misc code cleanup

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2190 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/walkers/recalibration/CovariateCounterWalker.java

@@ -101,7 +101,7 @@ public class CovariateCounterWalker extends LocusWalker<Integer, PrintStream> {
     private long countedBases = 0; // Number of bases used in the calculations, used for reporting in the output file
     private long skippedSites = 0; // Number of loci skipped because it was a dbSNP site, used for reporting in the output file
     private int numUnprocessed = 0; // Number of consecutive loci skipped because we are only processing every Nth site
-    private static final String versionString = "v2.0.10"; // Major version, minor version, and build number
+    private static final String versionString = "v2.0.11"; // Major version, minor version, and build number
     private Pair<Long, Long> dbSNP_counts = new Pair<Long, Long>(0L, 0L);  // mismatch/base counts for dbSNP loci
     private Pair<Long, Long> novel_counts = new Pair<Long, Long>(0L, 0L);  // mismatch/base counts for non-dbSNP loci
     private static final double DBSNP_VS_NOVEL_MISMATCH_RATE = 2.0;        // rate at which dbSNP sites (on an individual level) mismatch relative to novel sites (determined by looking at NA12878)
@@ -269,6 +269,7 @@ public class CovariateCounterWalker extends LocusWalker<Integer, PrintStream> {
             byte refBase;
             byte prevBase;
             byte[] colorSpaceQuals;
+            byte[] bases;
 
             // For each read at this locus
             for ( PileupElement p : context.getPileup() ) {
@@ -284,16 +285,18 @@ public class CovariateCounterWalker extends LocusWalker<Integer, PrintStream> {
                         // Skip if base quality is zero
                         if( read.getBaseQualities()[offset] > 0 ) {
 
+                            bases = read.getReadBases();
                             refBase = (byte)ref.getBase();
-                            prevBase = read.getReadBases()[offset - 1];
+                            prevBase = bases[offset - 1];
 
                             // DinucCovariate is responsible for getting the complement bases if needed
                             if( read.getReadNegativeStrandFlag() ) {
-                                prevBase = read.getReadBases()[offset + 1];
+                                prevBase = bases[offset + 1];
                             }
 
                             // Skip if this base or the previous one was an 'N' or etc.
-                            if( BaseUtils.isRegularBase( (char)prevBase ) && BaseUtils.isRegularBase( (char)(read.getReadBases()[offset]) ) ) {
+                            // BUGBUG: For DinucCovariate we should use previous reference base, not the previous base in this read. 
+                            if( BaseUtils.isRegularBase( (char)prevBase ) && BaseUtils.isRegularBase( (char)(bases[offset]) ) ) {
 
                                 // SOLID bams insert the reference base into the read if the color space quality is zero, so skip over them
                                 colorSpaceQuals = null;

java/src/org/broadinstitute/sting/gatk/walkers/recalibration/CycleCovariate.java

@@ -114,12 +114,9 @@ public class CycleCovariate implements Covariate {
                 }
                 warnedUserBadPlatform = true;
             }
-            //ReadHashDatum correctedReadDatum = new ReadHashDatum( readDatum );
+            read.getReadGroup().setPlatform( defaultPlatform );
-            //correctedReadDatum.platform = defaultPlatform;
+            return getValue( read, offset ); // a recursive call
-            return 0; // BUGBUG: broken at the moment
-            //return getValue( correctedReadDatum, offset ); // a recursive call
         }
-                                                        
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker

