedDatum = dataCollapsedQualityScore.get( newKey );
             if( collapsedDatum == null ) {
+            	//System.out.println("Added: " + newKey + "   " + newKey.hashCode());
                 dataCollapsedQualityScore.put( newKey, new RecalDatum( thisDatum ) );
             } else {
                 collapsedDatum.increment( thisDatum );
@@ -85,7 +86,7 @@ public class RecalDataManager {
                 newKey = new ArrayList<Comparable<?>>();
                 newKey.add( key.get(0) ); // make a new key with the read group ...
                 newKey.add( key.get(1) ); //                                    and quality score ...
-                newKey.add( key.get(iii) ); //                                                    and the given covariate
+                newKey.add( key.get(iii + 2) ); //                                                    and the given covariate
                 collapsedDatum = dataCollapsedByCovariate.get(iii).get( newKey );
                 if( collapsedDatum == null ) {
                     dataCollapsedByCovariate.get(iii).put( newKey, new RecalDatum( thisDatum ) );

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/RecalDatum.java

@@ -104,4 +104,8 @@ public class RecalDatum {
     public Long getNumObservations() {
         return numObservations;
     }
+    
+    public String toString() {
+    	return String.format( "RecalDatum: %d,%d,%d", numObservations, numMismatches, (int)empiricalQualByte() );
+    }
 }

java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/TableRecalibrationWalker.java

@@ -43,8 +43,8 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
     public String MODE_STRING = RecalibrationMode.SEQUENTIAL.toString();
     public RecalibrationMode MODE = RecalibrationMode.SEQUENTIAL; //BUGBUG: need some code here to set this properly
 
-    protected static RecalDataManager dataManager;
+    protected RecalDataManager dataManager;
-    protected static ArrayList<Covariate> requestedCovariates;
+    protected ArrayList<Covariate> requestedCovariates;
 
     private static Pattern COVARIATE_PATTERN = Pattern.compile("^@!.*");
     public final static String ORIGINAL_QUAL_ATTRIBUTE_TAG = "OQ";
@@ -117,11 +117,13 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
             dataManager.createCollapsedTables( requestedCovariates.size() );
             out.println( "...done!" );
         }
+        
+        //System.out.println(dataManager.getCollapsedTable(1));
     }
 
     private void addCSVData(String line) {
         String[] vals = line.split(",");
-        List<Comparable<?>> key = new ArrayList<Comparable<?>>();
+        ArrayList<Comparable<?>> key = new ArrayList<Comparable<?>>();
         Covariate cov; // preallocated for use in for loop below
         int iii;
         for( iii = 0; iii < requestedCovariates.size(); iii++ ) {
@@ -197,11 +199,12 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
         newKey.add( key.get(0) ); // read group
         RecalDatum globalDeltaQDatum = dataManager.getCollapsedTable(0).get( newKey );
         double globalDeltaQ = 0.0;
+        double aggregrateQreported = 0.0;
         if( globalDeltaQDatum != null ) {
-            globalDeltaQ = globalDeltaQDatum.empiricalQualDouble( SMOOTHING ) - ( dataManager.dataSumExpectedErrors.get( newKey ) / ((double) globalDeltaQDatum.getNumObservations()) );
+        	aggregrateQreported = QualityUtils.phredScaleErrorRate( dataManager.dataSumExpectedErrors.get( newKey ) / ((double) globalDeltaQDatum.getNumObservations()) );
+           globalDeltaQ = globalDeltaQDatum.empiricalQualDouble( SMOOTHING ) - aggregrateQreported;
         }
-        //System.out.printf("Global quality score shift is %.2f - %.2f = %.2f%n",
+        
-        //        globalDeltaQDatum.empiricalQualDouble( SMOOTHING ), ( dataManager.dataSumExpectedErrors.get( newKey ) / ((double) globalDeltaQDatum.getNumObservations())), globalDeltaQ);
        
         newKey = new ArrayList<Comparable<?>>();
         newKey.add( key.get(0) ); // read group
@@ -212,6 +215,7 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
             deltaQReported = deltaQReportedDatum.empiricalQualDouble( SMOOTHING ) - qualFromRead - globalDeltaQ;
         }
         
+        
         double deltaQCovariates = 0.0;
         RecalDatum deltaQCovariateDatum;
         for( int iii = 2; iii < key.size(); iii++ ) {
@@ -228,10 +232,6 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
         double newQuality = qualFromRead + globalDeltaQ + deltaQReported + deltaQCovariates;
         byte newQualityByte = QualityUtils.boundQual( (int)Math.round(newQuality), QualityUtils.MAX_REASONABLE_Q_SCORE );
         
-
-        //System.out.println( "base quality score calculated: " + key +
-        //                String.format( " => %d + %.2f + %.2f + %.2f = %d", qualFromRead, globalDeltaQ, deltaQReported, deltaQCovariates, newQualityByte ) );
-        
         if( newQualityByte <= 0 && newQualityByte >= QualityUtils.MAX_REASONABLE_Q_SCORE ) {
             throw new StingException( "Illegal base quality score calculated: " + key +
                         String.format( " => %d + %.2f + %.2f + %.2f = %d", qualFromRead, globalDeltaQ, deltaQReported, deltaQCovariates, newQualityByte ) );
@@ -260,7 +260,4 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
 
         return output;
     }
-
-    public void onTraversalDone( SAMFileWriter reduceResult ) {
-    }
 }