java/src/org/broadinstitute/sting/gatk/walkers/recalibration/DinucCovariate.java

@@ -60,23 +60,20 @@ public class DinucCovariate implements Covariate {
 
         byte base;
         byte prevBase;
+        byte[] bases = read.getReadBases();
         // If this is a negative strand read then we need to reverse the direction for our previous base
         if( read.getReadNegativeStrandFlag() ) {
-            base = (byte)BaseUtils.simpleComplement( (char)(read.getReadBases()[offset]) );
+            base = (byte)BaseUtils.simpleComplement( (char)(bases[offset]) );
-            prevBase = (byte)BaseUtils.simpleComplement( (char)(read.getReadBases()[offset + 1]) );
+            prevBase = (byte)BaseUtils.simpleComplement( (char)(bases[offset + 1]) );
         } else {
-            base = read.getReadBases()[offset];
+            base = bases[offset];
-            prevBase = read.getReadBases()[offset - 1];
+            prevBase = bases[offset - 1];
         }
-        //char[] charArray = {(char)prevBase, (char)base};
-        //return new String( charArray ); // This is an expensive call
         return dinucHashMap.get( Dinuc.hashBytes( prevBase, base ) );
-        //return String.format("%c%c", prevBase, base); // This return statement is too slow
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
     public final Comparable getValue( final String str ) {
-        //return str;
         return dinucHashMap.get( Dinuc.hashBytes( (byte)str.charAt(0), (byte)str.charAt(1) ) );
     }
 

java/src/org/broadinstitute/sting/gatk/walkers/recalibration/RecalDataManager.java

@@ -252,15 +252,15 @@ public class RecalDataManager {
             ((GATKSAMRecord)read).setReadGroup( readGroup );
         }
 
-        if( RAC.FORCE_READ_GROUP != null && !read.getReadGroup().getReadGroupId().equals(RAC.FORCE_READ_GROUP) ) { // Collapse all the read groups into a single common String provided by the user
+        if( RAC.FORCE_READ_GROUP != null && !readGroup.getReadGroupId().equals(RAC.FORCE_READ_GROUP) ) { // Collapse all the read groups into a single common String provided by the user
             String oldPlatform = readGroup.getPlatform();
             readGroup = new SAMReadGroupRecord( RAC.FORCE_READ_GROUP );
             readGroup.setPlatform( oldPlatform );
             ((GATKSAMRecord)read).setReadGroup( readGroup );
         }
 
-        if( RAC.FORCE_PLATFORM != null && !read.getReadGroup().getPlatform().equals(RAC.FORCE_PLATFORM)) {
+        if( RAC.FORCE_PLATFORM != null && !readGroup.getPlatform().equals(RAC.FORCE_PLATFORM)) {
-            read.getReadGroup().setPlatform( RAC.FORCE_PLATFORM );
+            readGroup.setPlatform( RAC.FORCE_PLATFORM );
         }
     }
 }

java/src/org/broadinstitute/sting/gatk/walkers/recalibration/TableRecalibrationWalker.java

@@ -93,7 +93,7 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
     private static final Pattern COMMENT_PATTERN = Pattern.compile("^#.*");
     private static final Pattern OLD_RECALIBRATOR_HEADER = Pattern.compile("^rg,.*");
     private static final Pattern COVARIATE_PATTERN = Pattern.compile("^ReadGroup,QualityScore,.*");
-    private static final String versionString = "v2.0.7"; // Major version, minor version, and build number
+    private static final String versionString = "v2.0.8"; // Major version, minor version, and build number
     private SAMFileWriter OUTPUT_BAM = null;// The File Writer that will write out the recalibrated bam
 
     //---------------------------------------------------------------------------------------------------------------
@@ -257,7 +257,7 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
         // Check if the data line is malformed, for example if the read group string contains a comma then it won't be parsed correctly
         if( vals.length != requestedCovariates.size() + 3 ) { // +3 because of nObservations, nMismatch, and Qempirical
             throw new StingException("Malformed input recalibration file. Found data line with too many fields: " + line +
-                    " --Perhaps the read group string contains a comma and isn't parsed correctly.");
+                    " --Perhaps the read group string contains a comma and isn't being parsed correctly.");
         }
 
         ArrayList<Comparable> key = new ArrayList<Comparable>();
@@ -483,6 +483,5 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
         if ( output != null ) {
             output.close();
         }
-        super.onTraversalDone(output);
     }
 }